2 * RAZF : Random Access compressed(Z) File
4 * Release Date: 2008-10-27
6 * Copyright 2008, Jue Ruan <ruanjue@gmail.com>, Heng Li <lh3@sanger.ac.uk>
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
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.
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
41 #if ZLIB_VERNUM < 0x1221
57 #warning "zlib < 1.2.2.1; RAZF writing is disabled."
60 #define DEF_MEM_LEVEL 8
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);
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);
73 static inline int is_big_endian(){
76 return (c[0] != 0x01);
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));
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];
91 static void save_zindex(RAZF *rz, int fd){
94 is_be = is_big_endian();
95 if(is_be) write(fd, &rz->index->size, sizeof(int));
97 v32 = byte_swap_4((uint32_t)rz->index->size);
98 write(fd, &v32, sizeof(uint32_t));
100 v32 = rz->index->size / RZ_BIN_SIZE + 1;
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]);
105 write(fd, rz->index->bin_offsets, sizeof(int64_t) * v32);
106 write(fd, rz->index->cell_offsets, sizeof(int32_t) * rz->index->size);
110 static void load_zindex(RAZF *rz, int fd){
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);
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]);
131 static RAZF* razf_open_w(int fd)
133 fprintf(stderr, "[razf_open_w] Writing is not available with zlib ver < 1.2.2.1\n");
137 static RAZF* razf_open_w(int fd){
140 setmode(fd, O_BINARY);
142 rz = calloc(1, sizeof(RAZF));
145 rz->stream = calloc(sizeof(z_stream), 1);
146 rz->inbuf = malloc(RZ_BUFFER_SIZE);
147 rz->outbuf = malloc(RZ_BUFFER_SIZE);
148 rz->index = calloc(sizeof(ZBlockIndex), 1);
149 deflateInit2(rz->stream, RZ_COMPRESS_LEVEL, Z_DEFLATED, WINDOW_BITS + 16, DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY);
150 rz->stream->avail_out = RZ_BUFFER_SIZE;
151 rz->stream->next_out = rz->outbuf;
152 rz->header = calloc(sizeof(gz_header), 1);
153 rz->header->os = 0x03; //Unix
154 rz->header->text = 0;
155 rz->header->time = 0;
156 rz->header->extra = malloc(7);
157 strncpy((char*)rz->header->extra, "RAZF", 4);
158 rz->header->extra[4] = 1; // obsolete field
159 // block size = RZ_BLOCK_SIZE, Big-Endian
160 rz->header->extra[5] = RZ_BLOCK_SIZE >> 8;
161 rz->header->extra[6] = RZ_BLOCK_SIZE & 0xFF;
162 rz->header->extra_len = 7;
163 rz->header->name = rz->header->comment = 0;
164 rz->header->hcrc = 0;
165 deflateSetHeader(rz->stream, rz->header);
166 rz->block_pos = rz->block_off = 0;
170 static void _razf_write(RAZF* rz, const void *data, int size){
172 rz->stream->avail_in = size;
173 rz->stream->next_in = (void*)data;
175 tout = rz->stream->avail_out;
176 deflate(rz->stream, Z_NO_FLUSH);
177 rz->out += tout - rz->stream->avail_out;
178 if(rz->stream->avail_out) break;
179 write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
180 rz->stream->avail_out = RZ_BUFFER_SIZE;
181 rz->stream->next_out = rz->outbuf;
182 if(rz->stream->avail_in == 0) break;
184 rz->in += size - rz->stream->avail_in;
185 rz->block_off += size - rz->stream->avail_in;
188 static void razf_flush(RAZF *rz){
191 _razf_write(rz, rz->inbuf, rz->buf_len);
192 rz->buf_off = rz->buf_len = 0;
194 if(rz->stream->avail_out){
195 write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
196 rz->stream->avail_out = RZ_BUFFER_SIZE;
197 rz->stream->next_out = rz->outbuf;
200 tout = rz->stream->avail_out;
201 deflate(rz->stream, Z_FULL_FLUSH);
202 rz->out += tout - rz->stream->avail_out;
203 if(rz->stream->avail_out == 0){
204 write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
205 rz->stream->avail_out = RZ_BUFFER_SIZE;
206 rz->stream->next_out = rz->outbuf;
209 rz->block_pos = rz->out;
213 static void razf_end_flush(RAZF *rz){
216 _razf_write(rz, rz->inbuf, rz->buf_len);
217 rz->buf_off = rz->buf_len = 0;
220 tout = rz->stream->avail_out;
221 deflate(rz->stream, Z_FINISH);
222 rz->out += tout - rz->stream->avail_out;
223 if(rz->stream->avail_out < RZ_BUFFER_SIZE){
224 write(rz->filedes, rz->outbuf, RZ_BUFFER_SIZE - rz->stream->avail_out);
225 rz->stream->avail_out = RZ_BUFFER_SIZE;
226 rz->stream->next_out = rz->outbuf;
231 static void _razf_buffered_write(RAZF *rz, const void *data, int size){
234 if(rz->buf_len == RZ_BUFFER_SIZE){
235 _razf_write(rz, rz->inbuf, rz->buf_len);
238 if(size + rz->buf_len < RZ_BUFFER_SIZE){
239 for(i=0;i<size;i++) ((char*)rz->inbuf + rz->buf_len)[i] = ((char*)data)[i];
243 n = RZ_BUFFER_SIZE - rz->buf_len;
244 for(i=0;i<n;i++) ((char*)rz->inbuf + rz->buf_len)[i] = ((char*)data)[i];
252 int razf_write(RAZF* rz, const void *data, int size){
256 next_block = ((rz->in / RZ_BLOCK_SIZE) + 1) * RZ_BLOCK_SIZE;
257 while(rz->in + rz->buf_len + size >= next_block){
258 n = next_block - rz->in - rz->buf_len;
259 _razf_buffered_write(rz, data, n);
263 add_zindex(rz, rz->in, rz->out);
264 next_block = ((rz->in / RZ_BLOCK_SIZE) + 1) * RZ_BLOCK_SIZE;
266 _razf_buffered_write(rz, data, size);
272 #define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
273 #define HEAD_CRC 0x02 /* bit 1 set: header CRC present */
274 #define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
275 #define ORIG_NAME 0x08 /* bit 3 set: original file name present */
276 #define COMMENT 0x10 /* bit 4 set: file comment present */
277 #define RESERVED 0xE0 /* bits 5..7: reserved */
279 static int _read_gz_header(unsigned char *data, int size, int *extra_off, int *extra_len){
280 int method, flags, n, len;
281 if(size < 2) return 0;
282 if(data[0] != 0x1f || data[1] != 0x8b) return 0;
283 if(size < 4) return 0;
286 if(method != Z_DEFLATED || (flags & RESERVED)) return 0;
287 n = 4 + 6; // Skip 6 bytes
290 if(flags & EXTRA_FIELD){
291 if(size < n + 2) return 0;
292 len = ((int)data[n + 1] << 8) | data[n];
296 if(n >= size) return 0;
300 *extra_len = n - (*extra_off);
302 if(flags & ORIG_NAME) while(n < size && data[n++]);
303 if(flags & COMMENT) while(n < size && data[n++]);
304 if(flags & HEAD_CRC){
305 if(n + 2 > size) return 0;
311 static RAZF* razf_open_r(int fd, int _load_index){
313 int ext_off, ext_len;
316 unsigned char c[] = "RAZF";
318 setmode(fd, O_BINARY);
320 rz = calloc(1, sizeof(RAZF));
323 rz->stream = calloc(sizeof(z_stream), 1);
324 rz->inbuf = malloc(RZ_BUFFER_SIZE);
325 rz->outbuf = malloc(RZ_BUFFER_SIZE);
326 rz->end = rz->src_end = 0x7FFFFFFFFFFFFFFFLL;
327 n = read(rz->filedes, rz->inbuf, RZ_BUFFER_SIZE);
328 ret = _read_gz_header(rz->inbuf, n, &ext_off, &ext_len);
332 rz->file_type = FILE_TYPE_PLAIN;
333 memcpy(rz->outbuf, rz->inbuf, n);
339 rz->header_size = ret;
340 ret = inflateInit2(rz->stream, -WINDOW_BITS);
341 if(ret != Z_OK){ inflateEnd(rz->stream); goto PLAIN_FILE;}
342 rz->stream->avail_in = n - rz->header_size;
343 rz->stream->next_in = rz->inbuf + rz->header_size;
344 rz->stream->avail_out = RZ_BUFFER_SIZE;
345 rz->stream->next_out = rz->outbuf;
346 rz->file_type = FILE_TYPE_GZ;
347 rz->in = rz->header_size;
348 rz->block_pos = rz->header_size;
349 rz->next_block_pos = rz->header_size;
351 if(ext_len < 7 || memcmp(rz->inbuf + ext_off, c, 4) != 0) return rz;
352 if(((((unsigned char*)rz->inbuf)[ext_off + 5] << 8) | ((unsigned char*)rz->inbuf)[ext_off + 6]) != RZ_BLOCK_SIZE){
353 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__);
356 rz->load_index = _load_index;
357 rz->file_type = FILE_TYPE_RZ;
358 if(lseek(fd, -16, SEEK_END) == -1){
362 rz->src_end = rz->end = 0x7FFFFFFFFFFFFFFFLL;
364 is_be = is_big_endian();
366 read(fd, &end, sizeof(int64_t));
367 if(!is_be) rz->src_end = (int64_t)byte_swap_8((uint64_t)end);
368 else rz->src_end = end;
369 read(fd, &end, sizeof(int64_t));
370 if(!is_be) rz->end = (int64_t)byte_swap_8((uint64_t)end);
373 rz->stream->avail_in -= n - rz->end;
376 if(rz->end > rz->src_end){
377 lseek(fd, rz->in, SEEK_SET);
380 if(lseek(fd, rz->end, SEEK_SET) != rz->end){
381 lseek(fd, rz->in, SEEK_SET);
385 lseek(fd, n, SEEK_SET);
390 RAZF* razf_dopen(int fd, const char *mode){
391 if(strstr(mode, "r")) return razf_open_r(fd, 1);
392 else if(strstr(mode, "w")) return razf_open_w(fd);
396 RAZF* razf_dopen2(int fd, const char *mode)
398 if(strstr(mode, "r")) return razf_open_r(fd, 0);
399 else if(strstr(mode, "w")) return razf_open_w(fd);
403 static inline RAZF* _razf_open(const char *filename, const char *mode, int _load_index){
406 if(strstr(mode, "r")){
408 fd = open(filename, O_RDONLY | O_BINARY);
410 fd = open(filename, O_RDONLY);
412 if(fd < 0) return NULL;
413 rz = razf_open_r(fd, _load_index);
414 } else if(strstr(mode, "w")){
416 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0644);
418 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC, 0644);
420 if(fd < 0) return NULL;
421 rz = razf_open_w(fd);
426 RAZF* razf_open(const char *filename, const char *mode){
427 return _razf_open(filename, mode, 1);
430 RAZF* razf_open2(const char *filename, const char *mode){
431 return _razf_open(filename, mode, 0);
434 int razf_get_data_size(RAZF *rz, int64_t *u_size, int64_t *c_size){
436 if(rz->mode != 'r' && rz->mode != 'R') return 0;
437 switch(rz->file_type){
438 case FILE_TYPE_PLAIN:
439 if(rz->end == 0x7fffffffffffffffLL){
440 if((n = lseek(rz->filedes, 0, SEEK_CUR)) == -1) return 0;
441 rz->end = lseek(rz->filedes, 0, SEEK_END);
442 lseek(rz->filedes, n, SEEK_SET);
444 *u_size = *c_size = rz->end;
449 if(rz->src_end == rz->end) return 0;
450 *u_size = rz->src_end;
458 static int _razf_read(RAZF* rz, void *data, int size){
460 if(rz->z_eof || rz->z_err) return 0;
461 if (rz->file_type == FILE_TYPE_PLAIN) {
462 ret = read(rz->filedes, data, size);
463 if (ret == 0) rz->z_eof = 1;
466 rz->stream->avail_out = size;
467 rz->stream->next_out = data;
468 while(rz->stream->avail_out){
469 if(rz->stream->avail_in == 0){
470 if(rz->in >= rz->end){ rz->z_eof = 1; break; }
471 if(rz->end - rz->in < RZ_BUFFER_SIZE){
472 rz->stream->avail_in = read(rz->filedes, rz->inbuf, rz->end -rz->in);
474 rz->stream->avail_in = read(rz->filedes, rz->inbuf, RZ_BUFFER_SIZE);
476 if(rz->stream->avail_in == 0){
480 rz->stream->next_in = rz->inbuf;
482 tin = rz->stream->avail_in;
483 ret = inflate(rz->stream, Z_BLOCK);
484 rz->in += tin - rz->stream->avail_in;
485 if(ret == Z_NEED_DICT || ret == Z_MEM_ERROR || ret == Z_DATA_ERROR){
486 fprintf(stderr, "[_razf_read] inflate error: %d (at %s:%d)\n", ret, __FILE__, __LINE__);
490 if(ret == Z_STREAM_END){
494 if ((rz->stream->data_type&128) && !(rz->stream->data_type&64)){
496 rz->next_block_pos = rz->in;
500 return size - rz->stream->avail_out;
503 int razf_read(RAZF *rz, void *data, int size){
508 if(size < rz->buf_len){
509 for(i=0;i<size;i++) ((char*)data)[i] = ((char*)rz->outbuf + rz->buf_off)[i];
513 rz->block_off += size;
517 for(i=0;i<rz->buf_len;i++) ((char*)data)[i] = ((char*)rz->outbuf + rz->buf_off)[i];
520 rz->block_off += rz->buf_len;
524 rz->block_pos = rz->next_block_pos;
529 } else if(rz->buf_flush){
530 rz->block_pos = rz->next_block_pos;
534 if(rz->buf_flush) continue;
535 rz->buf_len = _razf_read(rz, rz->outbuf, RZ_BUFFER_SIZE);
536 if(rz->z_eof && rz->buf_len == 0) break;
538 rz->out += ori_size - size;
539 return ori_size - size;
542 int razf_skip(RAZF* rz, int size){
547 if(size < rz->buf_len){
550 rz->block_off += size;
557 rz->block_off += rz->buf_len;
559 rz->block_pos = rz->next_block_pos;
564 } else if(rz->buf_flush){
565 rz->block_pos = rz->next_block_pos;
569 if(rz->buf_flush) continue;
570 rz->buf_len = _razf_read(rz, rz->outbuf, RZ_BUFFER_SIZE);
573 rz->out += ori_size - size;
574 return ori_size - size;
577 static void _razf_reset_read(RAZF *rz, int64_t in, int64_t out){
578 lseek(rz->filedes, in, SEEK_SET);
582 rz->next_block_pos = in;
585 rz->z_eof = rz->z_err = 0;
586 inflateReset(rz->stream);
587 rz->stream->avail_in = 0;
588 rz->buf_off = rz->buf_len = 0;
591 int64_t razf_jump(RAZF *rz, int64_t block_start, int block_offset){
594 if(rz->file_type == FILE_TYPE_PLAIN){
595 rz->buf_off = rz->buf_len = 0;
596 pos = block_start + block_offset;
597 pos = lseek(rz->filedes, pos, SEEK_SET);
598 rz->out = rz->in = pos;
601 if(block_start == rz->block_pos && block_offset >= rz->block_off) {
602 block_offset -= rz->block_off;
603 goto SKIP; // Needn't reset inflate
605 if(block_start == 0) block_start = rz->header_size; // Automaticly revist wrong block_start
606 _razf_reset_read(rz, block_start, 0);
608 if(block_offset) razf_skip(rz, block_offset);
609 return rz->block_off;
612 int64_t razf_seek(RAZF* rz, int64_t pos, int where){
614 int64_t seek_pos, new_out;
616 if (where == SEEK_CUR) pos += rz->out;
617 else if (where == SEEK_END) pos += rz->src_end;
618 if(rz->file_type == FILE_TYPE_PLAIN){
619 seek_pos = lseek(rz->filedes, pos, SEEK_SET);
620 rz->buf_off = rz->buf_len = 0;
621 rz->out = rz->in = seek_pos;
623 } else if(rz->file_type == FILE_TYPE_GZ){
624 if(pos >= rz->out) goto SKIP;
627 if(pos == rz->out) return pos;
628 if(pos > rz->src_end) return rz->out;
629 if(!rz->seekable || !rz->load_index){
630 if(pos >= rz->out) goto SKIP;
632 idx = pos / RZ_BLOCK_SIZE - 1;
633 seek_pos = (idx < 0)? rz->header_size:(rz->index->cell_offsets[idx] + rz->index->bin_offsets[idx / RZ_BIN_SIZE]);
634 new_out = (idx + 1) * RZ_BLOCK_SIZE;
635 if(pos > rz->out && new_out <= rz->out) goto SKIP;
636 _razf_reset_read(rz, seek_pos, new_out);
638 razf_skip(rz, (int)(pos - rz->out));
642 uint64_t razf_tell2(RAZF *rz)
645 if (rz->load_index) {
646 int64_t idx, seek_pos;
647 idx = rz->out / RZ_BLOCK_SIZE - 1;
648 seek_pos = (idx < 0)? rz->header_size:(rz->index->cell_offsets[idx] + rz->index->bin_offsets[idx / RZ_BIN_SIZE]);
649 if (seek_pos != rz->block_pos || rz->out%RZ_BLOCK_SIZE != rz->block_off)
650 fprintf(stderr, "[razf_tell2] inconsistent block offset: (%lld, %lld) != (%lld, %lld)\n",
651 (long long)seek_pos, (long long)rz->out%RZ_BLOCK_SIZE, (long long)rz->block_pos, (long long) rz->block_off);
654 return (uint64_t)rz->block_pos<<16 | (rz->block_off&0xffff);
657 int64_t razf_seek2(RAZF *rz, uint64_t voffset, int where)
659 if (where != SEEK_SET) return -1;
660 return razf_jump(rz, voffset>>16, voffset&0xffff);
663 void razf_close(RAZF *rz){
667 deflateEnd(rz->stream);
668 save_zindex(rz, rz->filedes);
670 write(rz->filedes, &rz->in, sizeof(int64_t));
671 write(rz->filedes, &rz->out, sizeof(int64_t));
673 uint64_t v64 = byte_swap_8((uint64_t)rz->in);
674 write(rz->filedes, &v64, sizeof(int64_t));
675 v64 = byte_swap_8((uint64_t)rz->out);
676 write(rz->filedes, &v64, sizeof(int64_t));
679 } else if(rz->mode == 'r'){
680 if(rz->stream) inflateEnd(rz->stream);
682 if(rz->inbuf) free(rz->inbuf);
683 if(rz->outbuf) free(rz->outbuf);
685 free(rz->header->extra);
686 free(rz->header->name);
687 free(rz->header->comment);
691 free(rz->index->bin_offsets);
692 free(rz->index->cell_offsets);