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