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
  42 #if ZLIB_VERNUM < 0x1221
  43 struct _gz_header_s {
  44     int     text;
  45     uLong   time;
  46     int     xflags;
  47     int     os;
  48     Bytef   *extra;
  49     uInt    extra_len;
  50     uInt    extra_max;
  51     Bytef   *name;
  52     uInt    name_max;
  53     Bytef   *comment;
  54     uInt    comm_max;
  55     int     hcrc;
  56     int     done;
  57 };
  58 #warning "zlib < 1.2.2.1; RAZF writing is disabled."
  59 #endif
  60
  61 #define DEF_MEM_LEVEL 8
  62
  63 static inline uint32_t byte_swap_4(uint32_t v){
  64         v = ((v & 0x0000FFFFU) << 16) | (v >> 16);
  65         return ((v & 0x00FF00FFU) << 8) | ((v & 0xFF00FF00U) >> 8);
  66 }
  67
  68 static inline uint64_t byte_swap_8(uint64_t v){
  69         v = ((v & 0x00000000FFFFFFFFLLU) << 32) | (v >> 32);
  70         v = ((v & 0x0000FFFF0000FFFFLLU) << 16) | ((v & 0xFFFF0000FFFF0000LLU) >> 16);
  71         return ((v & 0x00FF00FF00FF00FFLLU) << 8) | ((v & 0xFF00FF00FF00FF00LLU) >> 8);
  72 }
  73
  74 static inline int is_big_endian(){
  75         int x = 0x01;
  76         char *c = (char*)&x;
  77         return (c[0] != 0x01);
  78 }
  79
  80 #ifndef _RZ_READONLY
  81 static void add_zindex(RAZF *rz, int64_t in, int64_t out){
  82         uint32_t *cores;
  83         int64_t *bores;
  84
  85         if(rz->index->size == rz->index->cap){
  86                 rz->index->cap = rz->index->cap * 1.5 + 2;
  87                 cores = realloc(rz->index->cell_offsets, sizeof(int) * rz->index->cap);
  88                 bores = realloc(rz->index->bin_offsets, sizeof(int64_t) * (rz->index->cap/RZ_BIN_SIZE + 1));
  89                 if ((cores == NULL) || (bores == NULL)) {
  90                         fprintf(stderr, "[%s] failure to allocate space for new zindex.\n", __func__);
  91                         abort();
  92                 }
  93                 rz->index->cell_offsets = cores;
  94                 rz->index->bin_offsets = bores;
  95         }
  96         if(rz->index->size % RZ_BIN_SIZE == 0) rz->index->bin_offsets[rz->index->size / RZ_BIN_SIZE] = out;
  97         rz->index->cell_offsets[rz->index->size] = out - rz->index->bin_offsets[rz->index->size / RZ_BIN_SIZE];
  98         rz->index->size ++;
  99 }
 100
 101 static void save_zindex(RAZF *rz, int fd){
 102         size_t count;
 103         int32_t i, v32;
 104         int is_be;
 105         is_be = is_big_endian();
 106         if(is_be) {
 107                 if (write(fd, &rz->index->size, sizeof(int)) < 0) {
 108                         fprintf(stderr, "[%s] failure to write zindex size.\n", __func__);
 109                         abort();
 110                 }
 111         }
 112         else {
 113                 v32 = byte_swap_4((uint32_t)rz->index->size);
 114                 if (write(fd, &v32, sizeof(uint32_t)) < 0) {
 115                         fprintf(stderr, "[%s] failure to write zindex size.\n", __func__);
 116                         abort();
 117                 }
 118         }
 119         v32 = rz->index->size / RZ_BIN_SIZE + 1;
 120         if(!is_be){
 121                 for(i=0;i<v32;i++) rz->index->bin_offsets[i]  = byte_swap_8((uint64_t)rz->index->bin_offsets[i]);
 122                 for(i=0;i<rz->index->size;i++) rz->index->cell_offsets[i] = byte_swap_4((uint32_t)rz->index->cell_offsets[i]);
 123         }
 124         count = sizeof(int64_t) * v32;
 125         if (write(fd, rz->index->bin_offsets, count) < 0) {
 126                 fprintf(stderr, "[%s] failure to write zindex bin_offsets.\n", __func__);
 127                 abort();
 128         }
 129
 130         count = sizeof(int32_t) * rz->index->size;
 131         if (write(fd, rz->index->cell_offsets, count) < 0) {
 132                 fprintf(stderr, "[%s] failure to write zindex cell_offsets.\n", __func__);
 133                 abort();
 134         }
 135 }
 136 #endif
 137
 138 #ifdef _USE_KNETFILE
 139 static void load_zindex(RAZF *rz, knetFile *fp){
 140 #else
 141 static void load_zindex(RAZF *rz, int fd){
 142 #endif
 143         int32_t i, v32;
 144         int is_be;
 145         size_t count;
 146         if(!rz->load_index) return;
 147         if(rz->index == NULL) {
 148                 if ((rz->index = malloc(sizeof(ZBlockIndex))) == NULL) {
 149                         fprintf(stderr, "[%s] failure to allocate index.\n", __func__);
 150                         abort();
 151                 }
 152         }
 153         is_be = is_big_endian();
 154 #ifdef _USE_KNETFILE
 155         if (knet_read(fp, &rz->index->size, sizeof(int)) < 0) {
 156 #else
 157         if (read(fd, &rz->index->size, sizeof(int)) < 0) {
 158 #endif
 159                 fprintf(stderr, "[%s] failure to read zindex size.\n", __func__);
 160                 abort();
 161         }
 162         if(!is_be) rz->index->size = byte_swap_4((uint32_t)rz->index->size);
 163         rz->index->cap = rz->index->size;
 164         v32 = rz->index->size / RZ_BIN_SIZE + 1;
 165         count = sizeof(int64_t) * v32;
 166         if ((rz->index->bin_offsets = malloc(count)) == NULL) {
 167                 fprintf(stderr, "[%s] failure to allocate bin_offsets array.\n", __func__);
 168                 abort();
 169         }
 170 #ifdef _USE_KNETFILE
 171         if (knet_read(fp, rz->index->bin_offsets, count) < 0) {
 172 #else
 173         if (read(fd, rz->index->bin_offsets, count) < 0) {
 174 #endif
 175                 fprintf(stderr, "[%s] failure to read bin_offsets.\n", __func__);
 176                 abort();
 177         }
 178         count = sizeof(int) * rz->index->size;
 179         if ((rz->index->cell_offsets = malloc(count)) == NULL) {
 180                 fprintf(stderr, "[%s] failure to allocate cell_offsets array.\n", __func__);
 181                 abort();
 182         }
 183 #ifdef _USE_KNETFILE
 184         if (knet_read(fp, rz->index->cell_offsets, count) < count) {
 185 #else
 186         if (read(fd, rz->index->cell_offsets, count) < count) {
 187 #endif
 188                 fprintf(stderr, "[%s] failure to read cell_offsets.\n", __func__);
 189                 abort();
 190         }
 191         if(!is_be){
 192                 for(i=0;i<v32;i++) rz->index->bin_offsets[i] = byte_swap_8((uint64_t)rz->index->bin_offsets[i]);
 193                 for(i=0;i<rz->index->size;i++) rz->index->cell_offsets[i] = byte_swap_4((uint32_t)rz->index->cell_offsets[i]);
 194         }
 195 }
 196
 197 #ifdef _RZ_READONLY
 198 static RAZF* razf_open_w(int fd)
 199 {
 200         fprintf(stderr, "[razf_open_w] Writing is not available with zlib ver < 1.2.2.1\n");
 201         return 0;
 202 }
 203 #else
 204 static RAZF* razf_open_w(int fd){
 205         RAZF *rz;
 206 #ifdef _WIN32
 207         setmode(fd, O_BINARY);
 208 #endif
 209         if ((rz = calloc(1, sizeof(RAZF))) == NULL) {
 210                 fprintf(stderr, "[%s] failure to allocate RAZF structure.\n", __func__);
 211                 return NULL;
 212         }
 213         rz->mode = 'w';
 214 #ifdef _USE_KNETFILE
 215     rz->x.fpw = fd;
 216 #else
 217         rz->filedes = fd;
 218 #endif
 219         if ((rz->stream = calloc(sizeof(z_stream), 1)) == NULL) {
 220                 fprintf(stderr, "[%s] failure to allocate stream buffer.\n", __func__);
 221                 free(rz);
 222                 return NULL;
 223         }
 224         if ((rz->inbuf  = malloc(RZ_BUFFER_SIZE)) == NULL) {
 225                 fprintf(stderr, "[%s] failure to allocate input buffer.\n", __func__);
 226                 free(rz->stream);
 227                 free(rz);
 228                 return NULL;
 229         }
 230         if ((rz->outbuf = malloc(RZ_BUFFER_SIZE)) == NULL) {
 231                 fprintf(stderr, "[%s] failure to allocate output buffer.\n", __func__);
 232                 free(rz->stream);
 233                 free(rz->inbuf);
 234                 free(rz);
 235                 return NULL;
 236         }
 237         if ((rz->index = calloc(sizeof(ZBlockIndex), 1)) == NULL) {
 238                 fprintf(stderr, "[%s] failure to allocate index.\n", __func__);
 239                 free(rz->stream);
 240                 free(rz->inbuf);
 241                 free(rz->outbuf);
 242                 free(rz);
 243                 return NULL;
 244         }
 245
 246         deflateInit2(rz->stream, RZ_COMPRESS_LEVEL, Z_DEFLATED, WINDOW_BITS + 16, DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY);
 247         rz->stream->avail_out = RZ_BUFFER_SIZE;
 248         rz->stream->next_out  = rz->outbuf;
 249         if ((rz->header = calloc(sizeof(gz_header), 1)) == NULL) {
 250                 fprintf(stderr, "[%s] failure to allocate header buffer.\n", __func__);
 251                 free(rz->stream);
 252                 free(rz->inbuf);
 253                 free(rz->outbuf);
 254                 free(rz->index);
 255                 free(rz);
 256                 return NULL;
 257         }
 258         rz->header->os    = 0x03; //Unix
 259         rz->header->text  = 0;
 260         rz->header->time  = 0;
 261         if ((rz->header->extra = malloc(7)) == NULL) {
 262                 fprintf(stderr, "[%s] failure to allocate header buffer.\n", __func__);
 263                 free(rz->stream);
 264                 free(rz->inbuf);
 265                 free(rz->outbuf);
 266                 free(rz->index);
 267                 free(rz->header);
 268                 free(rz);
 269                 return NULL;
 270         }
 271         strncpy((char*)rz->header->extra, "RAZF", 4);
 272         rz->header->extra[4] = 1; // obsolete field
 273         // block size = RZ_BLOCK_SIZE, Big-Endian
 274         rz->header->extra[5] = RZ_BLOCK_SIZE >> 8;
 275         rz->header->extra[6] = RZ_BLOCK_SIZE & 0xFF;
 276         rz->header->extra_len = 7;
 277         rz->header->name = rz->header->comment  = 0;
 278         rz->header->hcrc = 0;
 279         deflateSetHeader(rz->stream, rz->header);
 280         rz->block_pos = rz->block_off = 0;
 281         return rz;
 282 }
 283
 284 static void _razf_write(RAZF* rz, const void *data, int size){
 285         int tout;
 286         size_t count;
 287         rz->stream->avail_in = size;
 288         rz->stream->next_in  = (void*)data;
 289         while(1){
 290                 tout = rz->stream->avail_out;
 291                 deflate(rz->stream, Z_NO_FLUSH);
 292                 rz->out += tout - rz->stream->avail_out;
 293                 if(rz->stream->avail_out) break;
 294                 count = RZ_BUFFER_SIZE - rz->stream->avail_out;
 295 #ifdef _USE_KNETFILE
 296                 if (write(rz->x.fpw, rz->outbuf, count) < 0) {
 297 #else
 298                 if (write(rz->filedes, rz->outbuf, count) < 0) {
 299 #endif
 300                         fprintf(stderr, "[%s] failed to write output buffer.\n", __func__);
 301                         abort();
 302                 }
 303                 rz->stream->avail_out = RZ_BUFFER_SIZE;
 304                 rz->stream->next_out  = rz->outbuf;
 305                 if(rz->stream->avail_in == 0) break;
 306         };
 307         rz->in += size - rz->stream->avail_in;
 308         rz->block_off += size - rz->stream->avail_in;
 309 }
 310
 311 static void razf_flush(RAZF *rz){
 312         size_t count;
 313         uint32_t tout;
 314         if(rz->buf_len){
 315                 _razf_write(rz, rz->inbuf, rz->buf_len);
 316                 rz->buf_off = rz->buf_len = 0;
 317         }
 318         if(rz->stream->avail_out){
 319                 count = RZ_BUFFER_SIZE - rz->stream->avail_out;
 320 #ifdef _USE_KNETFILE
 321                 if (write(rz->x.fpw, rz->outbuf, count) < 0) {
 322 #else
 323                 if (write(rz->filedes, rz->outbuf, count) < 0) {
 324 #endif
 325                         fprintf(stderr, "[%s] failed to flush output buffer.\n", __func__);
 326                         abort();
 327                 }
 328                 rz->stream->avail_out = RZ_BUFFER_SIZE;
 329                 rz->stream->next_out  = rz->outbuf;
 330         }
 331         while(1){
 332                 tout = rz->stream->avail_out;
 333                 deflate(rz->stream, Z_FULL_FLUSH);
 334                 rz->out += tout - rz->stream->avail_out;
 335                 if(rz->stream->avail_out == 0){
 336                         count = RZ_BUFFER_SIZE - rz->stream->avail_out;
 337 #ifdef _USE_KNETFILE
 338                         if (write(rz->x.fpw, rz->outbuf, count) < 0) {
 339 #else
 340                         if (write(rz->filedes, rz->outbuf, count) < 0) {
 341 #endif
 342                                 fprintf(stderr, "[%s] failed to flush output buffer.\n", __func__);
 343                                 abort();
 344                         }
 345                         rz->stream->avail_out = RZ_BUFFER_SIZE;
 346                         rz->stream->next_out  = rz->outbuf;
 347                 } else break;
 348         }
 349         rz->block_pos = rz->out;
 350         rz->block_off = 0;
 351 }
 352
 353 static void razf_end_flush(RAZF *rz){
 354         size_t count;
 355         uint32_t tout;
 356         if(rz->buf_len){
 357                 _razf_write(rz, rz->inbuf, rz->buf_len);
 358                 rz->buf_off = rz->buf_len = 0;
 359         }
 360         while(1){
 361                 tout = rz->stream->avail_out;
 362                 deflate(rz->stream, Z_FINISH);
 363                 rz->out += tout - rz->stream->avail_out;
 364                 if(rz->stream->avail_out < RZ_BUFFER_SIZE){
 365                         count = RZ_BUFFER_SIZE - rz->stream->avail_out;
 366 #ifdef _USE_KNETFILE
 367                         if (write(rz->x.fpw, rz->outbuf, count) < 0) {
 368 #else
 369                         if (write(rz->filedes, rz->outbuf, count) < 0) {
 370 #endif
 371                                 fprintf(stderr, "[%s] failed to flush output buffer.\n", __func__);
 372                                 abort();
 373                         }
 374                         rz->stream->avail_out = RZ_BUFFER_SIZE;
 375                         rz->stream->next_out  = rz->outbuf;
 376                 } else break;
 377         }
 378 }
 379
 380 static void _razf_buffered_write(RAZF *rz, const void *data, int size){
 381         int i, n;
 382         while(1){
 383                 if(rz->buf_len == RZ_BUFFER_SIZE){
 384                         _razf_write(rz, rz->inbuf, rz->buf_len);
 385                         rz->buf_len = 0;
 386                 }
 387                 if(size + rz->buf_len < RZ_BUFFER_SIZE){
 388                         for(i=0;i<size;i++) ((char*)rz->inbuf + rz->buf_len)[i] = ((char*)data)[i];
 389                         rz->buf_len += size;
 390                         return;
 391                 } else {
 392                         n = RZ_BUFFER_SIZE - rz->buf_len;
 393                         for(i=0;i<n;i++) ((char*)rz->inbuf + rz->buf_len)[i] = ((char*)data)[i];
 394                         size -= n;
 395                         data += n;
 396                         rz->buf_len += n;
 397                 }
 398         }
 399 }
 400
 401 int razf_write(RAZF* rz, const void *data, int size){
 402         int ori_size, n;
 403         int64_t next_block;
 404         ori_size = size;
 405         next_block = ((rz->in / RZ_BLOCK_SIZE) + 1) * RZ_BLOCK_SIZE;
 406         while(rz->in + rz->buf_len + size >= next_block){
 407                 n = next_block - rz->in - rz->buf_len;
 408                 _razf_buffered_write(rz, data, n);
 409                 data += n;
 410                 size -= n;
 411                 razf_flush(rz);
 412                 add_zindex(rz, rz->in, rz->out);
 413                 next_block = ((rz->in / RZ_BLOCK_SIZE) + 1) * RZ_BLOCK_SIZE;
 414         }
 415         _razf_buffered_write(rz, data, size);
 416         return ori_size;
 417 }
 418 #endif
 419
 420 /* gzip flag byte */
 421 #define ASCII_FLAG   0x01 /* bit 0 set: file probably ascii text */
 422 #define HEAD_CRC     0x02 /* bit 1 set: header CRC present */
 423 #define EXTRA_FIELD  0x04 /* bit 2 set: extra field present */
 424 #define ORIG_NAME    0x08 /* bit 3 set: original file name present */
 425 #define COMMENT      0x10 /* bit 4 set: file comment present */
 426 #define RESERVED     0xE0 /* bits 5..7: reserved */
 427
 428 static int _read_gz_header(unsigned char *data, int size, int *extra_off, int *extra_len){
 429         int method, flags, n, len;
 430         if(size < 2) return 0;
 431         if(data[0] != 0x1f || data[1] != 0x8b) return 0;
 432         if(size < 4) return 0;
 433         method = data[2];
 434         flags  = data[3];
 435         if(method != Z_DEFLATED || (flags & RESERVED)) return 0;
 436         n = 4 + 6; // Skip 6 bytes
 437         *extra_off = n + 2;
 438         *extra_len = 0;
 439         if(flags & EXTRA_FIELD){
 440                 if(size < n + 2) return 0;
 441                 len = ((int)data[n + 1] << 8) | data[n];
 442                 n += 2;
 443                 *extra_off = n;
 444                 while(len){
 445                         if(n >= size) return 0;
 446                         n ++;
 447                         len --;
 448                 }
 449                 *extra_len = n - (*extra_off);
 450         }
 451         if(flags & ORIG_NAME) while(n < size && data[n++]);
 452         if(flags & COMMENT) while(n < size && data[n++]);
 453         if(flags & HEAD_CRC){
 454                 if(n + 2 > size) return 0;
 455                 n += 2;
 456         }
 457         return n;
 458 }
 459
 460 #ifdef _USE_KNETFILE
 461 static RAZF* razf_open_r(knetFile *fp, int _load_index){
 462 #else
 463 static RAZF* razf_open_r(int fd, int _load_index){
 464 #endif
 465         RAZF *rz;
 466         int ext_off, ext_len;
 467         int n, is_be, ret;
 468         int64_t end;
 469         unsigned char c[] = "RAZF";
 470         if ((rz = calloc(1, sizeof(RAZF))) == NULL) {
 471                 fprintf(stderr, "[%s] failure to allocate RAZF structure.\n", __func__);
 472                 return NULL;
 473         }
 474         rz->mode = 'r';
 475 #ifdef _USE_KNETFILE
 476     rz->x.fpr = fp;
 477 #else
 478 #ifdef _WIN32
 479         setmode(fd, O_BINARY);
 480 #endif
 481         rz->filedes = fd;
 482 #endif
 483         if ((rz->stream = calloc(sizeof(z_stream), 1)) == NULL) {
 484                 fprintf(stderr, "[%s] failure to allocate z_stream.\n", __func__);
 485                 free(rz);
 486                 return NULL;
 487         }
 488         if ((rz->inbuf = malloc(RZ_BUFFER_SIZE)) == NULL) {
 489                 fprintf(stderr, "[%s] failure to allocate input buffer.\n", __func__);
 490                 free(rz->stream);
 491                 free(rz);
 492                 return NULL;
 493         }
 494         if ((rz->outbuf = malloc(RZ_BUFFER_SIZE)) == NULL) {
 495                 fprintf(stderr, "[%s] failure to allocate output buffer.\n", __func__);
 496                 free(rz->inbuf);
 497                 free(rz->stream);
 498                 free(rz);
 499                 return NULL;
 500         }
 501         rz->end = rz->src_end = 0x7FFFFFFFFFFFFFFFLL;
 502 #ifdef _USE_KNETFILE
 503     n = knet_read(rz->x.fpr, rz->inbuf, RZ_BUFFER_SIZE);
 504 #else
 505         n = read(rz->filedes, rz->inbuf, RZ_BUFFER_SIZE);
 506 #endif
 507         ret = _read_gz_header(rz->inbuf, n, &ext_off, &ext_len);
 508         if(ret == 0){
 509                 PLAIN_FILE:
 510                 rz->in = n;
 511                 rz->file_type = FILE_TYPE_PLAIN;
 512                 memcpy(rz->outbuf, rz->inbuf, n);
 513                 rz->buf_len = n;
 514                 free(rz->stream);
 515                 rz->stream = NULL;
 516                 return rz;
 517         }
 518         rz->header_size = ret;
 519         ret = inflateInit2(rz->stream, -WINDOW_BITS);
 520         if(ret != Z_OK){ inflateEnd(rz->stream); goto PLAIN_FILE;}
 521         rz->stream->avail_in = n - rz->header_size;
 522         rz->stream->next_in  = rz->inbuf + rz->header_size;
 523         rz->stream->avail_out = RZ_BUFFER_SIZE;
 524         rz->stream->next_out  = rz->outbuf;
 525         rz->file_type = FILE_TYPE_GZ;
 526         rz->in = rz->header_size;
 527         rz->block_pos = rz->header_size;
 528         rz->next_block_pos = rz->header_size;
 529         rz->block_off = 0;
 530         if(ext_len < 7 || memcmp(rz->inbuf + ext_off, c, 4) != 0) return rz;
 531         if(((((unsigned char*)rz->inbuf)[ext_off + 5] << 8) | ((unsigned char*)rz->inbuf)[ext_off + 6]) != RZ_BLOCK_SIZE){
 532                 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__);
 533                 return rz;
 534         }
 535         rz->load_index = _load_index;
 536         rz->file_type = FILE_TYPE_RZ;
 537 #ifdef _USE_KNETFILE
 538         if(knet_seek(fp, -16, SEEK_END) == -1){
 539 #else
 540         if(lseek(fd, -16, SEEK_END) == -1){
 541 #endif
 542                 UNSEEKABLE:
 543                 rz->seekable = 0;
 544                 rz->index = NULL;
 545                 rz->src_end = rz->end = 0x7FFFFFFFFFFFFFFFLL;
 546         } else {
 547                 is_be = is_big_endian();
 548                 rz->seekable = 1;
 549 #ifdef _USE_KNETFILE
 550         knet_read(fp, &end, sizeof(int64_t));
 551 #else
 552                 read(fd, &end, sizeof(int64_t));
 553 #endif
 554                 if(!is_be) rz->src_end = (int64_t)byte_swap_8((uint64_t)end);
 555                 else rz->src_end = end;
 556
 557 #ifdef _USE_KNETFILE
 558                 knet_read(fp, &end, sizeof(int64_t));
 559 #else
 560                 read(fd, &end, sizeof(int64_t));
 561 #endif
 562                 if(!is_be) rz->end = (int64_t)byte_swap_8((uint64_t)end);
 563                 else rz->end = end;
 564                 if(n > rz->end){
 565                         rz->stream->avail_in -= n - rz->end;
 566                         n = rz->end;
 567                 }
 568                 if(rz->end > rz->src_end){
 569 #ifdef _USE_KNETFILE
 570             knet_seek(fp, rz->in, SEEK_SET);
 571 #else
 572                         lseek(fd, rz->in, SEEK_SET);
 573 #endif
 574                         goto UNSEEKABLE;
 575                 }
 576 #ifdef _USE_KNETFILE
 577         knet_seek(fp, rz->end, SEEK_SET);
 578                 if(knet_tell(fp) != rz->end){
 579                         knet_seek(fp, rz->in, SEEK_SET);
 580 #else
 581                 if(lseek(fd, rz->end, SEEK_SET) != rz->end){
 582                         lseek(fd, rz->in, SEEK_SET);
 583 #endif
 584                         goto UNSEEKABLE;
 585                 }
 586 #ifdef _USE_KNETFILE
 587                 load_zindex(rz, fp);
 588                 knet_seek(fp, n, SEEK_SET);
 589 #else
 590                 load_zindex(rz, fd);
 591                 lseek(fd, n, SEEK_SET);
 592 #endif
 593         }
 594         return rz;
 595 }
 596
 597 #ifdef _USE_KNETFILE
 598 RAZF* razf_dopen(int fd, const char *mode){
 599     if (strstr(mode, "r")) fprintf(stderr,"[razf_dopen] implement me\n");
 600     else if(strstr(mode, "w")) return razf_open_w(fd);
 601         return NULL;
 602 }
 603
 604 RAZF* razf_dopen2(int fd, const char *mode)
 605 {
 606     fprintf(stderr,"[razf_dopen2] implement me\n");
 607     return NULL;
 608 }
 609 #else
 610 RAZF* razf_dopen(int fd, const char *mode){
 611         if(strstr(mode, "r")) return razf_open_r(fd, 1);
 612         else if(strstr(mode, "w")) return razf_open_w(fd);
 613         else return NULL;
 614 }
 615
 616 RAZF* razf_dopen2(int fd, const char *mode)
 617 {
 618         if(strstr(mode, "r")) return razf_open_r(fd, 0);
 619         else if(strstr(mode, "w")) return razf_open_w(fd);
 620         else return NULL;
 621 }
 622 #endif
 623
 624 static inline RAZF* _razf_open(const char *filename, const char *mode, int _load_index){
 625         int fd;
 626         RAZF *rz;
 627         if(strstr(mode, "r")){
 628 #ifdef _USE_KNETFILE
 629         knetFile *fd = knet_open(filename, "r");
 630         if (fd == 0) {
 631             fprintf(stderr, "[_razf_open] fail to open %s\n", filename);
 632             return NULL;
 633         }
 634 #else
 635 #ifdef _WIN32
 636                 fd = open(filename, O_RDONLY | O_BINARY);
 637 #else
 638                 fd = open(filename, O_RDONLY);
 639 #endif
 640 #endif
 641                 if(fd < 0) return NULL;
 642                 rz = razf_open_r(fd, _load_index);
 643         } else if(strstr(mode, "w")){
 644 #ifdef _WIN32
 645                 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0666);
 646 #else
 647                 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC, 0666);
 648 #endif
 649                 if(fd < 0) return NULL;
 650                 rz = razf_open_w(fd);
 651         } else return NULL;
 652         return rz;
 653 }
 654
 655 RAZF* razf_open(const char *filename, const char *mode){
 656         return _razf_open(filename, mode, 1);
 657 }
 658
 659 RAZF* razf_open2(const char *filename, const char *mode){
 660         return _razf_open(filename, mode, 0);
 661 }
 662
 663 int razf_get_data_size(RAZF *rz, int64_t *u_size, int64_t *c_size){
 664         int64_t n;
 665         if(rz->mode != 'r' && rz->mode != 'R') return 0;
 666         switch(rz->file_type){
 667                 case FILE_TYPE_PLAIN:
 668                         if(rz->end == 0x7fffffffffffffffLL){
 669 #ifdef _USE_KNETFILE
 670                                 if(knet_seek(rz->x.fpr, 0, SEEK_CUR) == -1) return 0;
 671                 n = knet_tell(rz->x.fpr);
 672                                 knet_seek(rz->x.fpr, 0, SEEK_END);
 673                 rz->end = knet_tell(rz->x.fpr);
 674                                 knet_seek(rz->x.fpr, n, SEEK_SET);
 675 #else
 676                                 if((n = lseek(rz->filedes, 0, SEEK_CUR)) == -1) return 0;
 677                                 rz->end = lseek(rz->filedes, 0, SEEK_END);
 678                                 lseek(rz->filedes, n, SEEK_SET);
 679 #endif
 680                         }
 681                         *u_size = *c_size = rz->end;
 682                         return 1;
 683                 case FILE_TYPE_GZ:
 684                         return 0;
 685                 case FILE_TYPE_RZ:
 686                         if(rz->src_end == rz->end) return 0;
 687                         *u_size = rz->src_end;
 688                         *c_size = rz->end;
 689                         return 1;
 690                 default:
 691                         return 0;
 692         }
 693 }
 694
 695 static int _razf_read(RAZF* rz, void *data, int size){
 696         int ret, tin;
 697         if(rz->z_eof || rz->z_err) return 0;
 698         if (rz->file_type == FILE_TYPE_PLAIN) {
 699 #ifdef _USE_KNETFILE
 700                 ret = knet_read(rz->x.fpr, data, size);
 701 #else
 702                 ret = read(rz->filedes, data, size);
 703 #endif
 704                 if (ret == 0) rz->z_eof = 1;
 705                 return ret;
 706         }
 707         rz->stream->avail_out = size;
 708         rz->stream->next_out  = data;
 709         while(rz->stream->avail_out){
 710                 if(rz->stream->avail_in == 0){
 711                         if(rz->in >= rz->end){ rz->z_eof = 1; break; }
 712                         if(rz->end - rz->in < RZ_BUFFER_SIZE){
 713 #ifdef _USE_KNETFILE
 714                                 rz->stream->avail_in = knet_read(rz->x.fpr, rz->inbuf, rz->end -rz->in);
 715 #else
 716                                 rz->stream->avail_in = read(rz->filedes, rz->inbuf, rz->end -rz->in);
 717 #endif
 718                         } else {
 719 #ifdef _USE_KNETFILE
 720                                 rz->stream->avail_in = knet_read(rz->x.fpr, rz->inbuf, RZ_BUFFER_SIZE);
 721 #else
 722                                 rz->stream->avail_in = read(rz->filedes, rz->inbuf, RZ_BUFFER_SIZE);
 723 #endif
 724                         }
 725                         if(rz->stream->avail_in == 0){
 726                                 rz->z_eof = 1;
 727                                 break;
 728                         }
 729                         rz->stream->next_in = rz->inbuf;
 730                 }
 731                 tin = rz->stream->avail_in;
 732                 ret = inflate(rz->stream, Z_BLOCK);
 733                 rz->in += tin - rz->stream->avail_in;
 734                 if(ret == Z_NEED_DICT || ret == Z_MEM_ERROR || ret == Z_DATA_ERROR){
 735                         fprintf(stderr, "[_razf_read] inflate error: %d %s (at %s:%d)\n", ret, rz->stream->msg ? rz->stream->msg : "", __FILE__, __LINE__);
 736                         rz->z_err = 1;
 737                         break;
 738                 }
 739                 if(ret == Z_STREAM_END){
 740                         rz->z_eof = 1;
 741                         break;
 742                 }
 743                 if ((rz->stream->data_type&128) && !(rz->stream->data_type&64)){
 744                         rz->buf_flush = 1;
 745                         rz->next_block_pos = rz->in;
 746                         break;
 747                 }
 748         }
 749         return size - rz->stream->avail_out;
 750 }
 751
 752 int razf_read(RAZF *rz, void *data, int size){
 753         int ori_size, i;
 754         ori_size = size;
 755         while(size > 0){
 756                 if(rz->buf_len){
 757                         if(size < rz->buf_len){
 758                                 for(i=0;i<size;i++) ((char*)data)[i] = ((char*)rz->outbuf + rz->buf_off)[i];
 759                                 rz->buf_off += size;
 760                                 rz->buf_len -= size;
 761                                 data += size;
 762                                 rz->block_off += size;
 763                                 size = 0;
 764                                 break;
 765                         } else {
 766                                 for(i=0;i<rz->buf_len;i++) ((char*)data)[i] = ((char*)rz->outbuf + rz->buf_off)[i];
 767                                 data += rz->buf_len;
 768                                 size -= rz->buf_len;
 769                                 rz->block_off += rz->buf_len;
 770                                 rz->buf_off = 0;
 771                                 rz->buf_len = 0;
 772                                 if(rz->buf_flush){
 773                                         rz->block_pos = rz->next_block_pos;
 774                                         rz->block_off = 0;
 775                                         rz->buf_flush = 0;
 776                                 }
 777                         }
 778                 } else if(rz->buf_flush){
 779                         rz->block_pos = rz->next_block_pos;
 780                         rz->block_off = 0;
 781                         rz->buf_flush = 0;
 782                 }
 783                 if(rz->buf_flush) continue;
 784                 rz->buf_len = _razf_read(rz, rz->outbuf, RZ_BUFFER_SIZE);
 785                 if(rz->z_eof && rz->buf_len == 0) break;
 786         }
 787         rz->out += ori_size - size;
 788         return ori_size - size;
 789 }
 790
 791 int razf_skip(RAZF* rz, int size){
 792         int ori_size;
 793         ori_size = size;
 794         while(size > 0){
 795                 if(rz->buf_len){
 796                         if(size < rz->buf_len){
 797                                 rz->buf_off += size;
 798                                 rz->buf_len -= size;
 799                                 rz->block_off += size;
 800                                 size = 0;
 801                                 break;
 802                         } else {
 803                                 size -= rz->buf_len;
 804                                 rz->buf_off = 0;
 805                                 rz->buf_len = 0;
 806                                 rz->block_off += rz->buf_len;
 807                                 if(rz->buf_flush){
 808                                         rz->block_pos = rz->next_block_pos;
 809                                         rz->block_off = 0;
 810                                         rz->buf_flush = 0;
 811                                 }
 812                         }
 813                 } else if(rz->buf_flush){
 814                         rz->block_pos = rz->next_block_pos;
 815                         rz->block_off = 0;
 816                         rz->buf_flush = 0;
 817                 }
 818                 if(rz->buf_flush) continue;
 819                 rz->buf_len = _razf_read(rz, rz->outbuf, RZ_BUFFER_SIZE);
 820                 if(rz->z_eof || rz->z_err) break;
 821         }
 822         rz->out += ori_size - size;
 823         return ori_size - size;
 824 }
 825
 826 static void _razf_reset_read(RAZF *rz, int64_t in, int64_t out){
 827 #ifdef _USE_KNETFILE
 828         knet_seek(rz->x.fpr, in, SEEK_SET);
 829 #else
 830         lseek(rz->filedes, in, SEEK_SET);
 831 #endif
 832         rz->in  = in;
 833         rz->out = out;
 834         rz->block_pos = in;
 835         rz->next_block_pos = in;
 836         rz->block_off = 0;
 837         rz->buf_flush = 0;
 838         rz->z_eof = rz->z_err = 0;
 839         inflateReset(rz->stream);
 840         rz->stream->avail_in = 0;
 841         rz->buf_off = rz->buf_len = 0;
 842 }
 843
 844 int64_t razf_jump(RAZF *rz, int64_t block_start, int block_offset){
 845         int64_t pos;
 846         rz->z_eof = 0;
 847         if(rz->file_type == FILE_TYPE_PLAIN){
 848                 rz->buf_off = rz->buf_len = 0;
 849                 pos = block_start + block_offset;
 850 #ifdef _USE_KNETFILE
 851                 knet_seek(rz->x.fpr, pos, SEEK_SET);
 852         pos = knet_tell(rz->x.fpr);
 853 #else
 854                 pos = lseek(rz->filedes, pos, SEEK_SET);
 855 #endif
 856                 rz->out = rz->in = pos;
 857                 return pos;
 858         }
 859         if(block_start == rz->block_pos && block_offset >= rz->block_off) {
 860                 block_offset -= rz->block_off;
 861                 goto SKIP; // Needn't reset inflate
 862         }
 863         if(block_start  == 0) block_start = rz->header_size; // Automaticly revist wrong block_start
 864         _razf_reset_read(rz, block_start, 0);
 865         SKIP:
 866         if(block_offset) razf_skip(rz, block_offset);
 867         return rz->block_off;
 868 }
 869
 870 int64_t razf_seek(RAZF* rz, int64_t pos, int where){
 871         int64_t idx;
 872         int64_t seek_pos, new_out;
 873         rz->z_eof = 0;
 874         if (where == SEEK_CUR) pos += rz->out;
 875         else if (where == SEEK_END) pos += rz->src_end;
 876         if(rz->file_type == FILE_TYPE_PLAIN){
 877 #ifdef _USE_KNETFILE
 878                 knet_seek(rz->x.fpr, pos, SEEK_SET);
 879         seek_pos = knet_tell(rz->x.fpr);
 880 #else
 881                 seek_pos = lseek(rz->filedes, pos, SEEK_SET);
 882 #endif
 883                 rz->buf_off = rz->buf_len = 0;
 884                 rz->out = rz->in = seek_pos;
 885                 return seek_pos;
 886         } else if(rz->file_type == FILE_TYPE_GZ){
 887                 if(pos >= rz->out) goto SKIP;
 888                 return rz->out;
 889         }
 890         if(pos == rz->out) return pos;
 891         if(pos > rz->src_end) return rz->out;
 892         if(!rz->seekable || !rz->load_index){
 893                 if(pos >= rz->out) goto SKIP;
 894         }
 895         idx = pos / RZ_BLOCK_SIZE - 1;
 896         seek_pos = (idx < 0)? rz->header_size:(rz->index->cell_offsets[idx] + rz->index->bin_offsets[idx / RZ_BIN_SIZE]);
 897         new_out  = (idx + 1) * RZ_BLOCK_SIZE;
 898         if(pos > rz->out && new_out <= rz->out) goto SKIP;
 899         _razf_reset_read(rz, seek_pos, new_out);
 900         SKIP:
 901         razf_skip(rz, (int)(pos - rz->out));
 902         return rz->out;
 903 }
 904
 905 uint64_t razf_tell2(RAZF *rz)
 906 {
 907         /*
 908         if (rz->load_index) {
 909                 int64_t idx, seek_pos;
 910                 idx = rz->out / RZ_BLOCK_SIZE - 1;
 911                 seek_pos = (idx < 0)? rz->header_size:(rz->index->cell_offsets[idx] + rz->index->bin_offsets[idx / RZ_BIN_SIZE]);
 912                 if (seek_pos != rz->block_pos || rz->out%RZ_BLOCK_SIZE != rz->block_off)
 913                         fprintf(stderr, "[razf_tell2] inconsistent block offset: (%lld, %lld) != (%lld, %lld)\n",
 914                                         (long long)seek_pos, (long long)rz->out%RZ_BLOCK_SIZE, (long long)rz->block_pos, (long long) rz->block_off);
 915         }
 916         */
 917         return (uint64_t)rz->block_pos<<16 | (rz->block_off&0xffff);
 918 }
 919
 920 int64_t razf_seek2(RAZF *rz, uint64_t voffset, int where)
 921 {
 922         if (where != SEEK_SET) return -1;
 923         return razf_jump(rz, voffset>>16, voffset&0xffff);
 924 }
 925
 926 void razf_close(RAZF *rz){
 927         if(rz->mode == 'w'){
 928 #ifndef _RZ_READONLY
 929                 razf_end_flush(rz);
 930                 deflateEnd(rz->stream);
 931 #ifdef _USE_KNETFILE
 932                 save_zindex(rz, rz->x.fpw);
 933                 if(is_big_endian()){
 934                         if (write(rz->x.fpw, &rz->in, sizeof(int64_t)) < 0) {
 935                                 fprintf(stderr, "[%s] failed to write rz.in.\n", __func__);
 936                                 abort();
 937                         }
 938                         if (write(rz->x.fpw, &rz->out, sizeof(int64_t)) < 0) {
 939                                 fprintf(stderr, "[%s] failed to write rz.out.\n", __func__);
 940                                 abort();
 941                         }
 942                 } else {
 943                         uint64_t v64 = byte_swap_8((uint64_t)rz->in);
 944                         if (write(rz->x.fpw, &v64, sizeof(int64_t)) < 0) {
 945                                 fprintf(stderr, "[%s] failed to write rz.in.\n", __func__);
 946                                 abort();
 947                         }
 948                         v64 = byte_swap_8((uint64_t)rz->out);
 949                         if (write(rz->x.fpw, &v64, sizeof(int64_t)) < 0) {
 950                                 fprintf(stderr, "[%s] failed to write rz.out.\n", __func__);
 951                                 abort();
 952                         }
 953                 }
 954 #else
 955                 save_zindex(rz, rz->filedes);
 956                 if(is_big_endian()){
 957                         if (write(rz->filedes, &rz->in, sizeof(int64_t)) < 0) {
 958                                 fprintf(stderr, "[%s] failed to write rz.in.\n", __func__);
 959                                 abort();
 960                         }
 961                         if (write(rz->filedes, &rz->out, sizeof(int64_t)) < 0) {
 962                                 fprintf(stderr, "[%s] failed to write rz.out.\n", __func__);
 963                                 abort();
 964                         }
 965                 } else {
 966                         uint64_t v64 = byte_swap_8((uint64_t)rz->in);
 967                         if (write(rz->filedes, &v64, sizeof(int64_t)) < 0) {
 968                                 fprintf(stderr, "[%s] failed to write rz.in.\n", __func__);
 969                                 abort();
 970                         }
 971                         v64 = byte_swap_8((uint64_t)rz->out);
 972                         if (write(rz->filedes, &v64, sizeof(int64_t)) < 0) {
 973                                 fprintf(stderr, "[%s] failed to write rz.out.\n", __func__);
 974                                 abort();
 975                         }
 976                 }
 977 #endif
 978 #endif
 979         } else if(rz->mode == 'r'){
 980                 if(rz->stream) inflateEnd(rz->stream);
 981         }
 982         if(rz->inbuf) free(rz->inbuf);
 983         if(rz->outbuf) free(rz->outbuf);
 984         if(rz->header){
 985                 free(rz->header->extra);
 986                 free(rz->header->name);
 987                 free(rz->header->comment);
 988                 free(rz->header);
 989         }
 990         if(rz->index){
 991                 free(rz->index->bin_offsets);
 992                 free(rz->index->cell_offsets);
 993                 free(rz->index);
 994         }
 995         free(rz->stream);
 996 #ifdef _USE_KNETFILE
 997     if (rz->mode == 'r')
 998         knet_close(rz->x.fpr);
 999     if (rz->mode == 'w')
1000         close(rz->x.fpw);
1001 #else
1002         close(rz->filedes);
1003 #endif
1004         free(rz);
1005 }
1006
1007 #endif