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 /* forward declaration */
  46 static open_method_decl(incore);
  47 static open_method_decl(loose);
  48 static open_method_decl(pack_non_delta);
  49 static struct git_istream *attach_stream_filter(struct git_istream *st,
  50                                                 struct stream_filter *filter);
  51
  52
  53 static open_istream_fn open_istream_tbl[] = {
  54         open_istream_incore,
  55         open_istream_loose,
  56         open_istream_pack_non_delta,
  57 };
  58
  59 #define FILTER_BUFFER (1024*16)
  60
  61 struct filtered_istream {
  62         struct git_istream *upstream;
  63         struct stream_filter *filter;
  64         char ibuf[FILTER_BUFFER];
  65         char obuf[FILTER_BUFFER];
  66         int i_end, i_ptr;
  67         int o_end, o_ptr;
  68         int input_finished;
  69 };
  70
  71 struct git_istream {
  72         const struct stream_vtbl *vtbl;
  73         unsigned long size; /* inflated size of full object */
  74         git_zstream z;
  75         enum { z_unused, z_used, z_done, z_error } z_state;
  76
  77         union {
  78                 struct {
  79                         char *buf; /* from read_object() */
  80                         unsigned long read_ptr;
  81                 } incore;
  82
  83                 struct {
  84                         void *mapped;
  85                         unsigned long mapsize;
  86                         char hdr[32];
  87                         int hdr_avail;
  88                         int hdr_used;
  89                 } loose;
  90
  91                 struct {
  92                         struct packed_git *pack;
  93                         off_t pos;
  94                 } in_pack;
  95
  96                 struct filtered_istream filtered;
  97         } u;
  98 };
  99
 100 int close_istream(struct git_istream *st)
 101 {
 102         int r = st->vtbl->close(st);
 103         free(st);
 104         return r;
 105 }
 106
 107 ssize_t read_istream(struct git_istream *st, void *buf, size_t sz)
 108 {
 109         return st->vtbl->read(st, buf, sz);
 110 }
 111
 112 static enum input_source istream_source(struct repository *r,
 113                                         const struct object_id *oid,
 114                                         enum object_type *type,
 115                                         struct object_info *oi)
 116 {
 117         unsigned long size;
 118         int status;
 119
 120         oi->typep = type;
 121         oi->sizep = &size;
 122         status = oid_object_info_extended(r, oid, oi, 0);
 123         if (status < 0)
 124                 return stream_error;
 125
 126         switch (oi->whence) {
 127         case OI_LOOSE:
 128                 return loose;
 129         case OI_PACKED:
 130                 if (!oi->u.packed.is_delta && big_file_threshold < size)
 131                         return pack_non_delta;
 132                 /* fallthru */
 133         default:
 134                 return incore;
 135         }
 136 }
 137
 138 struct git_istream *open_istream(struct repository *r,
 139                                  const struct object_id *oid,
 140                                  enum object_type *type,
 141                                  unsigned long *size,
 142                                  struct stream_filter *filter)
 143 {
 144         struct git_istream *st;
 145         struct object_info oi = OBJECT_INFO_INIT;
 146         const struct object_id *real = lookup_replace_object(r, oid);
 147         enum input_source src = istream_source(r, real, type, &oi);
 148
 149         if (src < 0)
 150                 return NULL;
 151
 152         st = xmalloc(sizeof(*st));
 153         if (open_istream_tbl[src](st, r, &oi, real, type)) {
 154                 if (open_istream_incore(st, r, &oi, real, type)) {
 155                         free(st);
 156                         return NULL;
 157                 }
 158         }
 159         if (filter) {
 160                 /* Add "&& !is_null_stream_filter(filter)" for performance */
 161                 struct git_istream *nst = attach_stream_filter(st, filter);
 162                 if (!nst) {
 163                         close_istream(st);
 164                         return NULL;
 165                 }
 166                 st = nst;
 167         }
 168
 169         *size = st->size;
 170         return st;
 171 }
 172
 173
 174 /*****************************************************************
 175  *
 176  * Common helpers
 177  *
 178  *****************************************************************/
 179
 180 static void close_deflated_stream(struct git_istream *st)
 181 {
 182         if (st->z_state == z_used)
 183                 git_inflate_end(&st->z);
 184 }
 185
 186
 187 /*****************************************************************
 188  *
 189  * Filtered stream
 190  *
 191  *****************************************************************/
 192
 193 static close_method_decl(filtered)
 194 {
 195         free_stream_filter(st->u.filtered.filter);
 196         return close_istream(st->u.filtered.upstream);
 197 }
 198
 199 static read_method_decl(filtered)
 200 {
 201         struct filtered_istream *fs = &(st->u.filtered);
 202         size_t filled = 0;
 203
 204         while (sz) {
 205                 /* do we already have filtered output? */
 206                 if (fs->o_ptr < fs->o_end) {
 207                         size_t to_move = fs->o_end - fs->o_ptr;
 208                         if (sz < to_move)
 209                                 to_move = sz;
 210                         memcpy(buf + filled, fs->obuf + fs->o_ptr, to_move);
 211                         fs->o_ptr += to_move;
 212                         sz -= to_move;
 213                         filled += to_move;
 214                         continue;
 215                 }
 216                 fs->o_end = fs->o_ptr = 0;
 217
 218                 /* do we have anything to feed the filter with? */
 219                 if (fs->i_ptr < fs->i_end) {
 220                         size_t to_feed = fs->i_end - fs->i_ptr;
 221                         size_t to_receive = FILTER_BUFFER;
 222                         if (stream_filter(fs->filter,
 223                                           fs->ibuf + fs->i_ptr, &to_feed,
 224                                           fs->obuf, &to_receive))
 225                                 return -1;
 226                         fs->i_ptr = fs->i_end - to_feed;
 227                         fs->o_end = FILTER_BUFFER - to_receive;
 228                         continue;
 229                 }
 230
 231                 /* tell the filter to drain upon no more input */
 232                 if (fs->input_finished) {
 233                         size_t to_receive = FILTER_BUFFER;
 234                         if (stream_filter(fs->filter,
 235                                           NULL, NULL,
 236                                           fs->obuf, &to_receive))
 237                                 return -1;
 238                         fs->o_end = FILTER_BUFFER - to_receive;
 239                         if (!fs->o_end)
 240                                 break;
 241                         continue;
 242                 }
 243                 fs->i_end = fs->i_ptr = 0;
 244
 245                 /* refill the input from the upstream */
 246                 if (!fs->input_finished) {
 247                         fs->i_end = read_istream(fs->upstream, fs->ibuf, FILTER_BUFFER);
 248                         if (fs->i_end < 0)
 249                                 return -1;
 250                         if (fs->i_end)
 251                                 continue;
 252                 }
 253                 fs->input_finished = 1;
 254         }
 255         return filled;
 256 }
 257
 258 static struct stream_vtbl filtered_vtbl = {
 259         close_istream_filtered,
 260         read_istream_filtered,
 261 };
 262
 263 static struct git_istream *attach_stream_filter(struct git_istream *st,
 264                                                 struct stream_filter *filter)
 265 {
 266         struct git_istream *ifs = xmalloc(sizeof(*ifs));
 267         struct filtered_istream *fs = &(ifs->u.filtered);
 268
 269         ifs->vtbl = &filtered_vtbl;
 270         fs->upstream = st;
 271         fs->filter = filter;
 272         fs->i_end = fs->i_ptr = 0;
 273         fs->o_end = fs->o_ptr = 0;
 274         fs->input_finished = 0;
 275         ifs->size = -1; /* unknown */
 276         return ifs;
 277 }
 278
 279 /*****************************************************************
 280  *
 281  * Loose object stream
 282  *
 283  *****************************************************************/
 284
 285 static read_method_decl(loose)
 286 {
 287         size_t total_read = 0;
 288
 289         switch (st->z_state) {
 290         case z_done:
 291                 return 0;
 292         case z_error:
 293                 return -1;
 294         default:
 295                 break;
 296         }
 297
 298         if (st->u.loose.hdr_used < st->u.loose.hdr_avail) {
 299                 size_t to_copy = st->u.loose.hdr_avail - st->u.loose.hdr_used;
 300                 if (sz < to_copy)
 301                         to_copy = sz;
 302                 memcpy(buf, st->u.loose.hdr + st->u.loose.hdr_used, to_copy);
 303                 st->u.loose.hdr_used += to_copy;
 304                 total_read += to_copy;
 305         }
 306
 307         while (total_read < sz) {
 308                 int status;
 309
 310                 st->z.next_out = (unsigned char *)buf + total_read;
 311                 st->z.avail_out = sz - total_read;
 312                 status = git_inflate(&st->z, Z_FINISH);
 313
 314                 total_read = st->z.next_out - (unsigned char *)buf;
 315
 316                 if (status == Z_STREAM_END) {
 317                         git_inflate_end(&st->z);
 318                         st->z_state = z_done;
 319                         break;
 320                 }
 321                 if (status != Z_OK && (status != Z_BUF_ERROR || total_read < sz)) {
 322                         git_inflate_end(&st->z);
 323                         st->z_state = z_error;
 324                         return -1;
 325                 }
 326         }
 327         return total_read;
 328 }
 329
 330 static close_method_decl(loose)
 331 {
 332         close_deflated_stream(st);
 333         munmap(st->u.loose.mapped, st->u.loose.mapsize);
 334         return 0;
 335 }
 336
 337 static struct stream_vtbl loose_vtbl = {
 338         close_istream_loose,
 339         read_istream_loose,
 340 };
 341
 342 static open_method_decl(loose)
 343 {
 344         st->u.loose.mapped = map_loose_object(r, oid, &st->u.loose.mapsize);
 345         if (!st->u.loose.mapped)
 346                 return -1;
 347         if ((unpack_loose_header(&st->z,
 348                                  st->u.loose.mapped,
 349                                  st->u.loose.mapsize,
 350                                  st->u.loose.hdr,
 351                                  sizeof(st->u.loose.hdr)) < 0) ||
 352             (parse_loose_header(st->u.loose.hdr, &st->size) < 0)) {
 353                 git_inflate_end(&st->z);
 354                 munmap(st->u.loose.mapped, st->u.loose.mapsize);
 355                 return -1;
 356         }
 357
 358         st->u.loose.hdr_used = strlen(st->u.loose.hdr) + 1;
 359         st->u.loose.hdr_avail = st->z.total_out;
 360         st->z_state = z_used;
 361
 362         st->vtbl = &loose_vtbl;
 363         return 0;
 364 }
 365
 366
 367 /*****************************************************************
 368  *
 369  * Non-delta packed object stream
 370  *
 371  *****************************************************************/
 372
 373 static read_method_decl(pack_non_delta)
 374 {
 375         size_t total_read = 0;
 376
 377         switch (st->z_state) {
 378         case z_unused:
 379                 memset(&st->z, 0, sizeof(st->z));
 380                 git_inflate_init(&st->z);
 381                 st->z_state = z_used;
 382                 break;
 383         case z_done:
 384                 return 0;
 385         case z_error:
 386                 return -1;
 387         case z_used:
 388                 break;
 389         }
 390
 391         while (total_read < sz) {
 392                 int status;
 393                 struct pack_window *window = NULL;
 394                 unsigned char *mapped;
 395
 396                 mapped = use_pack(st->u.in_pack.pack, &window,
 397                                   st->u.in_pack.pos, &st->z.avail_in);
 398
 399                 st->z.next_out = (unsigned char *)buf + total_read;
 400                 st->z.avail_out = sz - total_read;
 401                 st->z.next_in = mapped;
 402                 status = git_inflate(&st->z, Z_FINISH);
 403
 404                 st->u.in_pack.pos += st->z.next_in - mapped;
 405                 total_read = st->z.next_out - (unsigned char *)buf;
 406                 unuse_pack(&window);
 407
 408                 if (status == Z_STREAM_END) {
 409                         git_inflate_end(&st->z);
 410                         st->z_state = z_done;
 411                         break;
 412                 }
 413
 414                 /*
 415                  * Unlike the loose object case, we do not have to worry here
 416                  * about running out of input bytes and spinning infinitely. If
 417                  * we get Z_BUF_ERROR due to too few input bytes, then we'll
 418                  * replenish them in the next use_pack() call when we loop. If
 419                  * we truly hit the end of the pack (i.e., because it's corrupt
 420                  * or truncated), then use_pack() catches that and will die().
 421                  */
 422                 if (status != Z_OK && status != Z_BUF_ERROR) {
 423                         git_inflate_end(&st->z);
 424                         st->z_state = z_error;
 425                         return -1;
 426                 }
 427         }
 428         return total_read;
 429 }
 430
 431 static close_method_decl(pack_non_delta)
 432 {
 433         close_deflated_stream(st);
 434         return 0;
 435 }
 436
 437 static struct stream_vtbl pack_non_delta_vtbl = {
 438         close_istream_pack_non_delta,
 439         read_istream_pack_non_delta,
 440 };
 441
 442 static open_method_decl(pack_non_delta)
 443 {
 444         struct pack_window *window;
 445         enum object_type in_pack_type;
 446
 447         st->u.in_pack.pack = oi->u.packed.pack;
 448         st->u.in_pack.pos = oi->u.packed.offset;
 449         window = NULL;
 450
 451         in_pack_type = unpack_object_header(st->u.in_pack.pack,
 452                                             &window,
 453                                             &st->u.in_pack.pos,
 454                                             &st->size);
 455         unuse_pack(&window);
 456         switch (in_pack_type) {
 457         default:
 458                 return -1; /* we do not do deltas for now */
 459         case OBJ_COMMIT:
 460         case OBJ_TREE:
 461         case OBJ_BLOB:
 462         case OBJ_TAG:
 463                 break;
 464         }
 465         st->z_state = z_unused;
 466         st->vtbl = &pack_non_delta_vtbl;
 467         return 0;
 468 }
 469
 470
 471 /*****************************************************************
 472  *
 473  * In-core stream
 474  *
 475  *****************************************************************/
 476
 477 static close_method_decl(incore)
 478 {
 479         free(st->u.incore.buf);
 480         return 0;
 481 }
 482
 483 static read_method_decl(incore)
 484 {
 485         size_t read_size = sz;
 486         size_t remainder = st->size - st->u.incore.read_ptr;
 487
 488         if (remainder <= read_size)
 489                 read_size = remainder;
 490         if (read_size) {
 491                 memcpy(buf, st->u.incore.buf + st->u.incore.read_ptr, read_size);
 492                 st->u.incore.read_ptr += read_size;
 493         }
 494         return read_size;
 495 }
 496
 497 static struct stream_vtbl incore_vtbl = {
 498         close_istream_incore,
 499         read_istream_incore,
 500 };
 501
 502 static open_method_decl(incore)
 503 {
 504         st->u.incore.buf = read_object_file_extended(r, oid, type, &st->size, 0);
 505         st->u.incore.read_ptr = 0;
 506         st->vtbl = &incore_vtbl;
 507
 508         return st->u.incore.buf ? 0 : -1;
 509 }
 510
 511
 512 /****************************************************************
 513  * Users of streaming interface
 514  ****************************************************************/
 515
 516 int stream_blob_to_fd(int fd, const struct object_id *oid, struct stream_filter *filter,
 517                       int can_seek)
 518 {
 519         struct git_istream *st;
 520         enum object_type type;
 521         unsigned long sz;
 522         ssize_t kept = 0;
 523         int result = -1;
 524
 525         st = open_istream(the_repository, oid, &type, &sz, filter);
 526         if (!st) {
 527                 if (filter)
 528                         free_stream_filter(filter);
 529                 return result;
 530         }
 531         if (type != OBJ_BLOB)
 532                 goto close_and_exit;
 533         for (;;) {
 534                 char buf[1024 * 16];
 535                 ssize_t wrote, holeto;
 536                 ssize_t readlen = read_istream(st, buf, sizeof(buf));
 537
 538                 if (readlen < 0)
 539                         goto close_and_exit;
 540                 if (!readlen)
 541                         break;
 542                 if (can_seek && sizeof(buf) == readlen) {
 543                         for (holeto = 0; holeto < readlen; holeto++)
 544                                 if (buf[holeto])
 545                                         break;
 546                         if (readlen == holeto) {
 547                                 kept += holeto;
 548                                 continue;
 549                         }
 550                 }
 551
 552                 if (kept && lseek(fd, kept, SEEK_CUR) == (off_t) -1)
 553                         goto close_and_exit;
 554                 else
 555                         kept = 0;
 556                 wrote = write_in_full(fd, buf, readlen);
 557
 558                 if (wrote < 0)
 559                         goto close_and_exit;
 560         }
 561         if (kept && (lseek(fd, kept - 1, SEEK_CUR) == (off_t) -1 ||
 562                      xwrite(fd, "", 1) != 1))
 563                 goto close_and_exit;
 564         result = 0;
 565
 566  close_and_exit:
 567         close_istream(st);
 568         return result;
 569 }