streaming.c

   1 /*
   2  * Copyright (c) 2011, Google Inc.
   3  */
   4 #include "cache.h"
   5 #include "streaming.h"
   6
   7 enum input_source {
   8         stream_error = -1,
   9         incore = 0,
  10         loose = 1,
  11         pack_non_delta = 2
  12 };
  13
  14 typedef int (*open_istream_fn)(struct git_istream *,
  15                                struct object_info *,
  16                                const unsigned char *,
  17                                enum object_type *);
  18 typedef int (*close_istream_fn)(struct git_istream *);
  19 typedef ssize_t (*read_istream_fn)(struct git_istream *, char *, size_t);
  20
  21 struct stream_vtbl {
  22         close_istream_fn close;
  23         read_istream_fn read;
  24 };
  25
  26 #define open_method_decl(name) \
  27         int open_istream_ ##name \
  28         (struct git_istream *st, struct object_info *oi, \
  29          const unsigned char *sha1, \
  30          enum object_type *type)
  31
  32 #define close_method_decl(name) \
  33         int close_istream_ ##name \
  34         (struct git_istream *st)
  35
  36 #define read_method_decl(name) \
  37         ssize_t read_istream_ ##name \
  38         (struct git_istream *st, char *buf, size_t sz)
  39
  40 /* forward declaration */
  41 static open_method_decl(incore);
  42 static open_method_decl(loose);
  43 static open_method_decl(pack_non_delta);
  44 static struct git_istream *attach_stream_filter(struct git_istream *st,
  45                                                 struct stream_filter *filter);
  46
  47
  48 static open_istream_fn open_istream_tbl[] = {
  49         open_istream_incore,
  50         open_istream_loose,
  51         open_istream_pack_non_delta,
  52 };
  53
  54 #define FILTER_BUFFER (1024*16)
  55
  56 struct filtered_istream {
  57         struct git_istream *upstream;
  58         struct stream_filter *filter;
  59         char ibuf[FILTER_BUFFER];
  60         char obuf[FILTER_BUFFER];
  61         int i_end, i_ptr;
  62         int o_end, o_ptr;
  63         int input_finished;
  64 };
  65
  66 struct git_istream {
  67         const struct stream_vtbl *vtbl;
  68         unsigned long size; /* inflated size of full object */
  69         git_zstream z;
  70         enum { z_unused, z_used, z_done, z_error } z_state;
  71
  72         union {
  73                 struct {
  74                         char *buf; /* from read_object() */
  75                         unsigned long read_ptr;
  76                 } incore;
  77
  78                 struct {
  79                         void *mapped;
  80                         unsigned long mapsize;
  81                         char hdr[32];
  82                         int hdr_avail;
  83                         int hdr_used;
  84                 } loose;
  85
  86                 struct {
  87                         struct packed_git *pack;
  88                         off_t pos;
  89                 } in_pack;
  90
  91                 struct filtered_istream filtered;
  92         } u;
  93 };
  94
  95 int close_istream(struct git_istream *st)
  96 {
  97         int r = st->vtbl->close(st);
  98         free(st);
  99         return r;
 100 }
 101
 102 ssize_t read_istream(struct git_istream *st, void *buf, size_t sz)
 103 {
 104         return st->vtbl->read(st, buf, sz);
 105 }
 106
 107 static enum input_source istream_source(const unsigned char *sha1,
 108                                         enum object_type *type,
 109                                         struct object_info *oi)
 110 {
 111         unsigned long size;
 112         int status;
 113
 114         oi->typep = type;
 115         oi->sizep = &size;
 116         status = sha1_object_info_extended(sha1, oi, 0);
 117         if (status < 0)
 118                 return stream_error;
 119
 120         switch (oi->whence) {
 121         case OI_LOOSE:
 122                 return loose;
 123         case OI_PACKED:
 124                 if (!oi->u.packed.is_delta && big_file_threshold < size)
 125                         return pack_non_delta;
 126                 /* fallthru */
 127         default:
 128                 return incore;
 129         }
 130 }
 131
 132 struct git_istream *open_istream(const unsigned char *sha1,
 133                                  enum object_type *type,
 134                                  unsigned long *size,
 135                                  struct stream_filter *filter)
 136 {
 137         struct git_istream *st;
 138         struct object_info oi = OBJECT_INFO_INIT;
 139         const unsigned char *real = lookup_replace_object(sha1);
 140         enum input_source src = istream_source(real, type, &oi);
 141
 142         if (src < 0)
 143                 return NULL;
 144
 145         st = xmalloc(sizeof(*st));
 146         if (open_istream_tbl[src](st, &oi, real, type)) {
 147                 if (open_istream_incore(st, &oi, real, type)) {
 148                         free(st);
 149                         return NULL;
 150                 }
 151         }
 152         if (filter) {
 153                 /* Add "&& !is_null_stream_filter(filter)" for performance */
 154                 struct git_istream *nst = attach_stream_filter(st, filter);
 155                 if (!nst) {
 156                         close_istream(st);
 157                         return NULL;
 158                 }
 159                 st = nst;
 160         }
 161
 162         *size = st->size;
 163         return st;
 164 }
 165
 166
 167 /*****************************************************************
 168  *
 169  * Common helpers
 170  *
 171  *****************************************************************/
 172
 173 static void close_deflated_stream(struct git_istream *st)
 174 {
 175         if (st->z_state == z_used)
 176                 git_inflate_end(&st->z);
 177 }
 178
 179
 180 /*****************************************************************
 181  *
 182  * Filtered stream
 183  *
 184  *****************************************************************/
 185
 186 static close_method_decl(filtered)
 187 {
 188         free_stream_filter(st->u.filtered.filter);
 189         return close_istream(st->u.filtered.upstream);
 190 }
 191
 192 static read_method_decl(filtered)
 193 {
 194         struct filtered_istream *fs = &(st->u.filtered);
 195         size_t filled = 0;
 196
 197         while (sz) {
 198                 /* do we already have filtered output? */
 199                 if (fs->o_ptr < fs->o_end) {
 200                         size_t to_move = fs->o_end - fs->o_ptr;
 201                         if (sz < to_move)
 202                                 to_move = sz;
 203                         memcpy(buf + filled, fs->obuf + fs->o_ptr, to_move);
 204                         fs->o_ptr += to_move;
 205                         sz -= to_move;
 206                         filled += to_move;
 207                         continue;
 208                 }
 209                 fs->o_end = fs->o_ptr = 0;
 210
 211                 /* do we have anything to feed the filter with? */
 212                 if (fs->i_ptr < fs->i_end) {
 213                         size_t to_feed = fs->i_end - fs->i_ptr;
 214                         size_t to_receive = FILTER_BUFFER;
 215                         if (stream_filter(fs->filter,
 216                                           fs->ibuf + fs->i_ptr, &to_feed,
 217                                           fs->obuf, &to_receive))
 218                                 return -1;
 219                         fs->i_ptr = fs->i_end - to_feed;
 220                         fs->o_end = FILTER_BUFFER - to_receive;
 221                         continue;
 222                 }
 223
 224                 /* tell the filter to drain upon no more input */
 225                 if (fs->input_finished) {
 226                         size_t to_receive = FILTER_BUFFER;
 227                         if (stream_filter(fs->filter,
 228                                           NULL, NULL,
 229                                           fs->obuf, &to_receive))
 230                                 return -1;
 231                         fs->o_end = FILTER_BUFFER - to_receive;
 232                         if (!fs->o_end)
 233                                 break;
 234                         continue;
 235                 }
 236                 fs->i_end = fs->i_ptr = 0;
 237
 238                 /* refill the input from the upstream */
 239                 if (!fs->input_finished) {
 240                         fs->i_end = read_istream(fs->upstream, fs->ibuf, FILTER_BUFFER);
 241                         if (fs->i_end < 0)
 242                                 return -1;
 243                         if (fs->i_end)
 244                                 continue;
 245                 }
 246                 fs->input_finished = 1;
 247         }
 248         return filled;
 249 }
 250
 251 static struct stream_vtbl filtered_vtbl = {
 252         close_istream_filtered,
 253         read_istream_filtered,
 254 };
 255
 256 static struct git_istream *attach_stream_filter(struct git_istream *st,
 257                                                 struct stream_filter *filter)
 258 {
 259         struct git_istream *ifs = xmalloc(sizeof(*ifs));
 260         struct filtered_istream *fs = &(ifs->u.filtered);
 261
 262         ifs->vtbl = &filtered_vtbl;
 263         fs->upstream = st;
 264         fs->filter = filter;
 265         fs->i_end = fs->i_ptr = 0;
 266         fs->o_end = fs->o_ptr = 0;
 267         fs->input_finished = 0;
 268         ifs->size = -1; /* unknown */
 269         return ifs;
 270 }
 271
 272 /*****************************************************************
 273  *
 274  * Loose object stream
 275  *
 276  *****************************************************************/
 277
 278 static read_method_decl(loose)
 279 {
 280         size_t total_read = 0;
 281
 282         switch (st->z_state) {
 283         case z_done:
 284                 return 0;
 285         case z_error:
 286                 return -1;
 287         default:
 288                 break;
 289         }
 290
 291         if (st->u.loose.hdr_used < st->u.loose.hdr_avail) {
 292                 size_t to_copy = st->u.loose.hdr_avail - st->u.loose.hdr_used;
 293                 if (sz < to_copy)
 294                         to_copy = sz;
 295                 memcpy(buf, st->u.loose.hdr + st->u.loose.hdr_used, to_copy);
 296                 st->u.loose.hdr_used += to_copy;
 297                 total_read += to_copy;
 298         }
 299
 300         while (total_read < sz) {
 301                 int status;
 302
 303                 st->z.next_out = (unsigned char *)buf + total_read;
 304                 st->z.avail_out = sz - total_read;
 305                 status = git_inflate(&st->z, Z_FINISH);
 306
 307                 total_read = st->z.next_out - (unsigned char *)buf;
 308
 309                 if (status == Z_STREAM_END) {
 310                         git_inflate_end(&st->z);
 311                         st->z_state = z_done;
 312                         break;
 313                 }
 314                 if (status != Z_OK && (status != Z_BUF_ERROR || total_read < sz)) {
 315                         git_inflate_end(&st->z);
 316                         st->z_state = z_error;
 317                         return -1;
 318                 }
 319         }
 320         return total_read;
 321 }
 322
 323 static close_method_decl(loose)
 324 {
 325         close_deflated_stream(st);
 326         munmap(st->u.loose.mapped, st->u.loose.mapsize);
 327         return 0;
 328 }
 329
 330 static struct stream_vtbl loose_vtbl = {
 331         close_istream_loose,
 332         read_istream_loose,
 333 };
 334
 335 static open_method_decl(loose)
 336 {
 337         st->u.loose.mapped = map_sha1_file(sha1, &st->u.loose.mapsize);
 338         if (!st->u.loose.mapped)
 339                 return -1;
 340         if ((unpack_sha1_header(&st->z,
 341                                 st->u.loose.mapped,
 342                                 st->u.loose.mapsize,
 343                                 st->u.loose.hdr,
 344                                 sizeof(st->u.loose.hdr)) < 0) ||
 345             (parse_sha1_header(st->u.loose.hdr, &st->size) < 0)) {
 346                 git_inflate_end(&st->z);
 347                 munmap(st->u.loose.mapped, st->u.loose.mapsize);
 348                 return -1;
 349         }
 350
 351         st->u.loose.hdr_used = strlen(st->u.loose.hdr) + 1;
 352         st->u.loose.hdr_avail = st->z.total_out;
 353         st->z_state = z_used;
 354
 355         st->vtbl = &loose_vtbl;
 356         return 0;
 357 }
 358
 359
 360 /*****************************************************************
 361  *
 362  * Non-delta packed object stream
 363  *
 364  *****************************************************************/
 365
 366 static read_method_decl(pack_non_delta)
 367 {
 368         size_t total_read = 0;
 369
 370         switch (st->z_state) {
 371         case z_unused:
 372                 memset(&st->z, 0, sizeof(st->z));
 373                 git_inflate_init(&st->z);
 374                 st->z_state = z_used;
 375                 break;
 376         case z_done:
 377                 return 0;
 378         case z_error:
 379                 return -1;
 380         case z_used:
 381                 break;
 382         }
 383
 384         while (total_read < sz) {
 385                 int status;
 386                 struct pack_window *window = NULL;
 387                 unsigned char *mapped;
 388
 389                 mapped = use_pack(st->u.in_pack.pack, &window,
 390                                   st->u.in_pack.pos, &st->z.avail_in);
 391
 392                 st->z.next_out = (unsigned char *)buf + total_read;
 393                 st->z.avail_out = sz - total_read;
 394                 st->z.next_in = mapped;
 395                 status = git_inflate(&st->z, Z_FINISH);
 396
 397                 st->u.in_pack.pos += st->z.next_in - mapped;
 398                 total_read = st->z.next_out - (unsigned char *)buf;
 399                 unuse_pack(&window);
 400
 401                 if (status == Z_STREAM_END) {
 402                         git_inflate_end(&st->z);
 403                         st->z_state = z_done;
 404                         break;
 405                 }
 406                 if (status != Z_OK && status != Z_BUF_ERROR) {
 407                         git_inflate_end(&st->z);
 408                         st->z_state = z_error;
 409                         return -1;
 410                 }
 411         }
 412         return total_read;
 413 }
 414
 415 static close_method_decl(pack_non_delta)
 416 {
 417         close_deflated_stream(st);
 418         return 0;
 419 }
 420
 421 static struct stream_vtbl pack_non_delta_vtbl = {
 422         close_istream_pack_non_delta,
 423         read_istream_pack_non_delta,
 424 };
 425
 426 static open_method_decl(pack_non_delta)
 427 {
 428         struct pack_window *window;
 429         enum object_type in_pack_type;
 430
 431         st->u.in_pack.pack = oi->u.packed.pack;
 432         st->u.in_pack.pos = oi->u.packed.offset;
 433         window = NULL;
 434
 435         in_pack_type = unpack_object_header(st->u.in_pack.pack,
 436                                             &window,
 437                                             &st->u.in_pack.pos,
 438                                             &st->size);
 439         unuse_pack(&window);
 440         switch (in_pack_type) {
 441         default:
 442                 return -1; /* we do not do deltas for now */
 443         case OBJ_COMMIT:
 444         case OBJ_TREE:
 445         case OBJ_BLOB:
 446         case OBJ_TAG:
 447                 break;
 448         }
 449         st->z_state = z_unused;
 450         st->vtbl = &pack_non_delta_vtbl;
 451         return 0;
 452 }
 453
 454
 455 /*****************************************************************
 456  *
 457  * In-core stream
 458  *
 459  *****************************************************************/
 460
 461 static close_method_decl(incore)
 462 {
 463         free(st->u.incore.buf);
 464         return 0;
 465 }
 466
 467 static read_method_decl(incore)
 468 {
 469         size_t read_size = sz;
 470         size_t remainder = st->size - st->u.incore.read_ptr;
 471
 472         if (remainder <= read_size)
 473                 read_size = remainder;
 474         if (read_size) {
 475                 memcpy(buf, st->u.incore.buf + st->u.incore.read_ptr, read_size);
 476                 st->u.incore.read_ptr += read_size;
 477         }
 478         return read_size;
 479 }
 480
 481 static struct stream_vtbl incore_vtbl = {
 482         close_istream_incore,
 483         read_istream_incore,
 484 };
 485
 486 static open_method_decl(incore)
 487 {
 488         st->u.incore.buf = read_sha1_file_extended(sha1, type, &st->size, 0);
 489         st->u.incore.read_ptr = 0;
 490         st->vtbl = &incore_vtbl;
 491
 492         return st->u.incore.buf ? 0 : -1;
 493 }
 494
 495
 496 /****************************************************************
 497  * Users of streaming interface
 498  ****************************************************************/
 499
 500 int stream_blob_to_fd(int fd, const struct object_id *oid, struct stream_filter *filter,
 501                       int can_seek)
 502 {
 503         struct git_istream *st;
 504         enum object_type type;
 505         unsigned long sz;
 506         ssize_t kept = 0;
 507         int result = -1;
 508
 509         st = open_istream(oid->hash, &type, &sz, filter);
 510         if (!st) {
 511                 if (filter)
 512                         free_stream_filter(filter);
 513                 return result;
 514         }
 515         if (type != OBJ_BLOB)
 516                 goto close_and_exit;
 517         for (;;) {
 518                 char buf[1024 * 16];
 519                 ssize_t wrote, holeto;
 520                 ssize_t readlen = read_istream(st, buf, sizeof(buf));
 521
 522                 if (readlen < 0)
 523                         goto close_and_exit;
 524                 if (!readlen)
 525                         break;
 526                 if (can_seek && sizeof(buf) == readlen) {
 527                         for (holeto = 0; holeto < readlen; holeto++)
 528                                 if (buf[holeto])
 529                                         break;
 530                         if (readlen == holeto) {
 531                                 kept += holeto;
 532                                 continue;
 533                         }
 534                 }
 535
 536                 if (kept && lseek(fd, kept, SEEK_CUR) == (off_t) -1)
 537                         goto close_and_exit;
 538                 else
 539                         kept = 0;
 540                 wrote = write_in_full(fd, buf, readlen);
 541
 542                 if (wrote != readlen)
 543                         goto close_and_exit;
 544         }
 545         if (kept && (lseek(fd, kept - 1, SEEK_CUR) == (off_t) -1 ||
 546                      xwrite(fd, "", 1) != 1))
 547                 goto close_and_exit;
 548         result = 0;
 549
 550  close_and_exit:
 551         close_istream(st);
 552         return result;
 553 }