streaming.c

   1 /*
   2  * Copyright (c) 2011, Google Inc.
   3  */
   4 #include "cache.h"
   5 #include "streaming.h"
   6 #include "repository.h"
   7 #include "object-store.h"
   8 #include "replace-object.h"
   9 #include "packfile.h"
  10
  11 enum input_source {
  12         stream_error = -1,
  13         incore = 0,
  14         loose = 1,
  15         pack_non_delta = 2
  16 };
  17
  18 typedef int (*open_istream_fn)(struct git_istream *,
  19                                struct repository *,
  20                                struct object_info *,
  21                                const struct object_id *,
  22                                enum object_type *);
  23 typedef int (*close_istream_fn)(struct git_istream *);
  24 typedef ssize_t (*read_istream_fn)(struct git_istream *, char *, size_t);
  25
  26 struct stream_vtbl {
  27         close_istream_fn close;
  28         read_istream_fn read;
  29 };
  30
  31 #define open_method_decl(name) \
  32         int open_istream_ ##name \
  33         (struct git_istream *st, struct repository *r, \
  34          struct object_info *oi, const struct object_id *oid, \
  35          enum object_type *type)
  36
  37 #define close_method_decl(name) \
  38         int close_istream_ ##name \
  39         (struct git_istream *st)
  40
  41 #define read_method_decl(name) \
  42         ssize_t read_istream_ ##name \
  43         (struct git_istream *st, char *buf, size_t sz)
  44
  45 #define FILTER_BUFFER (1024*16)
  46
  47 struct filtered_istream {
  48         struct git_istream *upstream;
  49         struct stream_filter *filter;
  50         char ibuf[FILTER_BUFFER];
  51         char obuf[FILTER_BUFFER];
  52         int i_end, i_ptr;
  53         int o_end, o_ptr;
  54         int input_finished;
  55 };
  56
  57 struct git_istream {
  58         const struct stream_vtbl *vtbl;
  59         unsigned long size; /* inflated size of full object */
  60         git_zstream z;
  61         enum { z_unused, z_used, z_done, z_error } z_state;
  62
  63         union {
  64                 struct {
  65                         char *buf; /* from read_object() */
  66                         unsigned long read_ptr;
  67                 } incore;
  68
  69                 struct {
  70                         void *mapped;
  71                         unsigned long mapsize;
  72                         char hdr[32];
  73                         int hdr_avail;
  74                         int hdr_used;
  75                 } loose;
  76
  77                 struct {
  78                         struct packed_git *pack;
  79                         off_t pos;
  80                 } in_pack;
  81
  82                 struct filtered_istream filtered;
  83         } u;
  84 };
  85
  86 /*****************************************************************
  87  *
  88  * Common helpers
  89  *
  90  *****************************************************************/
  91
  92 static void close_deflated_stream(struct git_istream *st)
  93 {
  94         if (st->z_state == z_used)
  95                 git_inflate_end(&st->z);
  96 }
  97
  98
  99 /*****************************************************************
 100  *
 101  * Filtered stream
 102  *
 103  *****************************************************************/
 104
 105 static close_method_decl(filtered)
 106 {
 107         free_stream_filter(st->u.filtered.filter);
 108         return close_istream(st->u.filtered.upstream);
 109 }
 110
 111 static read_method_decl(filtered)
 112 {
 113         struct filtered_istream *fs = &(st->u.filtered);
 114         size_t filled = 0;
 115
 116         while (sz) {
 117                 /* do we already have filtered output? */
 118                 if (fs->o_ptr < fs->o_end) {
 119                         size_t to_move = fs->o_end - fs->o_ptr;
 120                         if (sz < to_move)
 121                                 to_move = sz;
 122                         memcpy(buf + filled, fs->obuf + fs->o_ptr, to_move);
 123                         fs->o_ptr += to_move;
 124                         sz -= to_move;
 125                         filled += to_move;
 126                         continue;
 127                 }
 128                 fs->o_end = fs->o_ptr = 0;
 129
 130                 /* do we have anything to feed the filter with? */
 131                 if (fs->i_ptr < fs->i_end) {
 132                         size_t to_feed = fs->i_end - fs->i_ptr;
 133                         size_t to_receive = FILTER_BUFFER;
 134                         if (stream_filter(fs->filter,
 135                                           fs->ibuf + fs->i_ptr, &to_feed,
 136                                           fs->obuf, &to_receive))
 137                                 return -1;
 138                         fs->i_ptr = fs->i_end - to_feed;
 139                         fs->o_end = FILTER_BUFFER - to_receive;
 140                         continue;
 141                 }
 142
 143                 /* tell the filter to drain upon no more input */
 144                 if (fs->input_finished) {
 145                         size_t to_receive = FILTER_BUFFER;
 146                         if (stream_filter(fs->filter,
 147                                           NULL, NULL,
 148                                           fs->obuf, &to_receive))
 149                                 return -1;
 150                         fs->o_end = FILTER_BUFFER - to_receive;
 151                         if (!fs->o_end)
 152                                 break;
 153                         continue;
 154                 }
 155                 fs->i_end = fs->i_ptr = 0;
 156
 157                 /* refill the input from the upstream */
 158                 if (!fs->input_finished) {
 159                         fs->i_end = read_istream(fs->upstream, fs->ibuf, FILTER_BUFFER);
 160                         if (fs->i_end < 0)
 161                                 return -1;
 162                         if (fs->i_end)
 163                                 continue;
 164                 }
 165                 fs->input_finished = 1;
 166         }
 167         return filled;
 168 }
 169
 170 static struct stream_vtbl filtered_vtbl = {
 171         close_istream_filtered,
 172         read_istream_filtered,
 173 };
 174
 175 static struct git_istream *attach_stream_filter(struct git_istream *st,
 176                                                 struct stream_filter *filter)
 177 {
 178         struct git_istream *ifs = xmalloc(sizeof(*ifs));
 179         struct filtered_istream *fs = &(ifs->u.filtered);
 180
 181         ifs->vtbl = &filtered_vtbl;
 182         fs->upstream = st;
 183         fs->filter = filter;
 184         fs->i_end = fs->i_ptr = 0;
 185         fs->o_end = fs->o_ptr = 0;
 186         fs->input_finished = 0;
 187         ifs->size = -1; /* unknown */
 188         return ifs;
 189 }
 190
 191 /*****************************************************************
 192  *
 193  * Loose object stream
 194  *
 195  *****************************************************************/
 196
 197 static read_method_decl(loose)
 198 {
 199         size_t total_read = 0;
 200
 201         switch (st->z_state) {
 202         case z_done:
 203                 return 0;
 204         case z_error:
 205                 return -1;
 206         default:
 207                 break;
 208         }
 209
 210         if (st->u.loose.hdr_used < st->u.loose.hdr_avail) {
 211                 size_t to_copy = st->u.loose.hdr_avail - st->u.loose.hdr_used;
 212                 if (sz < to_copy)
 213                         to_copy = sz;
 214                 memcpy(buf, st->u.loose.hdr + st->u.loose.hdr_used, to_copy);
 215                 st->u.loose.hdr_used += to_copy;
 216                 total_read += to_copy;
 217         }
 218
 219         while (total_read < sz) {
 220                 int status;
 221
 222                 st->z.next_out = (unsigned char *)buf + total_read;
 223                 st->z.avail_out = sz - total_read;
 224                 status = git_inflate(&st->z, Z_FINISH);
 225
 226                 total_read = st->z.next_out - (unsigned char *)buf;
 227
 228                 if (status == Z_STREAM_END) {
 229                         git_inflate_end(&st->z);
 230                         st->z_state = z_done;
 231                         break;
 232                 }
 233                 if (status != Z_OK && (status != Z_BUF_ERROR || total_read < sz)) {
 234                         git_inflate_end(&st->z);
 235                         st->z_state = z_error;
 236                         return -1;
 237                 }
 238         }
 239         return total_read;
 240 }
 241
 242 static close_method_decl(loose)
 243 {
 244         close_deflated_stream(st);
 245         munmap(st->u.loose.mapped, st->u.loose.mapsize);
 246         return 0;
 247 }
 248
 249 static struct stream_vtbl loose_vtbl = {
 250         close_istream_loose,
 251         read_istream_loose,
 252 };
 253
 254 static open_method_decl(loose)
 255 {
 256         st->u.loose.mapped = map_loose_object(r, oid, &st->u.loose.mapsize);
 257         if (!st->u.loose.mapped)
 258                 return -1;
 259         if ((unpack_loose_header(&st->z,
 260                                  st->u.loose.mapped,
 261                                  st->u.loose.mapsize,
 262                                  st->u.loose.hdr,
 263                                  sizeof(st->u.loose.hdr)) < 0) ||
 264             (parse_loose_header(st->u.loose.hdr, &st->size) < 0)) {
 265                 git_inflate_end(&st->z);
 266                 munmap(st->u.loose.mapped, st->u.loose.mapsize);
 267                 return -1;
 268         }
 269
 270         st->u.loose.hdr_used = strlen(st->u.loose.hdr) + 1;
 271         st->u.loose.hdr_avail = st->z.total_out;
 272         st->z_state = z_used;
 273
 274         st->vtbl = &loose_vtbl;
 275         return 0;
 276 }
 277
 278
 279 /*****************************************************************
 280  *
 281  * Non-delta packed object stream
 282  *
 283  *****************************************************************/
 284
 285 static read_method_decl(pack_non_delta)
 286 {
 287         size_t total_read = 0;
 288
 289         switch (st->z_state) {
 290         case z_unused:
 291                 memset(&st->z, 0, sizeof(st->z));
 292                 git_inflate_init(&st->z);
 293                 st->z_state = z_used;
 294                 break;
 295         case z_done:
 296                 return 0;
 297         case z_error:
 298                 return -1;
 299         case z_used:
 300                 break;
 301         }
 302
 303         while (total_read < sz) {
 304                 int status;
 305                 struct pack_window *window = NULL;
 306                 unsigned char *mapped;
 307
 308                 mapped = use_pack(st->u.in_pack.pack, &window,
 309                                   st->u.in_pack.pos, &st->z.avail_in);
 310
 311                 st->z.next_out = (unsigned char *)buf + total_read;
 312                 st->z.avail_out = sz - total_read;
 313                 st->z.next_in = mapped;
 314                 status = git_inflate(&st->z, Z_FINISH);
 315
 316                 st->u.in_pack.pos += st->z.next_in - mapped;
 317                 total_read = st->z.next_out - (unsigned char *)buf;
 318                 unuse_pack(&window);
 319
 320                 if (status == Z_STREAM_END) {
 321                         git_inflate_end(&st->z);
 322                         st->z_state = z_done;
 323                         break;
 324                 }
 325
 326                 /*
 327                  * Unlike the loose object case, we do not have to worry here
 328                  * about running out of input bytes and spinning infinitely. If
 329                  * we get Z_BUF_ERROR due to too few input bytes, then we'll
 330                  * replenish them in the next use_pack() call when we loop. If
 331                  * we truly hit the end of the pack (i.e., because it's corrupt
 332                  * or truncated), then use_pack() catches that and will die().
 333                  */
 334                 if (status != Z_OK && status != Z_BUF_ERROR) {
 335                         git_inflate_end(&st->z);
 336                         st->z_state = z_error;
 337                         return -1;
 338                 }
 339         }
 340         return total_read;
 341 }
 342
 343 static close_method_decl(pack_non_delta)
 344 {
 345         close_deflated_stream(st);
 346         return 0;
 347 }
 348
 349 static struct stream_vtbl pack_non_delta_vtbl = {
 350         close_istream_pack_non_delta,
 351         read_istream_pack_non_delta,
 352 };
 353
 354 static open_method_decl(pack_non_delta)
 355 {
 356         struct pack_window *window;
 357         enum object_type in_pack_type;
 358
 359         st->u.in_pack.pack = oi->u.packed.pack;
 360         st->u.in_pack.pos = oi->u.packed.offset;
 361         window = NULL;
 362
 363         in_pack_type = unpack_object_header(st->u.in_pack.pack,
 364                                             &window,
 365                                             &st->u.in_pack.pos,
 366                                             &st->size);
 367         unuse_pack(&window);
 368         switch (in_pack_type) {
 369         default:
 370                 return -1; /* we do not do deltas for now */
 371         case OBJ_COMMIT:
 372         case OBJ_TREE:
 373         case OBJ_BLOB:
 374         case OBJ_TAG:
 375                 break;
 376         }
 377         st->z_state = z_unused;
 378         st->vtbl = &pack_non_delta_vtbl;
 379         return 0;
 380 }
 381
 382
 383 /*****************************************************************
 384  *
 385  * In-core stream
 386  *
 387  *****************************************************************/
 388
 389 static close_method_decl(incore)
 390 {
 391         free(st->u.incore.buf);
 392         return 0;
 393 }
 394
 395 static read_method_decl(incore)
 396 {
 397         size_t read_size = sz;
 398         size_t remainder = st->size - st->u.incore.read_ptr;
 399
 400         if (remainder <= read_size)
 401                 read_size = remainder;
 402         if (read_size) {
 403                 memcpy(buf, st->u.incore.buf + st->u.incore.read_ptr, read_size);
 404                 st->u.incore.read_ptr += read_size;
 405         }
 406         return read_size;
 407 }
 408
 409 static struct stream_vtbl incore_vtbl = {
 410         close_istream_incore,
 411         read_istream_incore,
 412 };
 413
 414 static open_method_decl(incore)
 415 {
 416         st->u.incore.buf = read_object_file_extended(r, oid, type, &st->size, 0);
 417         st->u.incore.read_ptr = 0;
 418         st->vtbl = &incore_vtbl;
 419
 420         return st->u.incore.buf ? 0 : -1;
 421 }
 422
 423 /*****************************************************************************
 424  * static helpers variables and functions for users of streaming interface
 425  *****************************************************************************/
 426
 427 static open_istream_fn open_istream_tbl[] = {
 428         open_istream_incore,
 429         open_istream_loose,
 430         open_istream_pack_non_delta,
 431 };
 432
 433 static enum input_source istream_source(struct repository *r,
 434                                         const struct object_id *oid,
 435                                         enum object_type *type,
 436                                         struct object_info *oi)
 437 {
 438         unsigned long size;
 439         int status;
 440
 441         oi->typep = type;
 442         oi->sizep = &size;
 443         status = oid_object_info_extended(r, oid, oi, 0);
 444         if (status < 0)
 445                 return stream_error;
 446
 447         switch (oi->whence) {
 448         case OI_LOOSE:
 449                 return loose;
 450         case OI_PACKED:
 451                 if (!oi->u.packed.is_delta && big_file_threshold < size)
 452                         return pack_non_delta;
 453                 /* fallthru */
 454         default:
 455                 return incore;
 456         }
 457 }
 458
 459
 460 /****************************************************************
 461  * Users of streaming interface
 462  ****************************************************************/
 463
 464 int close_istream(struct git_istream *st)
 465 {
 466         int r = st->vtbl->close(st);
 467         free(st);
 468         return r;
 469 }
 470
 471 ssize_t read_istream(struct git_istream *st, void *buf, size_t sz)
 472 {
 473         return st->vtbl->read(st, buf, sz);
 474 }
 475
 476 struct git_istream *open_istream(struct repository *r,
 477                                  const struct object_id *oid,
 478                                  enum object_type *type,
 479                                  unsigned long *size,
 480                                  struct stream_filter *filter)
 481 {
 482         struct git_istream *st;
 483         struct object_info oi = OBJECT_INFO_INIT;
 484         const struct object_id *real = lookup_replace_object(r, oid);
 485         enum input_source src = istream_source(r, real, type, &oi);
 486
 487         if (src < 0)
 488                 return NULL;
 489
 490         st = xmalloc(sizeof(*st));
 491         if (open_istream_tbl[src](st, r, &oi, real, type)) {
 492                 if (open_istream_incore(st, r, &oi, real, type)) {
 493                         free(st);
 494                         return NULL;
 495                 }
 496         }
 497         if (filter) {
 498                 /* Add "&& !is_null_stream_filter(filter)" for performance */
 499                 struct git_istream *nst = attach_stream_filter(st, filter);
 500                 if (!nst) {
 501                         close_istream(st);
 502                         return NULL;
 503                 }
 504                 st = nst;
 505         }
 506
 507         *size = st->size;
 508         return st;
 509 }
 510
 511 int stream_blob_to_fd(int fd, const struct object_id *oid, struct stream_filter *filter,
 512                       int can_seek)
 513 {
 514         struct git_istream *st;
 515         enum object_type type;
 516         unsigned long sz;
 517         ssize_t kept = 0;
 518         int result = -1;
 519
 520         st = open_istream(the_repository, oid, &type, &sz, filter);
 521         if (!st) {
 522                 if (filter)
 523                         free_stream_filter(filter);
 524                 return result;
 525         }
 526         if (type != OBJ_BLOB)
 527                 goto close_and_exit;
 528         for (;;) {
 529                 char buf[1024 * 16];
 530                 ssize_t wrote, holeto;
 531                 ssize_t readlen = read_istream(st, buf, sizeof(buf));
 532
 533                 if (readlen < 0)
 534                         goto close_and_exit;
 535                 if (!readlen)
 536                         break;
 537                 if (can_seek && sizeof(buf) == readlen) {
 538                         for (holeto = 0; holeto < readlen; holeto++)
 539                                 if (buf[holeto])
 540                                         break;
 541                         if (readlen == holeto) {
 542                                 kept += holeto;
 543                                 continue;
 544                         }
 545                 }
 546
 547                 if (kept && lseek(fd, kept, SEEK_CUR) == (off_t) -1)
 548                         goto close_and_exit;
 549                 else
 550                         kept = 0;
 551                 wrote = write_in_full(fd, buf, readlen);
 552
 553                 if (wrote < 0)
 554                         goto close_and_exit;
 555         }
 556         if (kept && (lseek(fd, kept - 1, SEEK_CUR) == (off_t) -1 ||
 557                      xwrite(fd, "", 1) != 1))
 558                 goto close_and_exit;
 559         result = 0;
 560
 561  close_and_exit:
 562         close_istream(st);
 563         return result;
 564 }