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         return st->vtbl->close(st);
  98 }
  99
 100 ssize_t read_istream(struct git_istream *st, char *buf, size_t sz)
 101 {
 102         return st->vtbl->read(st, buf, sz);
 103 }
 104
 105 static enum input_source istream_source(const unsigned char *sha1,
 106                                         enum object_type *type,
 107                                         struct object_info *oi)
 108 {
 109         unsigned long size;
 110         int status;
 111
 112         oi->sizep = &size;
 113         status = sha1_object_info_extended(sha1, oi);
 114         if (status < 0)
 115                 return stream_error;
 116         *type = status;
 117
 118         switch (oi->whence) {
 119         case OI_LOOSE:
 120                 return loose;
 121         case OI_PACKED:
 122                 if (!oi->u.packed.is_delta && big_file_threshold <= size)
 123                         return pack_non_delta;
 124                 /* fallthru */
 125         default:
 126                 return incore;
 127         }
 128 }
 129
 130 struct git_istream *open_istream(const unsigned char *sha1,
 131                                  enum object_type *type,
 132                                  unsigned long *size,
 133                                  struct stream_filter *filter)
 134 {
 135         struct git_istream *st;
 136         struct object_info oi;
 137         const unsigned char *real = lookup_replace_object(sha1);
 138         enum input_source src = istream_source(real, type, &oi);
 139
 140         if (src < 0)
 141                 return NULL;
 142
 143         st = xmalloc(sizeof(*st));
 144         if (open_istream_tbl[src](st, &oi, real, type)) {
 145                 if (open_istream_incore(st, &oi, real, type)) {
 146                         free(st);
 147                         return NULL;
 148                 }
 149         }
 150         if (st && filter) {
 151                 /* Add "&& !is_null_stream_filter(filter)" for performance */
 152                 struct git_istream *nst = attach_stream_filter(st, filter);
 153                 if (!nst)
 154                         close_istream(st);
 155                 st = nst;
 156         }
 157
 158         *size = st->size;
 159         return st;
 160 }
 161
 162
 163 /*****************************************************************
 164  *
 165  * Common helpers
 166  *
 167  *****************************************************************/
 168
 169 static void close_deflated_stream(struct git_istream *st)
 170 {
 171         if (st->z_state == z_used)
 172                 git_inflate_end(&st->z);
 173 }
 174
 175
 176 /*****************************************************************
 177  *
 178  * Filtered stream
 179  *
 180  *****************************************************************/
 181
 182 static close_method_decl(filtered)
 183 {
 184         free_stream_filter(st->u.filtered.filter);
 185         return close_istream(st->u.filtered.upstream);
 186 }
 187
 188 static read_method_decl(filtered)
 189 {
 190         struct filtered_istream *fs = &(st->u.filtered);
 191         size_t filled = 0;
 192
 193         while (sz) {
 194                 /* do we already have filtered output? */
 195                 if (fs->o_ptr < fs->o_end) {
 196                         size_t to_move = fs->o_end - fs->o_ptr;
 197                         if (sz < to_move)
 198                                 to_move = sz;
 199                         memcpy(buf + filled, fs->obuf + fs->o_ptr, to_move);
 200                         fs->o_ptr += to_move;
 201                         sz -= to_move;
 202                         filled += to_move;
 203                         continue;
 204                 }
 205                 fs->o_end = fs->o_ptr = 0;
 206
 207                 /* do we have anything to feed the filter with? */
 208                 if (fs->i_ptr < fs->i_end) {
 209                         size_t to_feed = fs->i_end - fs->i_ptr;
 210                         size_t to_receive = FILTER_BUFFER;
 211                         if (stream_filter(fs->filter,
 212                                           fs->ibuf + fs->i_ptr, &to_feed,
 213                                           fs->obuf, &to_receive))
 214                                 return -1;
 215                         fs->i_ptr = fs->i_end - to_feed;
 216                         fs->o_end = FILTER_BUFFER - to_receive;
 217                         continue;
 218                 }
 219
 220                 /* tell the filter to drain upon no more input */
 221                 if (fs->input_finished) {
 222                         size_t to_receive = FILTER_BUFFER;
 223                         if (stream_filter(fs->filter,
 224                                           NULL, NULL,
 225                                           fs->obuf, &to_receive))
 226                                 return -1;
 227                         fs->o_end = FILTER_BUFFER - to_receive;
 228                         if (!fs->o_end)
 229                                 break;
 230                         continue;
 231                 }
 232                 fs->i_end = fs->i_ptr = 0;
 233
 234                 /* refill the input from the upstream */
 235                 if (!fs->input_finished) {
 236                         fs->i_end = read_istream(fs->upstream, fs->ibuf, FILTER_BUFFER);
 237                         if (fs->i_end < 0)
 238                                 break;
 239                         if (fs->i_end)
 240                                 continue;
 241                 }
 242                 fs->input_finished = 1;
 243         }
 244         return filled;
 245 }
 246
 247 static struct stream_vtbl filtered_vtbl = {
 248         close_istream_filtered,
 249         read_istream_filtered,
 250 };
 251
 252 static struct git_istream *attach_stream_filter(struct git_istream *st,
 253                                                 struct stream_filter *filter)
 254 {
 255         struct git_istream *ifs = xmalloc(sizeof(*ifs));
 256         struct filtered_istream *fs = &(ifs->u.filtered);
 257
 258         ifs->vtbl = &filtered_vtbl;
 259         fs->upstream = st;
 260         fs->filter = filter;
 261         fs->i_end = fs->i_ptr = 0;
 262         fs->o_end = fs->o_ptr = 0;
 263         fs->input_finished = 0;
 264         ifs->size = -1; /* unknown */
 265         return ifs;
 266 }
 267
 268 /*****************************************************************
 269  *
 270  * Loose object stream
 271  *
 272  *****************************************************************/
 273
 274 static read_method_decl(loose)
 275 {
 276         size_t total_read = 0;
 277
 278         switch (st->z_state) {
 279         case z_done:
 280                 return 0;
 281         case z_error:
 282                 return -1;
 283         default:
 284                 break;
 285         }
 286
 287         if (st->u.loose.hdr_used < st->u.loose.hdr_avail) {
 288                 size_t to_copy = st->u.loose.hdr_avail - st->u.loose.hdr_used;
 289                 if (sz < to_copy)
 290                         to_copy = sz;
 291                 memcpy(buf, st->u.loose.hdr + st->u.loose.hdr_used, to_copy);
 292                 st->u.loose.hdr_used += to_copy;
 293                 total_read += to_copy;
 294         }
 295
 296         while (total_read < sz) {
 297                 int status;
 298
 299                 st->z.next_out = (unsigned char *)buf + total_read;
 300                 st->z.avail_out = sz - total_read;
 301                 status = git_inflate(&st->z, Z_FINISH);
 302
 303                 total_read = st->z.next_out - (unsigned char *)buf;
 304
 305                 if (status == Z_STREAM_END) {
 306                         git_inflate_end(&st->z);
 307                         st->z_state = z_done;
 308                         break;
 309                 }
 310                 if (status != Z_OK && status != Z_BUF_ERROR) {
 311                         git_inflate_end(&st->z);
 312                         st->z_state = z_error;
 313                         return -1;
 314                 }
 315         }
 316         return total_read;
 317 }
 318
 319 static close_method_decl(loose)
 320 {
 321         close_deflated_stream(st);
 322         munmap(st->u.loose.mapped, st->u.loose.mapsize);
 323         return 0;
 324 }
 325
 326 static struct stream_vtbl loose_vtbl = {
 327         close_istream_loose,
 328         read_istream_loose,
 329 };
 330
 331 static open_method_decl(loose)
 332 {
 333         st->u.loose.mapped = map_sha1_file(sha1, &st->u.loose.mapsize);
 334         if (!st->u.loose.mapped)
 335                 return -1;
 336         if (unpack_sha1_header(&st->z,
 337                                st->u.loose.mapped,
 338                                st->u.loose.mapsize,
 339                                st->u.loose.hdr,
 340                                sizeof(st->u.loose.hdr)) < 0) {
 341                 git_inflate_end(&st->z);
 342                 munmap(st->u.loose.mapped, st->u.loose.mapsize);
 343                 return -1;
 344         }
 345
 346         parse_sha1_header(st->u.loose.hdr, &st->size);
 347         st->u.loose.hdr_used = strlen(st->u.loose.hdr) + 1;
 348         st->u.loose.hdr_avail = st->z.total_out;
 349         st->z_state = z_used;
 350
 351         st->vtbl = &loose_vtbl;
 352         return 0;
 353 }
 354
 355
 356 /*****************************************************************
 357  *
 358  * Non-delta packed object stream
 359  *
 360  *****************************************************************/
 361
 362 static read_method_decl(pack_non_delta)
 363 {
 364         size_t total_read = 0;
 365
 366         switch (st->z_state) {
 367         case z_unused:
 368                 memset(&st->z, 0, sizeof(st->z));
 369                 git_inflate_init(&st->z);
 370                 st->z_state = z_used;
 371                 break;
 372         case z_done:
 373                 return 0;
 374         case z_error:
 375                 return -1;
 376         case z_used:
 377                 break;
 378         }
 379
 380         while (total_read < sz) {
 381                 int status;
 382                 struct pack_window *window = NULL;
 383                 unsigned char *mapped;
 384
 385                 mapped = use_pack(st->u.in_pack.pack, &window,
 386                                   st->u.in_pack.pos, &st->z.avail_in);
 387
 388                 st->z.next_out = (unsigned char *)buf + total_read;
 389                 st->z.avail_out = sz - total_read;
 390                 st->z.next_in = mapped;
 391                 status = git_inflate(&st->z, Z_FINISH);
 392
 393                 st->u.in_pack.pos += st->z.next_in - mapped;
 394                 total_read = st->z.next_out - (unsigned char *)buf;
 395                 unuse_pack(&window);
 396
 397                 if (status == Z_STREAM_END) {
 398                         git_inflate_end(&st->z);
 399                         st->z_state = z_done;
 400                         break;
 401                 }
 402                 if (status != Z_OK && status != Z_BUF_ERROR) {
 403                         git_inflate_end(&st->z);
 404                         st->z_state = z_error;
 405                         return -1;
 406                 }
 407         }
 408         return total_read;
 409 }
 410
 411 static close_method_decl(pack_non_delta)
 412 {
 413         close_deflated_stream(st);
 414         return 0;
 415 }
 416
 417 static struct stream_vtbl pack_non_delta_vtbl = {
 418         close_istream_pack_non_delta,
 419         read_istream_pack_non_delta,
 420 };
 421
 422 static open_method_decl(pack_non_delta)
 423 {
 424         struct pack_window *window;
 425         enum object_type in_pack_type;
 426
 427         st->u.in_pack.pack = oi->u.packed.pack;
 428         st->u.in_pack.pos = oi->u.packed.offset;
 429         window = NULL;
 430
 431         in_pack_type = unpack_object_header(st->u.in_pack.pack,
 432                                             &window,
 433                                             &st->u.in_pack.pos,
 434                                             &st->size);
 435         unuse_pack(&window);
 436         switch (in_pack_type) {
 437         default:
 438                 return -1; /* we do not do deltas for now */
 439         case OBJ_COMMIT:
 440         case OBJ_TREE:
 441         case OBJ_BLOB:
 442         case OBJ_TAG:
 443                 break;
 444         }
 445         st->z_state = z_unused;
 446         st->vtbl = &pack_non_delta_vtbl;
 447         return 0;
 448 }
 449
 450
 451 /*****************************************************************
 452  *
 453  * In-core stream
 454  *
 455  *****************************************************************/
 456
 457 static close_method_decl(incore)
 458 {
 459         free(st->u.incore.buf);
 460         return 0;
 461 }
 462
 463 static read_method_decl(incore)
 464 {
 465         size_t read_size = sz;
 466         size_t remainder = st->size - st->u.incore.read_ptr;
 467
 468         if (remainder <= read_size)
 469                 read_size = remainder;
 470         if (read_size) {
 471                 memcpy(buf, st->u.incore.buf + st->u.incore.read_ptr, read_size);
 472                 st->u.incore.read_ptr += read_size;
 473         }
 474         return read_size;
 475 }
 476
 477 static struct stream_vtbl incore_vtbl = {
 478         close_istream_incore,
 479         read_istream_incore,
 480 };
 481
 482 static open_method_decl(incore)
 483 {
 484         st->u.incore.buf = read_sha1_file_extended(sha1, type, &st->size, 0);
 485         st->u.incore.read_ptr = 0;
 486         st->vtbl = &incore_vtbl;
 487
 488         return st->u.incore.buf ? 0 : -1;
 489 }