streaming.c

   1 /*
   2  * Copyright (c) 2011, Google Inc.
   3  */
   4 #include "cache.h"
   5 #include "streaming.h"
   6 #include "packfile.h"
   7
   8 enum input_source {
   9         stream_error = -1,
  10         incore = 0,
  11         loose = 1,
  12         pack_non_delta = 2
  13 };
  14
  15 typedef int (*open_istream_fn)(struct git_istream *,
  16                                struct object_info *,
  17                                const unsigned char *,
  18                                enum object_type *);
  19 typedef int (*close_istream_fn)(struct git_istream *);
  20 typedef ssize_t (*read_istream_fn)(struct git_istream *, char *, size_t);
  21
  22 struct stream_vtbl {
  23         close_istream_fn close;
  24         read_istream_fn read;
  25 };
  26
  27 #define open_method_decl(name) \
  28         int open_istream_ ##name \
  29         (struct git_istream *st, struct object_info *oi, \
  30          const unsigned char *sha1, \
  31          enum object_type *type)
  32
  33 #define close_method_decl(name) \
  34         int close_istream_ ##name \
  35         (struct git_istream *st)
  36
  37 #define read_method_decl(name) \
  38         ssize_t read_istream_ ##name \
  39         (struct git_istream *st, char *buf, size_t sz)
  40
  41 /* forward declaration */
  42 static open_method_decl(incore);
  43 static open_method_decl(loose);
  44 static open_method_decl(pack_non_delta);
  45 static struct git_istream *attach_stream_filter(struct git_istream *st,
  46                                                 struct stream_filter *filter);
  47
  48
  49 static open_istream_fn open_istream_tbl[] = {
  50         open_istream_incore,
  51         open_istream_loose,
  52         open_istream_pack_non_delta,
  53 };
  54
  55 #define FILTER_BUFFER (1024*16)
  56
  57 struct filtered_istream {
  58         struct git_istream *upstream;
  59         struct stream_filter *filter;
  60         char ibuf[FILTER_BUFFER];
  61         char obuf[FILTER_BUFFER];
  62         int i_end, i_ptr;
  63         int o_end, o_ptr;
  64         int input_finished;
  65 };
  66
  67 struct git_istream {
  68         const struct stream_vtbl *vtbl;
  69         unsigned long size; /* inflated size of full object */
  70         git_zstream z;
  71         enum { z_unused, z_used, z_done, z_error } z_state;
  72
  73         union {
  74                 struct {
  75                         char *buf; /* from read_object() */
  76                         unsigned long read_ptr;
  77                 } incore;
  78
  79                 struct {
  80                         void *mapped;
  81                         unsigned long mapsize;
  82                         char hdr[32];
  83                         int hdr_avail;
  84                         int hdr_used;
  85                 } loose;
  86
  87                 struct {
  88                         struct packed_git *pack;
  89                         off_t pos;
  90                 } in_pack;
  91
  92                 struct filtered_istream filtered;
  93         } u;
  94 };
  95
  96 int close_istream(struct git_istream *st)
  97 {
  98         int r = st->vtbl->close(st);
  99         free(st);
 100         return r;
 101 }
 102
 103 ssize_t read_istream(struct git_istream *st, void *buf, size_t sz)
 104 {
 105         return st->vtbl->read(st, buf, sz);
 106 }
 107
 108 static enum input_source istream_source(const unsigned char *sha1,
 109                                         enum object_type *type,
 110                                         struct object_info *oi)
 111 {
 112         unsigned long size;
 113         int status;
 114         struct object_id oid;
 115
 116         hashcpy(oid.hash, sha1);
 117
 118         oi->typep = type;
 119         oi->sizep = &size;
 120         status = oid_object_info_extended(&oid, oi, 0);
 121         if (status < 0)
 122                 return stream_error;
 123
 124         switch (oi->whence) {
 125         case OI_LOOSE:
 126                 return loose;
 127         case OI_PACKED:
 128                 if (!oi->u.packed.is_delta && big_file_threshold < size)
 129                         return pack_non_delta;
 130                 /* fallthru */
 131         default:
 132                 return incore;
 133         }
 134 }
 135
 136 struct git_istream *open_istream(const struct object_id *oid,
 137                                  enum object_type *type,
 138                                  unsigned long *size,
 139                                  struct stream_filter *filter)
 140 {
 141         struct git_istream *st;
 142         struct object_info oi = OBJECT_INFO_INIT;
 143         const unsigned char *real = lookup_replace_object(oid->hash);
 144         enum input_source src = istream_source(real, type, &oi);
 145
 146         if (src < 0)
 147                 return NULL;
 148
 149         st = xmalloc(sizeof(*st));
 150         if (open_istream_tbl[src](st, &oi, real, type)) {
 151                 if (open_istream_incore(st, &oi, real, type)) {
 152                         free(st);
 153                         return NULL;
 154                 }
 155         }
 156         if (filter) {
 157                 /* Add "&& !is_null_stream_filter(filter)" for performance */
 158                 struct git_istream *nst = attach_stream_filter(st, filter);
 159                 if (!nst) {
 160                         close_istream(st);
 161                         return NULL;
 162                 }
 163                 st = nst;
 164         }
 165
 166         *size = st->size;
 167         return st;
 168 }
 169
 170
 171 /*****************************************************************
 172  *
 173  * Common helpers
 174  *
 175  *****************************************************************/
 176
 177 static void close_deflated_stream(struct git_istream *st)
 178 {
 179         if (st->z_state == z_used)
 180                 git_inflate_end(&st->z);
 181 }
 182
 183
 184 /*****************************************************************
 185  *
 186  * Filtered stream
 187  *
 188  *****************************************************************/
 189
 190 static close_method_decl(filtered)
 191 {
 192         free_stream_filter(st->u.filtered.filter);
 193         return close_istream(st->u.filtered.upstream);
 194 }
 195
 196 static read_method_decl(filtered)
 197 {
 198         struct filtered_istream *fs = &(st->u.filtered);
 199         size_t filled = 0;
 200
 201         while (sz) {
 202                 /* do we already have filtered output? */
 203                 if (fs->o_ptr < fs->o_end) {
 204                         size_t to_move = fs->o_end - fs->o_ptr;
 205                         if (sz < to_move)
 206                                 to_move = sz;
 207                         memcpy(buf + filled, fs->obuf + fs->o_ptr, to_move);
 208                         fs->o_ptr += to_move;
 209                         sz -= to_move;
 210                         filled += to_move;
 211                         continue;
 212                 }
 213                 fs->o_end = fs->o_ptr = 0;
 214
 215                 /* do we have anything to feed the filter with? */
 216                 if (fs->i_ptr < fs->i_end) {
 217                         size_t to_feed = fs->i_end - fs->i_ptr;
 218                         size_t to_receive = FILTER_BUFFER;
 219                         if (stream_filter(fs->filter,
 220                                           fs->ibuf + fs->i_ptr, &to_feed,
 221                                           fs->obuf, &to_receive))
 222                                 return -1;
 223                         fs->i_ptr = fs->i_end - to_feed;
 224                         fs->o_end = FILTER_BUFFER - to_receive;
 225                         continue;
 226                 }
 227
 228                 /* tell the filter to drain upon no more input */
 229                 if (fs->input_finished) {
 230                         size_t to_receive = FILTER_BUFFER;
 231                         if (stream_filter(fs->filter,
 232                                           NULL, NULL,
 233                                           fs->obuf, &to_receive))
 234                                 return -1;
 235                         fs->o_end = FILTER_BUFFER - to_receive;
 236                         if (!fs->o_end)
 237                                 break;
 238                         continue;
 239                 }
 240                 fs->i_end = fs->i_ptr = 0;
 241
 242                 /* refill the input from the upstream */
 243                 if (!fs->input_finished) {
 244                         fs->i_end = read_istream(fs->upstream, fs->ibuf, FILTER_BUFFER);
 245                         if (fs->i_end < 0)
 246                                 return -1;
 247                         if (fs->i_end)
 248                                 continue;
 249                 }
 250                 fs->input_finished = 1;
 251         }
 252         return filled;
 253 }
 254
 255 static struct stream_vtbl filtered_vtbl = {
 256         close_istream_filtered,
 257         read_istream_filtered,
 258 };
 259
 260 static struct git_istream *attach_stream_filter(struct git_istream *st,
 261                                                 struct stream_filter *filter)
 262 {
 263         struct git_istream *ifs = xmalloc(sizeof(*ifs));
 264         struct filtered_istream *fs = &(ifs->u.filtered);
 265
 266         ifs->vtbl = &filtered_vtbl;
 267         fs->upstream = st;
 268         fs->filter = filter;
 269         fs->i_end = fs->i_ptr = 0;
 270         fs->o_end = fs->o_ptr = 0;
 271         fs->input_finished = 0;
 272         ifs->size = -1; /* unknown */
 273         return ifs;
 274 }
 275
 276 /*****************************************************************
 277  *
 278  * Loose object stream
 279  *
 280  *****************************************************************/
 281
 282 static read_method_decl(loose)
 283 {
 284         size_t total_read = 0;
 285
 286         switch (st->z_state) {
 287         case z_done:
 288                 return 0;
 289         case z_error:
 290                 return -1;
 291         default:
 292                 break;
 293         }
 294
 295         if (st->u.loose.hdr_used < st->u.loose.hdr_avail) {
 296                 size_t to_copy = st->u.loose.hdr_avail - st->u.loose.hdr_used;
 297                 if (sz < to_copy)
 298                         to_copy = sz;
 299                 memcpy(buf, st->u.loose.hdr + st->u.loose.hdr_used, to_copy);
 300                 st->u.loose.hdr_used += to_copy;
 301                 total_read += to_copy;
 302         }
 303
 304         while (total_read < sz) {
 305                 int status;
 306
 307                 st->z.next_out = (unsigned char *)buf + total_read;
 308                 st->z.avail_out = sz - total_read;
 309                 status = git_inflate(&st->z, Z_FINISH);
 310
 311                 total_read = st->z.next_out - (unsigned char *)buf;
 312
 313                 if (status == Z_STREAM_END) {
 314                         git_inflate_end(&st->z);
 315                         st->z_state = z_done;
 316                         break;
 317                 }
 318                 if (status != Z_OK && (status != Z_BUF_ERROR || total_read < sz)) {
 319                         git_inflate_end(&st->z);
 320                         st->z_state = z_error;
 321                         return -1;
 322                 }
 323         }
 324         return total_read;
 325 }
 326
 327 static close_method_decl(loose)
 328 {
 329         close_deflated_stream(st);
 330         munmap(st->u.loose.mapped, st->u.loose.mapsize);
 331         return 0;
 332 }
 333
 334 static struct stream_vtbl loose_vtbl = {
 335         close_istream_loose,
 336         read_istream_loose,
 337 };
 338
 339 static open_method_decl(loose)
 340 {
 341         st->u.loose.mapped = map_sha1_file(sha1, &st->u.loose.mapsize);
 342         if (!st->u.loose.mapped)
 343                 return -1;
 344         if ((unpack_sha1_header(&st->z,
 345                                 st->u.loose.mapped,
 346                                 st->u.loose.mapsize,
 347                                 st->u.loose.hdr,
 348                                 sizeof(st->u.loose.hdr)) < 0) ||
 349             (parse_sha1_header(st->u.loose.hdr, &st->size) < 0)) {
 350                 git_inflate_end(&st->z);
 351                 munmap(st->u.loose.mapped, st->u.loose.mapsize);
 352                 return -1;
 353         }
 354
 355         st->u.loose.hdr_used = strlen(st->u.loose.hdr) + 1;
 356         st->u.loose.hdr_avail = st->z.total_out;
 357         st->z_state = z_used;
 358
 359         st->vtbl = &loose_vtbl;
 360         return 0;
 361 }
 362
 363
 364 /*****************************************************************
 365  *
 366  * Non-delta packed object stream
 367  *
 368  *****************************************************************/
 369
 370 static read_method_decl(pack_non_delta)
 371 {
 372         size_t total_read = 0;
 373
 374         switch (st->z_state) {
 375         case z_unused:
 376                 memset(&st->z, 0, sizeof(st->z));
 377                 git_inflate_init(&st->z);
 378                 st->z_state = z_used;
 379                 break;
 380         case z_done:
 381                 return 0;
 382         case z_error:
 383                 return -1;
 384         case z_used:
 385                 break;
 386         }
 387
 388         while (total_read < sz) {
 389                 int status;
 390                 struct pack_window *window = NULL;
 391                 unsigned char *mapped;
 392
 393                 mapped = use_pack(st->u.in_pack.pack, &window,
 394                                   st->u.in_pack.pos, &st->z.avail_in);
 395
 396                 st->z.next_out = (unsigned char *)buf + total_read;
 397                 st->z.avail_out = sz - total_read;
 398                 st->z.next_in = mapped;
 399                 status = git_inflate(&st->z, Z_FINISH);
 400
 401                 st->u.in_pack.pos += st->z.next_in - mapped;
 402                 total_read = st->z.next_out - (unsigned char *)buf;
 403                 unuse_pack(&window);
 404
 405                 if (status == Z_STREAM_END) {
 406                         git_inflate_end(&st->z);
 407                         st->z_state = z_done;
 408                         break;
 409                 }
 410                 if (status != Z_OK && status != Z_BUF_ERROR) {
 411                         git_inflate_end(&st->z);
 412                         st->z_state = z_error;
 413                         return -1;
 414                 }
 415         }
 416         return total_read;
 417 }
 418
 419 static close_method_decl(pack_non_delta)
 420 {
 421         close_deflated_stream(st);
 422         return 0;
 423 }
 424
 425 static struct stream_vtbl pack_non_delta_vtbl = {
 426         close_istream_pack_non_delta,
 427         read_istream_pack_non_delta,
 428 };
 429
 430 static open_method_decl(pack_non_delta)
 431 {
 432         struct pack_window *window;
 433         enum object_type in_pack_type;
 434
 435         st->u.in_pack.pack = oi->u.packed.pack;
 436         st->u.in_pack.pos = oi->u.packed.offset;
 437         window = NULL;
 438
 439         in_pack_type = unpack_object_header(st->u.in_pack.pack,
 440                                             &window,
 441                                             &st->u.in_pack.pos,
 442                                             &st->size);
 443         unuse_pack(&window);
 444         switch (in_pack_type) {
 445         default:
 446                 return -1; /* we do not do deltas for now */
 447         case OBJ_COMMIT:
 448         case OBJ_TREE:
 449         case OBJ_BLOB:
 450         case OBJ_TAG:
 451                 break;
 452         }
 453         st->z_state = z_unused;
 454         st->vtbl = &pack_non_delta_vtbl;
 455         return 0;
 456 }
 457
 458
 459 /*****************************************************************
 460  *
 461  * In-core stream
 462  *
 463  *****************************************************************/
 464
 465 static close_method_decl(incore)
 466 {
 467         free(st->u.incore.buf);
 468         return 0;
 469 }
 470
 471 static read_method_decl(incore)
 472 {
 473         size_t read_size = sz;
 474         size_t remainder = st->size - st->u.incore.read_ptr;
 475
 476         if (remainder <= read_size)
 477                 read_size = remainder;
 478         if (read_size) {
 479                 memcpy(buf, st->u.incore.buf + st->u.incore.read_ptr, read_size);
 480                 st->u.incore.read_ptr += read_size;
 481         }
 482         return read_size;
 483 }
 484
 485 static struct stream_vtbl incore_vtbl = {
 486         close_istream_incore,
 487         read_istream_incore,
 488 };
 489
 490 static open_method_decl(incore)
 491 {
 492         st->u.incore.buf = read_sha1_file_extended(sha1, type, &st->size, 0);
 493         st->u.incore.read_ptr = 0;
 494         st->vtbl = &incore_vtbl;
 495
 496         return st->u.incore.buf ? 0 : -1;
 497 }
 498
 499
 500 /****************************************************************
 501  * Users of streaming interface
 502  ****************************************************************/
 503
 504 int stream_blob_to_fd(int fd, const struct object_id *oid, struct stream_filter *filter,
 505                       int can_seek)
 506 {
 507         struct git_istream *st;
 508         enum object_type type;
 509         unsigned long sz;
 510         ssize_t kept = 0;
 511         int result = -1;
 512
 513         st = open_istream(oid, &type, &sz, filter);
 514         if (!st) {
 515                 if (filter)
 516                         free_stream_filter(filter);
 517                 return result;
 518         }
 519         if (type != OBJ_BLOB)
 520                 goto close_and_exit;
 521         for (;;) {
 522                 char buf[1024 * 16];
 523                 ssize_t wrote, holeto;
 524                 ssize_t readlen = read_istream(st, buf, sizeof(buf));
 525
 526                 if (readlen < 0)
 527                         goto close_and_exit;
 528                 if (!readlen)
 529                         break;
 530                 if (can_seek && sizeof(buf) == readlen) {
 531                         for (holeto = 0; holeto < readlen; holeto++)
 532                                 if (buf[holeto])
 533                                         break;
 534                         if (readlen == holeto) {
 535                                 kept += holeto;
 536                                 continue;
 537                         }
 538                 }
 539
 540                 if (kept && lseek(fd, kept, SEEK_CUR) == (off_t) -1)
 541                         goto close_and_exit;
 542                 else
 543                         kept = 0;
 544                 wrote = write_in_full(fd, buf, readlen);
 545
 546                 if (wrote < 0)
 547                         goto close_and_exit;
 548         }
 549         if (kept && (lseek(fd, kept - 1, SEEK_CUR) == (off_t) -1 ||
 550                      xwrite(fd, "", 1) != 1))
 551                 goto close_and_exit;
 552         result = 0;
 553
 554  close_and_exit:
 555         close_istream(st);
 556         return result;
 557 }