razf.c

   1 /*
   2  * RAZF : Random Access compressed(Z) File
   3  * Version: 1.0
   4  * Release Date: 2008-10-27
   5  *
   6  * Copyright 2008, Jue Ruan <ruanjue@gmail.com>, Heng Li <lh3@sanger.ac.uk>
   7  *
   8  * All rights reserved.
   9  *
  10  * Redistribution and use in source and binary forms, with or without
  11  * modification, are permitted provided that the following conditions
  12  * are met:
  13  * 1. Redistributions of source code must retain the above copyright
  14  *    notice, this list of conditions and the following disclaimer.
  15  * 2. Redistributions in binary form must reproduce the above copyright
  16  *    notice, this list of conditions and the following disclaimer in the
  17  *    documentation and/or other materials provided with the distribution.
  18  *
  19  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  29  * SUCH DAMAGE.
  30  */
  31
  32 #ifndef _NO_RAZF
  33
  34 #include <fcntl.h>
  35 #include <stdio.h>
  36 #include <stdlib.h>
  37 #include <string.h>
  38 #include <unistd.h>
  39 #include "razf.h"
  40
  41 #if ZLIB_VERNUM < 0x1221
  42 struct _gz_header_s {
  43     int     text;
  44     uLong   time;
  45     int     xflags;
  46     int     os;
  47     Bytef   *extra;
  48     uInt    extra_len;
  49     uInt    extra_max;
  50     Bytef   *name;
  51     uInt    name_max;
  52     Bytef   *comment;
  53     uInt    comm_max;
  54     int     hcrc;
  55     int     done;
  56 };
  57 #warning "zlib < 1.2.2.1; RAZF writing is disabled."
  58 #endif
  59
  60 #define DEF_MEM_LEVEL 8
  61
  62 static inline uint32_t byte_swap_4(uint32_t v){
  63         v = ((v & 0x0000FFFFU) << 16) | (v >> 16);
  64         return ((v & 0x00FF00FFU) << 8) | ((v & 0xFF00FF00U) >> 8);
  65 }
  66
  67 static inline uint64_t byte_swap_8(uint64_t v){
  68         v = ((v & 0x00000000FFFFFFFFLLU) << 32) | (v >> 32);
  69         v = ((v & 0x0000FFFF0000FFFFLLU) << 16) | ((v & 0xFFFF0000FFFF0000LLU) >> 16);
  70         return ((v & 0x00FF00FF00FF00FFLLU) << 8) | ((v & 0xFF00FF00FF00FF00LLU) >> 8);
  71 }
  72
  73 static inline int is_big_endian(){
  74         int x = 0x01;
  75         char *c = (char*)&x;
  76         return (c[0] != 0x01);
  77 }
  78
  79 #ifndef _RZ_READONLY
  80 static void add_zindex(RAZF *rz, int64_t in, int64_t out){
  81         if(rz->index->size == rz->index->cap){
  82                 rz->index->cap = rz->index->cap * 1.5 + 2;
  83                 rz->index->cell_offsets = realloc(rz->index->cell_offsets, sizeof(int) * rz->index->cap);
  84                 rz->index->bin_offsets  = realloc(rz->index->bin_offsets, sizeof(int64_t) * (rz->index->cap/RZ_BIN_SIZE + 1));
  85         }
  86         if(rz->index->size % RZ_BIN_SIZE == 0) rz->index->bin_offsets[rz->index->size / RZ_BIN_SIZE] = out;
  87         rz->index->cell_offsets[rz->index->size] = out - rz->index->bin_offsets[rz->index->size / RZ_BIN_SIZE];
  88         rz->index->size ++;
  89 }
  90
  91 static void save_zindex(RAZF *rz, int fd){
  92         int32_t i, v32;
  93         int is_be;
  94         is_be = is_big_endian();
  95         if(is_be) write(fd, &rz->index->size, sizeof(int));
  96         else {
  97                 v32 = byte_swap_4((uint32_t)rz->index->size);
  98                 write(fd, &v32, sizeof(uint32_t));
  99         }
 100         v32 = rz->index->size / RZ_BIN_SIZE + 1;
 101         if(!is_be){
 102                 for(i=0;i<v32;i++) rz->index->bin_offsets[i]  = byte_swap_8((uint64_t)rz->index->bin_offsets[i]);
 103                 for(i=0;i<rz->index->size;i++) rz->index->cell_offsets[i] = byte_swap_4((uint32_t)rz->index->cell_offsets[i]);
 104         }
 105         write(fd, rz->index->bin_offsets, sizeof(int64_t) * v32);
 106         write(fd, rz->index->cell_offsets, sizeof(int32_t) * rz->index->size);
 107 }
 108 #endif
 109
 110 static void load_zindex(RAZF *rz, int fd){
 111         int32_t i, v32;
 112         int is_be;
 113         if(!rz->load_index) return;
 114         if(rz->index == NULL) rz->index = malloc(sizeof(ZBlockIndex));
 115         is_be = is_big_endian();
 116         read(fd, &rz->index->size, sizeof(int));
 117         if(!is_be) rz->index->size = byte_swap_4((uint32_t)rz->index->size);
 118         rz->index->cap = rz->index->size;
 119         v32 = rz->index->size / RZ_BIN_SIZE + 1;
 120         rz->index->bin_offsets  = malloc(sizeof(int64_t) * v32);
 121         read(fd, rz->index->bin_offsets, sizeof(int64_t) * v32);
 122         rz->index->cell_offsets = malloc(sizeof(int) * rz->index->size);
 123         read(fd, rz->index->cell_offsets, sizeof(int) * rz->index->size);
 124         if(!is_be){
 125                 for(i=0;i<v32;i++) rz->index->bin_offsets[i] = byte_swap_8((uint64_t)rz->index->bin_offsets[i]);
 126                 for(i=0;i<rz->index->size;i++) rz->index->cell_offsets[i] = byte_swap_4((uint32_t)rz->index->cell_offsets[i]);
 127         }
 128 }
 129
 130 #ifdef _RZ_READONLY
 131 static RAZF* razf_open_w(int fd)
 132 {
 133         fprintf(stderr, "[razf_open_w] Writing is not available with zlib ver < 1.2.2.1\n");
 134         return 0;
 135 }
 136 #else
 137 static RAZF* razf_open_w(int fd){
 138         RAZF *rz;
 139         rz = calloc(1, sizeof(RAZF));
 140         rz->mode = 'w';
 141         rz->filedes = fd;
 142         rz->stream = calloc(sizeof(z_stream), 1);
 143         rz->inbuf  = malloc(RZ_BUFFER_SIZE);
 144         rz->outbuf = malloc(RZ_BUFFER_SIZE);
 145         rz->index = calloc(sizeof(ZBlockIndex), 1);
 146         deflateInit2(rz->stream, RZ_COMPRESS_LEVEL, Z_DEFLATED, WINDOW_BITS + 16, DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY);
 147         rz->stream->avail_out = RZ_BUFFER_SIZE;
 148         rz->stream->next_out  = rz->outbuf;
 149         rz->header = calloc(sizeof(gz_header), 1);
 150         rz->header->os    = 0x03; //Unix
 151         rz->header->text  = 0;
 152         rz->header->time  = 0;
 153         rz->header->extra = malloc(7);
 154         strncpy((char*)rz->header->extra, "RAZF", 4);
 155         rz->header->extra[4] = 1; // obsolete field
 156         // block size = RZ_BLOCK_SIZE, Big-Endian
 157         rz->header->extra[5] = RZ_BLOCK_SIZE >> 8;
 158         rz->header->extra[6] = RZ_BLOCK_SIZE & 0xFF;
 159         rz->header->extra_len = 7;
 160         rz->header->name = rz->header->comment  = 0;
 161         rz->header->hcrc = 0;
 162         deflateSetHeader(rz->stream, rz->header);
 163         rz->block_pos = rz->block_off = 0;
 164         return rz;
 165 }
 166
 167 static void _razf_write(RAZF* rz, const void *data, int size){
 168         int tout;
 169         rz->stream->avail_in = size;
 170         rz->stream->next_in  = (void*)data;
 171         while(1){
 172                 tout = rz->stream->avail_out;
 173                 deflate(rz->stream, Z_NO_FLUSH);
 174                 rz->out += tout - rz->stream->avail_out;
 175                 if(rz->stream->avail_out) break;
 176                 write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
 177                 rz->stream->avail_out = RZ_BUFFER_SIZE;
 178                 rz->stream->next_out  = rz->outbuf;
 179                 if(rz->stream->avail_in == 0) break;
 180         };
 181         rz->in += size - rz->stream->avail_in;
 182         rz->block_off += size - rz->stream->avail_in;
 183 }
 184
 185 static void razf_flush(RAZF *rz){
 186         uint32_t tout;
 187         if(rz->buf_len){
 188                 _razf_write(rz, rz->inbuf, rz->buf_len);
 189                 rz->buf_off = rz->buf_len = 0;
 190         }
 191         if(rz->stream->avail_out){
 192                 write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
 193                 rz->stream->avail_out = RZ_BUFFER_SIZE;
 194                 rz->stream->next_out  = rz->outbuf;
 195         }
 196         while(1){
 197                 tout = rz->stream->avail_out;
 198                 deflate(rz->stream, Z_FULL_FLUSH);
 199                 rz->out += tout - rz->stream->avail_out;
 200                 if(rz->stream->avail_out == 0){
 201                         write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
 202                         rz->stream->avail_out = RZ_BUFFER_SIZE;
 203                         rz->stream->next_out  = rz->outbuf;
 204                 } else break;
 205         }
 206         rz->block_pos = rz->out;
 207         rz->block_off = 0;
 208 }
 209
 210 static void razf_end_flush(RAZF *rz){
 211         uint32_t tout;
 212         if(rz->buf_len){
 213                 _razf_write(rz, rz->inbuf, rz->buf_len);
 214                 rz->buf_off = rz->buf_len = 0;
 215         }
 216         while(1){
 217                 tout = rz->stream->avail_out;
 218                 deflate(rz->stream, Z_FINISH);
 219                 rz->out += tout - rz->stream->avail_out;
 220                 if(rz->stream->avail_out < RZ_BUFFER_SIZE){
 221                         write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
 222                         rz->stream->avail_out = RZ_BUFFER_SIZE;
 223                         rz->stream->next_out  = rz->outbuf;
 224                 } else break;
 225         }
 226 }
 227
 228 static void _razf_buffered_write(RAZF *rz, const void *data, int size){
 229         int i, n;
 230         while(1){
 231                 if(rz->buf_len == RZ_BUFFER_SIZE){
 232                         _razf_write(rz, rz->inbuf, rz->buf_len);
 233                         rz->buf_len = 0;
 234                 }
 235                 if(size + rz->buf_len < RZ_BUFFER_SIZE){
 236                         for(i=0;i<size;i++) ((char*)rz->inbuf + rz->buf_len)[i] = ((char*)data)[i];
 237                         rz->buf_len += size;
 238                         return;
 239                 } else {
 240                         n = RZ_BUFFER_SIZE - rz->buf_len;
 241                         for(i=0;i<n;i++) ((char*)rz->inbuf + rz->buf_len)[i] = ((char*)data)[i];
 242                         size -= n;
 243                         data += n;
 244                         rz->buf_len += n;
 245                 }
 246         }
 247 }
 248
 249 int razf_write(RAZF* rz, const void *data, int size){
 250         int ori_size, n;
 251         int64_t next_block;
 252         ori_size = size;
 253         next_block = ((rz->in / RZ_BLOCK_SIZE) + 1) * RZ_BLOCK_SIZE;
 254         while(rz->in + rz->buf_len + size >= next_block){
 255                 n = next_block - rz->in - rz->buf_len;
 256                 _razf_buffered_write(rz, data, n);
 257                 data += n;
 258                 size -= n;
 259                 razf_flush(rz);
 260                 add_zindex(rz, rz->in, rz->out);
 261                 next_block = ((rz->in / RZ_BLOCK_SIZE) + 1) * RZ_BLOCK_SIZE;
 262         }
 263         _razf_buffered_write(rz, data, size);
 264         return ori_size;
 265 }
 266 #endif
 267
 268 /* gzip flag byte */
 269 #define ASCII_FLAG   0x01 /* bit 0 set: file probably ascii text */
 270 #define HEAD_CRC     0x02 /* bit 1 set: header CRC present */
 271 #define EXTRA_FIELD  0x04 /* bit 2 set: extra field present */
 272 #define ORIG_NAME    0x08 /* bit 3 set: original file name present */
 273 #define COMMENT      0x10 /* bit 4 set: file comment present */
 274 #define RESERVED     0xE0 /* bits 5..7: reserved */
 275
 276 static int _read_gz_header(unsigned char *data, int size, int *extra_off, int *extra_len){
 277         int method, flags, n, len;
 278         if(size < 2) return 0;
 279         if(data[0] != 0x1f || data[1] != 0x8b) return 0;
 280         if(size < 4) return 0;
 281         method = data[2];
 282         flags  = data[3];
 283         if(method != Z_DEFLATED || (flags & RESERVED)) return 0;
 284         n = 4 + 6; // Skip 6 bytes
 285         *extra_off = n + 2;
 286         *extra_len = 0;
 287         if(flags & EXTRA_FIELD){
 288                 if(size < n + 2) return 0;
 289                 len = ((int)data[n + 1] << 8) | data[n];
 290                 n += 2;
 291                 *extra_off = n;
 292                 while(len){
 293                         if(n >= size) return 0;
 294                         n ++;
 295                         len --;
 296                 }
 297                 *extra_len = n - (*extra_off);
 298         }
 299         if(flags & ORIG_NAME) while(n < size && data[n++]);
 300         if(flags & COMMENT) while(n < size && data[n++]);
 301         if(flags & HEAD_CRC){
 302                 if(n + 2 > size) return 0;
 303                 n += 2;
 304         }
 305         return n;
 306 }
 307
 308 static RAZF* razf_open_r(int fd, int _load_index){
 309         RAZF *rz;
 310         int ext_off, ext_len;
 311         int n, is_be, ret;
 312         int64_t end;
 313         unsigned char c[] = "RAZF";
 314         rz = calloc(1, sizeof(RAZF));
 315         rz->mode = 'r';
 316         rz->filedes = fd;
 317         rz->stream = calloc(sizeof(z_stream), 1);
 318         rz->inbuf  = malloc(RZ_BUFFER_SIZE);
 319         rz->outbuf = malloc(RZ_BUFFER_SIZE);
 320         rz->end = rz->src_end = 0x7FFFFFFFFFFFFFFFLL;
 321         n = read(rz->filedes, rz->inbuf, RZ_BUFFER_SIZE);
 322         ret = _read_gz_header(rz->inbuf, n, &ext_off, &ext_len);
 323         if(ret == 0){
 324                 PLAIN_FILE:
 325                 rz->in = n;
 326                 rz->file_type = FILE_TYPE_PLAIN;
 327                 memcpy(rz->outbuf, rz->inbuf, n);
 328                 rz->buf_len = n;
 329                 free(rz->stream);
 330                 rz->stream = NULL;
 331                 return rz;
 332         }
 333         rz->header_size = ret;
 334         ret = inflateInit2(rz->stream, -WINDOW_BITS);
 335         if(ret != Z_OK){ inflateEnd(rz->stream); goto PLAIN_FILE;}
 336         rz->stream->avail_in = n - rz->header_size;
 337         rz->stream->next_in  = rz->inbuf + rz->header_size;
 338         rz->stream->avail_out = RZ_BUFFER_SIZE;
 339         rz->stream->next_out  = rz->outbuf;
 340         rz->file_type = FILE_TYPE_GZ;
 341         rz->in = rz->header_size;
 342         rz->block_pos = rz->header_size;
 343         rz->next_block_pos = rz->header_size;
 344         rz->block_off = 0;
 345         if(ext_len < 7 || memcmp(rz->inbuf + ext_off, c, 4) != 0) return rz;
 346         if(((((unsigned char*)rz->inbuf)[ext_off + 5] << 8) | ((unsigned char*)rz->inbuf)[ext_off + 6]) != RZ_BLOCK_SIZE){
 347                 fprintf(stderr, " -- WARNING: RZ_BLOCK_SIZE is not %d, treat source as gz file.  in %s -- %s:%d --\n", RZ_BLOCK_SIZE, __FUNCTION__, __FILE__, __LINE__);
 348                 return rz;
 349         }
 350         rz->load_index = _load_index;
 351         rz->file_type = FILE_TYPE_RZ;
 352         if(lseek(fd, -16, SEEK_END) == -1){
 353                 UNSEEKABLE:
 354                 rz->seekable = 0;
 355                 rz->index = NULL;
 356                 rz->src_end = rz->end = 0x7FFFFFFFFFFFFFFFLL;
 357         } else {
 358                 is_be = is_big_endian();
 359                 rz->seekable = 1;
 360                 read(fd, &end, sizeof(int64_t));
 361                 if(!is_be) rz->src_end = (int64_t)byte_swap_8((uint64_t)end);
 362                 else rz->src_end = end;
 363                 read(fd, &end, sizeof(int64_t));
 364                 if(!is_be) rz->end = (int64_t)byte_swap_8((uint64_t)end);
 365                 else rz->end = end;
 366                 if(n > rz->end){
 367                         rz->stream->avail_in -= n - rz->end;
 368                         n = rz->end;
 369                 }
 370                 if(rz->end > rz->src_end){
 371                         lseek(fd, rz->in, SEEK_SET);
 372                         goto UNSEEKABLE;
 373                 }
 374                 if(lseek(fd, rz->end, SEEK_SET) != rz->end){
 375                         lseek(fd, rz->in, SEEK_SET);
 376                         goto UNSEEKABLE;
 377                 }
 378                 load_zindex(rz, fd);
 379                 lseek(fd, n, SEEK_SET);
 380         }
 381         return rz;
 382 }
 383
 384 RAZF* razf_dopen(int fd, const char *mode){
 385         if(strcasecmp(mode, "r") == 0) return razf_open_r(fd, 1);
 386         else if(strcasecmp(mode, "w") == 0) return razf_open_w(fd);
 387         else return NULL;
 388 }
 389
 390 RAZF* razf_dopen2(int fd, const char *mode)
 391 {
 392         if(strcasecmp(mode, "r") == 0) return razf_open_r(fd, 0);
 393         else if(strcasecmp(mode, "w") == 0) return razf_open_w(fd);
 394         else return NULL;
 395 }
 396
 397 static inline RAZF* _razf_open(const char *filename, const char *mode, int _load_index){
 398         int fd;
 399         RAZF *rz;
 400         if(strcasecmp(mode, "r") == 0){
 401                 fd = open(filename, O_RDONLY);
 402                 rz = razf_open_r(fd, _load_index);
 403         } else if(strcasecmp(mode, "w") == 0){
 404                 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC, 0644);
 405                 rz = razf_open_w(fd);
 406         } else return NULL;
 407         return rz;
 408 }
 409
 410 RAZF* razf_open(const char *filename, const char *mode){
 411         return _razf_open(filename, mode, 1);
 412 }
 413
 414 RAZF* razf_open2(const char *filename, const char *mode){
 415         return _razf_open(filename, mode, 0);
 416 }
 417
 418 int razf_get_data_size(RAZF *rz, int64_t *u_size, int64_t *c_size){
 419         int64_t n;
 420         if(rz->mode != 'r' && rz->mode != 'R') return 0;
 421         switch(rz->file_type){
 422                 case FILE_TYPE_PLAIN:
 423                         if(rz->end == 0x7fffffffffffffffLL){
 424                                 if((n = lseek(rz->filedes, 0, SEEK_CUR)) == -1) return 0;
 425                                 rz->end = lseek(rz->filedes, 0, SEEK_END);
 426                                 lseek(rz->filedes, n, SEEK_SET);
 427                         }
 428                         *u_size = *c_size = rz->end;
 429                         return 1;
 430                 case FILE_TYPE_GZ:
 431                         return 0;
 432                 case FILE_TYPE_RZ:
 433                         if(rz->src_end == rz->end) return 0;
 434                         *u_size = rz->src_end;
 435                         *c_size = rz->end;
 436                         return 1;
 437                 default:
 438                         return 0;
 439         }
 440 }
 441
 442 static int _razf_read(RAZF* rz, void *data, int size){
 443         int ret, tin;
 444         if(rz->z_eof || rz->z_err) return 0;
 445         if (rz->file_type == FILE_TYPE_PLAIN) {
 446                 ret = read(rz->filedes, data, size);
 447                 if (ret == 0) rz->z_eof = 1;
 448                 return ret;
 449         }
 450         rz->stream->avail_out = size;
 451         rz->stream->next_out  = data;
 452         while(rz->stream->avail_out){
 453                 if(rz->stream->avail_in == 0){
 454                         if(rz->in >= rz->end){ rz->z_eof = 1; break; }
 455                         if(rz->end - rz->in < RZ_BUFFER_SIZE){
 456                                 rz->stream->avail_in = read(rz->filedes, rz->inbuf, rz->end -rz->in);
 457                         } else {
 458                                 rz->stream->avail_in = read(rz->filedes, rz->inbuf, RZ_BUFFER_SIZE);
 459                         }
 460                         if(rz->stream->avail_in == 0){
 461                                 rz->z_eof = 1;
 462                                 break;
 463                         }
 464                         rz->stream->next_in = rz->inbuf;
 465                 }
 466                 tin = rz->stream->avail_in;
 467                 ret = inflate(rz->stream, Z_BLOCK);
 468                 rz->in += tin - rz->stream->avail_in;
 469                 if(ret == Z_NEED_DICT || ret == Z_MEM_ERROR || ret == Z_DATA_ERROR){
 470                         fprintf(stderr, "[_razf_read] inflate error: %d (at %s:%d)\n", ret, __FILE__, __LINE__);
 471                         rz->z_err = 1;
 472                         break;
 473                 }
 474                 if(ret == Z_STREAM_END){
 475                         rz->z_eof = 1;
 476                         break;
 477                 }
 478                 if ((rz->stream->data_type&128) && !(rz->stream->data_type&64)){
 479                         rz->buf_flush = 1;
 480                         rz->next_block_pos = rz->in;
 481                         break;
 482                 }
 483         }
 484         return size - rz->stream->avail_out;
 485 }
 486
 487 int razf_read(RAZF *rz, void *data, int size){
 488         int ori_size, i;
 489         ori_size = size;
 490         while(size > 0){
 491                 if(rz->buf_len){
 492                         if(size < rz->buf_len){
 493                                 for(i=0;i<size;i++) ((char*)data)[i] = ((char*)rz->outbuf + rz->buf_off)[i];
 494                                 rz->buf_off += size;
 495                                 rz->buf_len -= size;
 496                                 data += size;
 497                                 rz->block_off += size;
 498                                 size = 0;
 499                                 break;
 500                         } else {
 501                                 for(i=0;i<rz->buf_len;i++) ((char*)data)[i] = ((char*)rz->outbuf + rz->buf_off)[i];
 502                                 data += rz->buf_len;
 503                                 size -= rz->buf_len;
 504                                 rz->block_off += rz->buf_len;
 505                                 rz->buf_off = 0;
 506                                 rz->buf_len = 0;
 507                                 if(rz->buf_flush){
 508                                         rz->block_pos = rz->next_block_pos;
 509                                         rz->block_off = 0;
 510                                         rz->buf_flush = 0;
 511                                 }
 512                         }
 513                 } else if(rz->buf_flush){
 514                         rz->block_pos = rz->next_block_pos;
 515                         rz->block_off = 0;
 516                         rz->buf_flush = 0;
 517                 }
 518                 if(rz->buf_flush) continue;
 519                 rz->buf_len = _razf_read(rz, rz->outbuf, RZ_BUFFER_SIZE);
 520                 if(rz->z_eof && rz->buf_len == 0) break;
 521         }
 522         rz->out += ori_size - size;
 523         return ori_size - size;
 524 }
 525
 526 int razf_skip(RAZF* rz, int size){
 527         int ori_size;
 528         ori_size = size;
 529         while(size > 0){
 530                 if(rz->buf_len){
 531                         if(size < rz->buf_len){
 532                                 rz->buf_off += size;
 533                                 rz->buf_len -= size;
 534                                 rz->block_off += size;
 535                                 size = 0;
 536                                 break;
 537                         } else {
 538                                 size -= rz->buf_len;
 539                                 rz->buf_off = 0;
 540                                 rz->buf_len = 0;
 541                                 rz->block_off += rz->buf_len;
 542                                 if(rz->buf_flush){
 543                                         rz->block_pos = rz->next_block_pos;
 544                                         rz->block_off = 0;
 545                                         rz->buf_flush = 0;
 546                                 }
 547                         }
 548                 } else if(rz->buf_flush){
 549                         rz->block_pos = rz->next_block_pos;
 550                         rz->block_off = 0;
 551                         rz->buf_flush = 0;
 552                 }
 553                 if(rz->buf_flush) continue;
 554                 rz->buf_len = _razf_read(rz, rz->outbuf, RZ_BUFFER_SIZE);
 555                 if(rz->z_eof) break;
 556         }
 557         rz->out += ori_size - size;
 558         return ori_size - size;
 559 }
 560
 561 static void _razf_reset_read(RAZF *rz, int64_t in, int64_t out){
 562         lseek(rz->filedes, in, SEEK_SET);
 563         rz->in  = in;
 564         rz->out = out;
 565         rz->block_pos = in;
 566         rz->next_block_pos = in;
 567         rz->block_off = 0;
 568         rz->buf_flush = 0;
 569         rz->z_eof = rz->z_err = 0;
 570         inflateReset(rz->stream);
 571         rz->stream->avail_in = 0;
 572         rz->buf_off = rz->buf_len = 0;
 573 }
 574
 575 int64_t razf_jump(RAZF *rz, int64_t block_start, int block_offset){
 576         int64_t pos;
 577         rz->z_eof = 0;
 578         if(rz->file_type == FILE_TYPE_PLAIN){
 579                 rz->buf_off = rz->buf_len = 0;
 580                 pos = block_start + block_offset;
 581                 pos = lseek(rz->filedes, pos, SEEK_SET);
 582                 rz->out = rz->in = pos;
 583                 return pos;
 584         }
 585         if(block_start == rz->block_pos && block_offset >= rz->block_off) {
 586                 block_offset -= rz->block_off;
 587                 goto SKIP; // Needn't reset inflate
 588         }
 589         if(block_start  == 0) block_start = rz->header_size; // Automaticly revist wrong block_start
 590         _razf_reset_read(rz, block_start, 0);
 591         SKIP:
 592         if(block_offset) razf_skip(rz, block_offset);
 593         return rz->block_off;
 594 }
 595
 596 int64_t razf_seek(RAZF* rz, int64_t pos, int where){
 597         int64_t idx;
 598         int64_t seek_pos, new_out;
 599         rz->z_eof = 0;
 600         if (where == SEEK_CUR) pos += rz->out;
 601         else if (where == SEEK_END) pos += rz->src_end;
 602         if(rz->file_type == FILE_TYPE_PLAIN){
 603                 seek_pos = lseek(rz->filedes, pos, SEEK_SET);
 604                 rz->buf_off = rz->buf_len = 0;
 605                 rz->out = rz->in = seek_pos;
 606                 return seek_pos;
 607         } else if(rz->file_type == FILE_TYPE_GZ){
 608                 if(pos >= rz->out) goto SKIP;
 609                 return rz->out;
 610         }
 611         if(pos == rz->out) return pos;
 612         if(pos > rz->src_end) return rz->out;
 613         if(!rz->seekable || !rz->load_index){
 614                 if(pos >= rz->out) goto SKIP;
 615         }
 616         idx = pos / RZ_BLOCK_SIZE - 1;
 617         seek_pos = (idx < 0)? rz->header_size:(rz->index->cell_offsets[idx] + rz->index->bin_offsets[idx / RZ_BIN_SIZE]);
 618         new_out  = (idx + 1) * RZ_BLOCK_SIZE;
 619         if(pos > rz->out && new_out <= rz->out) goto SKIP;
 620         _razf_reset_read(rz, seek_pos, new_out);
 621         SKIP:
 622         razf_skip(rz, (int)(pos - rz->out));
 623         return rz->out;
 624 }
 625
 626 uint64_t razf_tell2(RAZF *rz)
 627 {
 628         /*
 629         if (rz->load_index) {
 630                 int64_t idx, seek_pos;
 631                 idx = rz->out / RZ_BLOCK_SIZE - 1;
 632                 seek_pos = (idx < 0)? rz->header_size:(rz->index->cell_offsets[idx] + rz->index->bin_offsets[idx / RZ_BIN_SIZE]);
 633                 if (seek_pos != rz->block_pos || rz->out%RZ_BLOCK_SIZE != rz->block_off)
 634                         fprintf(stderr, "[razf_tell2] inconsistent block offset: (%lld, %lld) != (%lld, %lld)\n",
 635                                         (long long)seek_pos, (long long)rz->out%RZ_BLOCK_SIZE, (long long)rz->block_pos, (long long) rz->block_off);
 636         }
 637         */
 638         return (uint64_t)rz->block_pos<<16 | (rz->block_off&0xffff);
 639 }
 640
 641 int64_t razf_seek2(RAZF *rz, uint64_t voffset, int where)
 642 {
 643         if (where != SEEK_SET) return -1;
 644         return razf_jump(rz, voffset>>16, voffset&0xffff);
 645 }
 646
 647 void razf_close(RAZF *rz){
 648         if(rz->mode == 'w'){
 649 #ifndef _RZ_READONLY
 650                 razf_end_flush(rz);
 651                 deflateEnd(rz->stream);
 652                 save_zindex(rz, rz->filedes);
 653                 if(is_big_endian()){
 654                         write(rz->filedes, &rz->in, sizeof(int64_t));
 655                         write(rz->filedes, &rz->out, sizeof(int64_t));
 656                 } else {
 657                         uint64_t v64 = byte_swap_8((uint64_t)rz->in);
 658                         write(rz->filedes, &v64, sizeof(int64_t));
 659                         v64 = byte_swap_8((uint64_t)rz->out);
 660                         write(rz->filedes, &v64, sizeof(int64_t));
 661                 }
 662 #endif
 663         } else if(rz->mode == 'r'){
 664                 if(rz->stream) inflateEnd(rz->stream);
 665         }
 666         if(rz->inbuf) free(rz->inbuf);
 667         if(rz->outbuf) free(rz->outbuf);
 668         if(rz->header){
 669                 free(rz->header->extra);
 670                 free(rz->header->name);
 671                 free(rz->header->comment);
 672                 free(rz->header);
 673         }
 674         if(rz->index){
 675                 free(rz->index->bin_offsets);
 676                 free(rz->index->cell_offsets);
 677                 free(rz->index);
 678         }
 679         free(rz->stream);
 680         close(rz->filedes);
 681         free(rz);
 682 }
 683
 684 #endif