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         struct object_info oi = OBJECT_INFO_INIT;
 227         oi.sizep = &st->size;
 228
 229         st->u.loose.mapped = map_loose_object(r, oid, &st->u.loose.mapsize);
 230         if (!st->u.loose.mapped)
 231                 return -1;
 232         switch (unpack_loose_header(&st->z, st->u.loose.mapped,
 233                                     st->u.loose.mapsize, st->u.loose.hdr,
 234                                     sizeof(st->u.loose.hdr), NULL)) {
 235         case ULHR_OK:
 236                 break;
 237         case ULHR_BAD:
 238                 goto error;
 239         }
 240         if (parse_loose_header(st->u.loose.hdr, &oi, 0) < 0)
 241                 goto error;
 242
 243         st->u.loose.hdr_used = strlen(st->u.loose.hdr) + 1;
 244         st->u.loose.hdr_avail = st->z.total_out;
 245         st->z_state = z_used;
 246         st->close = close_istream_loose;
 247         st->read = read_istream_loose;
 248
 249         return 0;
 250 error:
 251         git_inflate_end(&st->z);
 252         munmap(st->u.loose.mapped, st->u.loose.mapsize);
 253         return -1;
 254 }
 255
 256
 257 /*****************************************************************
 258  *
 259  * Non-delta packed object stream
 260  *
 261  *****************************************************************/
 262
 263 static ssize_t read_istream_pack_non_delta(struct git_istream *st, char *buf,
 264                                            size_t sz)
 265 {
 266         size_t total_read = 0;
 267
 268         switch (st->z_state) {
 269         case z_unused:
 270                 memset(&st->z, 0, sizeof(st->z));
 271                 git_inflate_init(&st->z);
 272                 st->z_state = z_used;
 273                 break;
 274         case z_done:
 275                 return 0;
 276         case z_error:
 277                 return -1;
 278         case z_used:
 279                 break;
 280         }
 281
 282         while (total_read < sz) {
 283                 int status;
 284                 struct pack_window *window = NULL;
 285                 unsigned char *mapped;
 286
 287                 mapped = use_pack(st->u.in_pack.pack, &window,
 288                                   st->u.in_pack.pos, &st->z.avail_in);
 289
 290                 st->z.next_out = (unsigned char *)buf + total_read;
 291                 st->z.avail_out = sz - total_read;
 292                 st->z.next_in = mapped;
 293                 status = git_inflate(&st->z, Z_FINISH);
 294
 295                 st->u.in_pack.pos += st->z.next_in - mapped;
 296                 total_read = st->z.next_out - (unsigned char *)buf;
 297                 unuse_pack(&window);
 298
 299                 if (status == Z_STREAM_END) {
 300                         git_inflate_end(&st->z);
 301                         st->z_state = z_done;
 302                         break;
 303                 }
 304
 305                 /*
 306                  * Unlike the loose object case, we do not have to worry here
 307                  * about running out of input bytes and spinning infinitely. If
 308                  * we get Z_BUF_ERROR due to too few input bytes, then we'll
 309                  * replenish them in the next use_pack() call when we loop. If
 310                  * we truly hit the end of the pack (i.e., because it's corrupt
 311                  * or truncated), then use_pack() catches that and will die().
 312                  */
 313                 if (status != Z_OK && status != Z_BUF_ERROR) {
 314                         git_inflate_end(&st->z);
 315                         st->z_state = z_error;
 316                         return -1;
 317                 }
 318         }
 319         return total_read;
 320 }
 321
 322 static int close_istream_pack_non_delta(struct git_istream *st)
 323 {
 324         close_deflated_stream(st);
 325         return 0;
 326 }
 327
 328 static int open_istream_pack_non_delta(struct git_istream *st,
 329                                        struct repository *r,
 330                                        const struct object_id *oid,
 331                                        enum object_type *type)
 332 {
 333         struct pack_window *window;
 334         enum object_type in_pack_type;
 335
 336         window = NULL;
 337
 338         in_pack_type = unpack_object_header(st->u.in_pack.pack,
 339                                             &window,
 340                                             &st->u.in_pack.pos,
 341                                             &st->size);
 342         unuse_pack(&window);
 343         switch (in_pack_type) {
 344         default:
 345                 return -1; /* we do not do deltas for now */
 346         case OBJ_COMMIT:
 347         case OBJ_TREE:
 348         case OBJ_BLOB:
 349         case OBJ_TAG:
 350                 break;
 351         }
 352         st->z_state = z_unused;
 353         st->close = close_istream_pack_non_delta;
 354         st->read = read_istream_pack_non_delta;
 355
 356         return 0;
 357 }
 358
 359
 360 /*****************************************************************
 361  *
 362  * In-core stream
 363  *
 364  *****************************************************************/
 365
 366 static int close_istream_incore(struct git_istream *st)
 367 {
 368         free(st->u.incore.buf);
 369         return 0;
 370 }
 371
 372 static ssize_t read_istream_incore(struct git_istream *st, char *buf, size_t sz)
 373 {
 374         size_t read_size = sz;
 375         size_t remainder = st->size - st->u.incore.read_ptr;
 376
 377         if (remainder <= read_size)
 378                 read_size = remainder;
 379         if (read_size) {
 380                 memcpy(buf, st->u.incore.buf + st->u.incore.read_ptr, read_size);
 381                 st->u.incore.read_ptr += read_size;
 382         }
 383         return read_size;
 384 }
 385
 386 static int open_istream_incore(struct git_istream *st, struct repository *r,
 387                                const struct object_id *oid, enum object_type *type)
 388 {
 389         st->u.incore.buf = read_object_file_extended(r, oid, type, &st->size, 0);
 390         st->u.incore.read_ptr = 0;
 391         st->close = close_istream_incore;
 392         st->read = read_istream_incore;
 393
 394         return st->u.incore.buf ? 0 : -1;
 395 }
 396
 397 /*****************************************************************************
 398  * static helpers variables and functions for users of streaming interface
 399  *****************************************************************************/
 400
 401 static int istream_source(struct git_istream *st,
 402                           struct repository *r,
 403                           const struct object_id *oid,
 404                           enum object_type *type)
 405 {
 406         unsigned long size;
 407         int status;
 408         struct object_info oi = OBJECT_INFO_INIT;
 409
 410         oi.typep = type;
 411         oi.sizep = &size;
 412         status = oid_object_info_extended(r, oid, &oi, 0);
 413         if (status < 0)
 414                 return status;
 415
 416         switch (oi.whence) {
 417         case OI_LOOSE:
 418                 st->open = open_istream_loose;
 419                 return 0;
 420         case OI_PACKED:
 421                 if (!oi.u.packed.is_delta && big_file_threshold < size) {
 422                         st->u.in_pack.pack = oi.u.packed.pack;
 423                         st->u.in_pack.pos = oi.u.packed.offset;
 424                         st->open = open_istream_pack_non_delta;
 425                         return 0;
 426                 }
 427                 /* fallthru */
 428         default:
 429                 st->open = open_istream_incore;
 430                 return 0;
 431         }
 432 }
 433
 434 /****************************************************************
 435  * Users of streaming interface
 436  ****************************************************************/
 437
 438 int close_istream(struct git_istream *st)
 439 {
 440         int r = st->close(st);
 441         free(st);
 442         return r;
 443 }
 444
 445 ssize_t read_istream(struct git_istream *st, void *buf, size_t sz)
 446 {
 447         return st->read(st, buf, sz);
 448 }
 449
 450 struct git_istream *open_istream(struct repository *r,
 451                                  const struct object_id *oid,
 452                                  enum object_type *type,
 453                                  unsigned long *size,
 454                                  struct stream_filter *filter)
 455 {
 456         struct git_istream *st = xmalloc(sizeof(*st));
 457         const struct object_id *real = lookup_replace_object(r, oid);
 458         int ret = istream_source(st, r, real, type);
 459
 460         if (ret) {
 461                 free(st);
 462                 return NULL;
 463         }
 464
 465         if (st->open(st, r, real, type)) {
 466                 if (open_istream_incore(st, r, real, type)) {
 467                         free(st);
 468                         return NULL;
 469                 }
 470         }
 471         if (filter) {
 472                 /* Add "&& !is_null_stream_filter(filter)" for performance */
 473                 struct git_istream *nst = attach_stream_filter(st, filter);
 474                 if (!nst) {
 475                         close_istream(st);
 476                         return NULL;
 477                 }
 478                 st = nst;
 479         }
 480
 481         *size = st->size;
 482         return st;
 483 }
 484
 485 int stream_blob_to_fd(int fd, const struct object_id *oid, struct stream_filter *filter,
 486                       int can_seek)
 487 {
 488         struct git_istream *st;
 489         enum object_type type;
 490         unsigned long sz;
 491         ssize_t kept = 0;
 492         int result = -1;
 493
 494         st = open_istream(the_repository, oid, &type, &sz, filter);
 495         if (!st) {
 496                 if (filter)
 497                         free_stream_filter(filter);
 498                 return result;
 499         }
 500         if (type != OBJ_BLOB)
 501                 goto close_and_exit;
 502         for (;;) {
 503                 char buf[1024 * 16];
 504                 ssize_t wrote, holeto;
 505                 ssize_t readlen = read_istream(st, buf, sizeof(buf));
 506
 507                 if (readlen < 0)
 508                         goto close_and_exit;
 509                 if (!readlen)
 510                         break;
 511                 if (can_seek && sizeof(buf) == readlen) {
 512                         for (holeto = 0; holeto < readlen; holeto++)
 513                                 if (buf[holeto])
 514                                         break;
 515                         if (readlen == holeto) {
 516                                 kept += holeto;
 517                                 continue;
 518                         }
 519                 }
 520
 521                 if (kept && lseek(fd, kept, SEEK_CUR) == (off_t) -1)
 522                         goto close_and_exit;
 523                 else
 524                         kept = 0;
 525                 wrote = write_in_full(fd, buf, readlen);
 526
 527                 if (wrote < 0)
 528                         goto close_and_exit;
 529         }
 530         if (kept && (lseek(fd, kept - 1, SEEK_CUR) == (off_t) -1 ||
 531                      xwrite(fd, "", 1) != 1))
 532                 goto close_and_exit;
 533         result = 0;
 534
 535  close_and_exit:
 536         close_istream(st);
 537         return result;
 538 }