ngs_server 0.4 → 0.5
Sign up to get free protection for your applications and to get access to all the features.
- data/ext/tabix/ChangeLog +593 -0
- data/ext/tabix/Makefile +65 -0
- data/ext/tabix/NEWS +126 -0
- data/ext/tabix/TabixReader.java +395 -0
- data/ext/tabix/bam_endian.h +42 -0
- data/ext/tabix/bedidx.c +156 -0
- data/ext/tabix/bgzf.c +714 -0
- data/ext/tabix/bgzf.h +157 -0
- data/ext/tabix/bgzip.c +206 -0
- data/ext/tabix/example.gtf.gz +0 -0
- data/ext/tabix/example.gtf.gz.tbi +0 -0
- data/ext/tabix/extconf.rb +1 -0
- data/ext/tabix/index.c +998 -0
- data/ext/tabix/khash.h +486 -0
- data/ext/tabix/knetfile.c +632 -0
- data/ext/tabix/knetfile.h +75 -0
- data/ext/tabix/kseq.h +227 -0
- data/ext/tabix/ksort.h +271 -0
- data/ext/tabix/kstring.c +165 -0
- data/ext/tabix/kstring.h +68 -0
- data/ext/tabix/main.c +290 -0
- data/ext/tabix/perl/MANIFEST +8 -0
- data/ext/tabix/perl/Makefile.PL +8 -0
- data/ext/tabix/perl/Tabix.pm +76 -0
- data/ext/tabix/perl/Tabix.xs +71 -0
- data/ext/tabix/perl/TabixIterator.pm +41 -0
- data/ext/tabix/perl/t/01local.t +28 -0
- data/ext/tabix/perl/t/02remote.t +28 -0
- data/ext/tabix/perl/typemap +3 -0
- data/ext/tabix/python/setup.py +55 -0
- data/ext/tabix/python/tabixmodule.c +408 -0
- data/ext/tabix/python/test.py +91 -0
- data/ext/tabix/tabix.1 +132 -0
- data/ext/tabix/tabix.h +145 -0
- data/ext/tabix/tabix.py +87 -0
- data/ext/tabix/tabix.tex +121 -0
- data/ext/vcftools/perl/Vcf.pm +5 -3
- data/ext/vcftools/perl/vcf-query +2 -0
- data/lib/ngs_server/version.rb +1 -1
- data/lib/ngs_server.rb +12 -11
- data/ngs_server.gemspec +1 -2
- metadata +39 -2
data/ext/tabix/bgzf.c
ADDED
@@ -0,0 +1,714 @@
|
|
1
|
+
/* The MIT License
|
2
|
+
|
3
|
+
Copyright (c) 2008 Broad Institute / Massachusetts Institute of Technology
|
4
|
+
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
7
|
+
in the Software without restriction, including without limitation the rights
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
10
|
+
furnished to do so, subject to the following conditions:
|
11
|
+
|
12
|
+
The above copyright notice and this permission notice shall be included in
|
13
|
+
all copies or substantial portions of the Software.
|
14
|
+
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
|
21
|
+
THE SOFTWARE.
|
22
|
+
*/
|
23
|
+
|
24
|
+
/*
|
25
|
+
2009-06-29 by lh3: cache recent uncompressed blocks.
|
26
|
+
2009-06-25 by lh3: optionally use my knetfile library to access file on a FTP.
|
27
|
+
2009-06-12 by lh3: support a mode string like "wu" where 'u' for uncompressed output */
|
28
|
+
|
29
|
+
#include <stdio.h>
|
30
|
+
#include <stdlib.h>
|
31
|
+
#include <string.h>
|
32
|
+
#include <unistd.h>
|
33
|
+
#include <fcntl.h>
|
34
|
+
#include <sys/types.h>
|
35
|
+
#include <sys/stat.h>
|
36
|
+
#include "bgzf.h"
|
37
|
+
|
38
|
+
#include "khash.h"
|
39
|
+
typedef struct {
|
40
|
+
int size;
|
41
|
+
uint8_t *block;
|
42
|
+
int64_t end_offset;
|
43
|
+
} cache_t;
|
44
|
+
KHASH_MAP_INIT_INT64(cache, cache_t)
|
45
|
+
|
46
|
+
#if defined(_WIN32) || defined(_MSC_VER)
|
47
|
+
#define ftello(fp) ftell(fp)
|
48
|
+
#define fseeko(fp, offset, whence) fseek(fp, offset, whence)
|
49
|
+
#else
|
50
|
+
extern off_t ftello(FILE *stream);
|
51
|
+
extern int fseeko(FILE *stream, off_t offset, int whence);
|
52
|
+
#endif
|
53
|
+
|
54
|
+
typedef int8_t bgzf_byte_t;
|
55
|
+
|
56
|
+
static const int DEFAULT_BLOCK_SIZE = 64 * 1024;
|
57
|
+
static const int MAX_BLOCK_SIZE = 64 * 1024;
|
58
|
+
|
59
|
+
static const int BLOCK_HEADER_LENGTH = 18;
|
60
|
+
static const int BLOCK_FOOTER_LENGTH = 8;
|
61
|
+
|
62
|
+
static const int GZIP_ID1 = 31;
|
63
|
+
static const int GZIP_ID2 = 139;
|
64
|
+
static const int CM_DEFLATE = 8;
|
65
|
+
static const int FLG_FEXTRA = 4;
|
66
|
+
static const int OS_UNKNOWN = 255;
|
67
|
+
static const int BGZF_ID1 = 66; // 'B'
|
68
|
+
static const int BGZF_ID2 = 67; // 'C'
|
69
|
+
static const int BGZF_LEN = 2;
|
70
|
+
static const int BGZF_XLEN = 6; // BGZF_LEN+4
|
71
|
+
|
72
|
+
static const int GZIP_WINDOW_BITS = -15; // no zlib header
|
73
|
+
static const int Z_DEFAULT_MEM_LEVEL = 8;
|
74
|
+
|
75
|
+
|
76
|
+
inline
|
77
|
+
void
|
78
|
+
packInt16(uint8_t* buffer, uint16_t value)
|
79
|
+
{
|
80
|
+
buffer[0] = value;
|
81
|
+
buffer[1] = value >> 8;
|
82
|
+
}
|
83
|
+
|
84
|
+
inline
|
85
|
+
int
|
86
|
+
unpackInt16(const uint8_t* buffer)
|
87
|
+
{
|
88
|
+
return (buffer[0] | (buffer[1] << 8));
|
89
|
+
}
|
90
|
+
|
91
|
+
inline
|
92
|
+
void
|
93
|
+
packInt32(uint8_t* buffer, uint32_t value)
|
94
|
+
{
|
95
|
+
buffer[0] = value;
|
96
|
+
buffer[1] = value >> 8;
|
97
|
+
buffer[2] = value >> 16;
|
98
|
+
buffer[3] = value >> 24;
|
99
|
+
}
|
100
|
+
|
101
|
+
static inline
|
102
|
+
int
|
103
|
+
bgzf_min(int x, int y)
|
104
|
+
{
|
105
|
+
return (x < y) ? x : y;
|
106
|
+
}
|
107
|
+
|
108
|
+
static
|
109
|
+
void
|
110
|
+
report_error(BGZF* fp, const char* message) {
|
111
|
+
fp->error = message;
|
112
|
+
}
|
113
|
+
|
114
|
+
int bgzf_check_bgzf(const char *fn)
|
115
|
+
{
|
116
|
+
BGZF *fp;
|
117
|
+
uint8_t buf[10],magic[10]="\037\213\010\4\0\0\0\0\0\377";
|
118
|
+
int n;
|
119
|
+
|
120
|
+
if ((fp = bgzf_open(fn, "r")) == 0)
|
121
|
+
{
|
122
|
+
fprintf(stderr, "[bgzf_check_bgzf] failed to open the file: %s\n",fn);
|
123
|
+
return -1;
|
124
|
+
}
|
125
|
+
|
126
|
+
#ifdef _USE_KNETFILE
|
127
|
+
n = knet_read(fp->x.fpr, buf, 10);
|
128
|
+
#else
|
129
|
+
n = fread(buf, 1, 10, fp->file);
|
130
|
+
#endif
|
131
|
+
bgzf_close(fp);
|
132
|
+
|
133
|
+
if ( n!=10 )
|
134
|
+
return -1;
|
135
|
+
|
136
|
+
if ( !memcmp(magic, buf, 10) ) return 1;
|
137
|
+
return 0;
|
138
|
+
}
|
139
|
+
|
140
|
+
static BGZF *bgzf_read_init()
|
141
|
+
{
|
142
|
+
BGZF *fp;
|
143
|
+
fp = calloc(1, sizeof(BGZF));
|
144
|
+
fp->uncompressed_block_size = MAX_BLOCK_SIZE;
|
145
|
+
fp->uncompressed_block = malloc(MAX_BLOCK_SIZE);
|
146
|
+
fp->compressed_block_size = MAX_BLOCK_SIZE;
|
147
|
+
fp->compressed_block = malloc(MAX_BLOCK_SIZE);
|
148
|
+
fp->cache_size = 0;
|
149
|
+
fp->cache = kh_init(cache);
|
150
|
+
return fp;
|
151
|
+
}
|
152
|
+
|
153
|
+
static
|
154
|
+
BGZF*
|
155
|
+
open_read(int fd)
|
156
|
+
{
|
157
|
+
#ifdef _USE_KNETFILE
|
158
|
+
knetFile *file = knet_dopen(fd, "r");
|
159
|
+
#else
|
160
|
+
FILE* file = fdopen(fd, "r");
|
161
|
+
#endif
|
162
|
+
BGZF* fp;
|
163
|
+
if (file == 0) return 0;
|
164
|
+
fp = bgzf_read_init();
|
165
|
+
fp->file_descriptor = fd;
|
166
|
+
fp->open_mode = 'r';
|
167
|
+
#ifdef _USE_KNETFILE
|
168
|
+
fp->x.fpr = file;
|
169
|
+
#else
|
170
|
+
fp->file = file;
|
171
|
+
#endif
|
172
|
+
return fp;
|
173
|
+
}
|
174
|
+
|
175
|
+
static
|
176
|
+
BGZF*
|
177
|
+
open_write(int fd, int compress_level) // compress_level==-1 for the default level
|
178
|
+
{
|
179
|
+
FILE* file = fdopen(fd, "w");
|
180
|
+
BGZF* fp;
|
181
|
+
if (file == 0) return 0;
|
182
|
+
fp = malloc(sizeof(BGZF));
|
183
|
+
fp->file_descriptor = fd;
|
184
|
+
fp->open_mode = 'w';
|
185
|
+
fp->owned_file = 0;
|
186
|
+
fp->compress_level = compress_level < 0? Z_DEFAULT_COMPRESSION : compress_level; // Z_DEFAULT_COMPRESSION==-1
|
187
|
+
if (fp->compress_level > 9) fp->compress_level = Z_DEFAULT_COMPRESSION;
|
188
|
+
#ifdef _USE_KNETFILE
|
189
|
+
fp->x.fpw = file;
|
190
|
+
#else
|
191
|
+
fp->file = file;
|
192
|
+
#endif
|
193
|
+
fp->uncompressed_block_size = DEFAULT_BLOCK_SIZE;
|
194
|
+
fp->uncompressed_block = NULL;
|
195
|
+
fp->compressed_block_size = MAX_BLOCK_SIZE;
|
196
|
+
fp->compressed_block = malloc(MAX_BLOCK_SIZE);
|
197
|
+
fp->block_address = 0;
|
198
|
+
fp->block_offset = 0;
|
199
|
+
fp->block_length = 0;
|
200
|
+
fp->error = NULL;
|
201
|
+
return fp;
|
202
|
+
}
|
203
|
+
|
204
|
+
BGZF*
|
205
|
+
bgzf_open(const char* __restrict path, const char* __restrict mode)
|
206
|
+
{
|
207
|
+
BGZF* fp = NULL;
|
208
|
+
if (strchr(mode, 'r') || strchr(mode, 'R')) { /* The reading mode is preferred. */
|
209
|
+
#ifdef _USE_KNETFILE
|
210
|
+
knetFile *file = knet_open(path, mode);
|
211
|
+
if (file == 0) return 0;
|
212
|
+
fp = bgzf_read_init();
|
213
|
+
fp->file_descriptor = -1;
|
214
|
+
fp->open_mode = 'r';
|
215
|
+
fp->x.fpr = file;
|
216
|
+
#else
|
217
|
+
int fd, oflag = O_RDONLY;
|
218
|
+
#ifdef _WIN32
|
219
|
+
oflag |= O_BINARY;
|
220
|
+
#endif
|
221
|
+
fd = open(path, oflag);
|
222
|
+
if (fd == -1) return 0;
|
223
|
+
fp = open_read(fd);
|
224
|
+
#endif
|
225
|
+
} else if (strchr(mode, 'w') || strchr(mode, 'W')) {
|
226
|
+
int fd, compress_level = -1, oflag = O_WRONLY | O_CREAT | O_TRUNC;
|
227
|
+
#ifdef _WIN32
|
228
|
+
oflag |= O_BINARY;
|
229
|
+
#endif
|
230
|
+
fd = open(path, oflag, 0666);
|
231
|
+
if (fd == -1) return 0;
|
232
|
+
{ // set compress_level
|
233
|
+
int i;
|
234
|
+
for (i = 0; mode[i]; ++i)
|
235
|
+
if (mode[i] >= '0' && mode[i] <= '9') break;
|
236
|
+
if (mode[i]) compress_level = (int)mode[i] - '0';
|
237
|
+
if (strchr(mode, 'u')) compress_level = 0;
|
238
|
+
}
|
239
|
+
fp = open_write(fd, compress_level);
|
240
|
+
}
|
241
|
+
if (fp != NULL) fp->owned_file = 1;
|
242
|
+
return fp;
|
243
|
+
}
|
244
|
+
|
245
|
+
BGZF*
|
246
|
+
bgzf_fdopen(int fd, const char * __restrict mode)
|
247
|
+
{
|
248
|
+
if (fd == -1) return 0;
|
249
|
+
if (mode[0] == 'r' || mode[0] == 'R') {
|
250
|
+
return open_read(fd);
|
251
|
+
} else if (mode[0] == 'w' || mode[0] == 'W') {
|
252
|
+
int i, compress_level = -1;
|
253
|
+
for (i = 0; mode[i]; ++i)
|
254
|
+
if (mode[i] >= '0' && mode[i] <= '9') break;
|
255
|
+
if (mode[i]) compress_level = (int)mode[i] - '0';
|
256
|
+
if (strchr(mode, 'u')) compress_level = 0;
|
257
|
+
return open_write(fd, compress_level);
|
258
|
+
} else {
|
259
|
+
return NULL;
|
260
|
+
}
|
261
|
+
}
|
262
|
+
|
263
|
+
static
|
264
|
+
int
|
265
|
+
deflate_block(BGZF* fp, int block_length)
|
266
|
+
{
|
267
|
+
// Deflate the block in fp->uncompressed_block into fp->compressed_block.
|
268
|
+
// Also adds an extra field that stores the compressed block length.
|
269
|
+
|
270
|
+
bgzf_byte_t* buffer = fp->compressed_block;
|
271
|
+
int buffer_size = fp->compressed_block_size;
|
272
|
+
|
273
|
+
// Init gzip header
|
274
|
+
buffer[0] = GZIP_ID1;
|
275
|
+
buffer[1] = GZIP_ID2;
|
276
|
+
buffer[2] = CM_DEFLATE;
|
277
|
+
buffer[3] = FLG_FEXTRA;
|
278
|
+
buffer[4] = 0; // mtime
|
279
|
+
buffer[5] = 0;
|
280
|
+
buffer[6] = 0;
|
281
|
+
buffer[7] = 0;
|
282
|
+
buffer[8] = 0;
|
283
|
+
buffer[9] = OS_UNKNOWN;
|
284
|
+
buffer[10] = BGZF_XLEN;
|
285
|
+
buffer[11] = 0;
|
286
|
+
buffer[12] = BGZF_ID1;
|
287
|
+
buffer[13] = BGZF_ID2;
|
288
|
+
buffer[14] = BGZF_LEN;
|
289
|
+
buffer[15] = 0;
|
290
|
+
buffer[16] = 0; // placeholder for block length
|
291
|
+
buffer[17] = 0;
|
292
|
+
|
293
|
+
// loop to retry for blocks that do not compress enough
|
294
|
+
int input_length = block_length;
|
295
|
+
int compressed_length = 0;
|
296
|
+
while (1) {
|
297
|
+
z_stream zs;
|
298
|
+
zs.zalloc = NULL;
|
299
|
+
zs.zfree = NULL;
|
300
|
+
zs.next_in = fp->uncompressed_block;
|
301
|
+
zs.avail_in = input_length;
|
302
|
+
zs.next_out = (void*)&buffer[BLOCK_HEADER_LENGTH];
|
303
|
+
zs.avail_out = buffer_size - BLOCK_HEADER_LENGTH - BLOCK_FOOTER_LENGTH;
|
304
|
+
|
305
|
+
int status = deflateInit2(&zs, fp->compress_level, Z_DEFLATED,
|
306
|
+
GZIP_WINDOW_BITS, Z_DEFAULT_MEM_LEVEL, Z_DEFAULT_STRATEGY);
|
307
|
+
if (status != Z_OK) {
|
308
|
+
report_error(fp, "deflate init failed");
|
309
|
+
return -1;
|
310
|
+
}
|
311
|
+
status = deflate(&zs, Z_FINISH);
|
312
|
+
if (status != Z_STREAM_END) {
|
313
|
+
deflateEnd(&zs);
|
314
|
+
if (status == Z_OK) {
|
315
|
+
// Not enough space in buffer.
|
316
|
+
// Can happen in the rare case the input doesn't compress enough.
|
317
|
+
// Reduce the amount of input until it fits.
|
318
|
+
input_length -= 1024;
|
319
|
+
if (input_length <= 0) {
|
320
|
+
// should never happen
|
321
|
+
report_error(fp, "input reduction failed");
|
322
|
+
return -1;
|
323
|
+
}
|
324
|
+
continue;
|
325
|
+
}
|
326
|
+
report_error(fp, "deflate failed");
|
327
|
+
return -1;
|
328
|
+
}
|
329
|
+
status = deflateEnd(&zs);
|
330
|
+
if (status != Z_OK) {
|
331
|
+
report_error(fp, "deflate end failed");
|
332
|
+
return -1;
|
333
|
+
}
|
334
|
+
compressed_length = zs.total_out;
|
335
|
+
compressed_length += BLOCK_HEADER_LENGTH + BLOCK_FOOTER_LENGTH;
|
336
|
+
if (compressed_length > MAX_BLOCK_SIZE) {
|
337
|
+
// should never happen
|
338
|
+
report_error(fp, "deflate overflow");
|
339
|
+
return -1;
|
340
|
+
}
|
341
|
+
break;
|
342
|
+
}
|
343
|
+
|
344
|
+
packInt16((uint8_t*)&buffer[16], compressed_length-1);
|
345
|
+
uint32_t crc = crc32(0L, NULL, 0L);
|
346
|
+
crc = crc32(crc, fp->uncompressed_block, input_length);
|
347
|
+
packInt32((uint8_t*)&buffer[compressed_length-8], crc);
|
348
|
+
packInt32((uint8_t*)&buffer[compressed_length-4], input_length);
|
349
|
+
|
350
|
+
int remaining = block_length - input_length;
|
351
|
+
if (remaining > 0) {
|
352
|
+
if (remaining > input_length) {
|
353
|
+
// should never happen (check so we can use memcpy)
|
354
|
+
report_error(fp, "remainder too large");
|
355
|
+
return -1;
|
356
|
+
}
|
357
|
+
memcpy(fp->uncompressed_block,
|
358
|
+
fp->uncompressed_block + input_length,
|
359
|
+
remaining);
|
360
|
+
}
|
361
|
+
fp->block_offset = remaining;
|
362
|
+
return compressed_length;
|
363
|
+
}
|
364
|
+
|
365
|
+
static
|
366
|
+
int
|
367
|
+
inflate_block(BGZF* fp, int block_length)
|
368
|
+
{
|
369
|
+
// Inflate the block in fp->compressed_block into fp->uncompressed_block
|
370
|
+
|
371
|
+
z_stream zs;
|
372
|
+
int status;
|
373
|
+
zs.zalloc = NULL;
|
374
|
+
zs.zfree = NULL;
|
375
|
+
zs.next_in = fp->compressed_block + 18;
|
376
|
+
zs.avail_in = block_length - 16;
|
377
|
+
zs.next_out = fp->uncompressed_block;
|
378
|
+
zs.avail_out = fp->uncompressed_block_size;
|
379
|
+
|
380
|
+
status = inflateInit2(&zs, GZIP_WINDOW_BITS);
|
381
|
+
if (status != Z_OK) {
|
382
|
+
report_error(fp, "inflate init failed");
|
383
|
+
return -1;
|
384
|
+
}
|
385
|
+
status = inflate(&zs, Z_FINISH);
|
386
|
+
if (status != Z_STREAM_END) {
|
387
|
+
inflateEnd(&zs);
|
388
|
+
report_error(fp, "inflate failed");
|
389
|
+
return -1;
|
390
|
+
}
|
391
|
+
status = inflateEnd(&zs);
|
392
|
+
if (status != Z_OK) {
|
393
|
+
report_error(fp, "inflate failed");
|
394
|
+
return -1;
|
395
|
+
}
|
396
|
+
return zs.total_out;
|
397
|
+
}
|
398
|
+
|
399
|
+
static
|
400
|
+
int
|
401
|
+
check_header(const bgzf_byte_t* header)
|
402
|
+
{
|
403
|
+
return (header[0] == GZIP_ID1 &&
|
404
|
+
header[1] == (bgzf_byte_t) GZIP_ID2 &&
|
405
|
+
header[2] == Z_DEFLATED &&
|
406
|
+
(header[3] & FLG_FEXTRA) != 0 &&
|
407
|
+
unpackInt16((uint8_t*)&header[10]) == BGZF_XLEN &&
|
408
|
+
header[12] == BGZF_ID1 &&
|
409
|
+
header[13] == BGZF_ID2 &&
|
410
|
+
unpackInt16((uint8_t*)&header[14]) == BGZF_LEN);
|
411
|
+
}
|
412
|
+
|
413
|
+
static void free_cache(BGZF *fp)
|
414
|
+
{
|
415
|
+
khint_t k;
|
416
|
+
khash_t(cache) *h = (khash_t(cache)*)fp->cache;
|
417
|
+
if (fp->open_mode != 'r') return;
|
418
|
+
for (k = kh_begin(h); k < kh_end(h); ++k)
|
419
|
+
if (kh_exist(h, k)) free(kh_val(h, k).block);
|
420
|
+
kh_destroy(cache, h);
|
421
|
+
}
|
422
|
+
|
423
|
+
static int load_block_from_cache(BGZF *fp, int64_t block_address)
|
424
|
+
{
|
425
|
+
khint_t k;
|
426
|
+
cache_t *p;
|
427
|
+
khash_t(cache) *h = (khash_t(cache)*)fp->cache;
|
428
|
+
k = kh_get(cache, h, block_address);
|
429
|
+
if (k == kh_end(h)) return 0;
|
430
|
+
p = &kh_val(h, k);
|
431
|
+
if (fp->block_length != 0) fp->block_offset = 0;
|
432
|
+
fp->block_address = block_address;
|
433
|
+
fp->block_length = p->size;
|
434
|
+
memcpy(fp->uncompressed_block, p->block, MAX_BLOCK_SIZE);
|
435
|
+
#ifdef _USE_KNETFILE
|
436
|
+
knet_seek(fp->x.fpr, p->end_offset, SEEK_SET);
|
437
|
+
#else
|
438
|
+
fseeko(fp->file, p->end_offset, SEEK_SET);
|
439
|
+
#endif
|
440
|
+
return p->size;
|
441
|
+
}
|
442
|
+
|
443
|
+
static void cache_block(BGZF *fp, int size)
|
444
|
+
{
|
445
|
+
int ret;
|
446
|
+
khint_t k;
|
447
|
+
cache_t *p;
|
448
|
+
khash_t(cache) *h = (khash_t(cache)*)fp->cache;
|
449
|
+
if (MAX_BLOCK_SIZE >= fp->cache_size) return;
|
450
|
+
if ((kh_size(h) + 1) * MAX_BLOCK_SIZE > fp->cache_size) {
|
451
|
+
/* A better way would be to remove the oldest block in the
|
452
|
+
* cache, but here we remove a random one for simplicity. This
|
453
|
+
* should not have a big impact on performance. */
|
454
|
+
for (k = kh_begin(h); k < kh_end(h); ++k)
|
455
|
+
if (kh_exist(h, k)) break;
|
456
|
+
if (k < kh_end(h)) {
|
457
|
+
free(kh_val(h, k).block);
|
458
|
+
kh_del(cache, h, k);
|
459
|
+
}
|
460
|
+
}
|
461
|
+
k = kh_put(cache, h, fp->block_address, &ret);
|
462
|
+
if (ret == 0) return; // if this happens, a bug!
|
463
|
+
p = &kh_val(h, k);
|
464
|
+
p->size = fp->block_length;
|
465
|
+
p->end_offset = fp->block_address + size;
|
466
|
+
p->block = malloc(MAX_BLOCK_SIZE);
|
467
|
+
memcpy(kh_val(h, k).block, fp->uncompressed_block, MAX_BLOCK_SIZE);
|
468
|
+
}
|
469
|
+
|
470
|
+
int
|
471
|
+
bgzf_read_block(BGZF* fp)
|
472
|
+
{
|
473
|
+
bgzf_byte_t header[BLOCK_HEADER_LENGTH];
|
474
|
+
int count, size = 0, block_length, remaining;
|
475
|
+
#ifdef _USE_KNETFILE
|
476
|
+
int64_t block_address = knet_tell(fp->x.fpr);
|
477
|
+
if (load_block_from_cache(fp, block_address)) return 0;
|
478
|
+
count = knet_read(fp->x.fpr, header, sizeof(header));
|
479
|
+
#else
|
480
|
+
int64_t block_address = ftello(fp->file);
|
481
|
+
if (load_block_from_cache(fp, block_address)) return 0;
|
482
|
+
count = fread(header, 1, sizeof(header), fp->file);
|
483
|
+
#endif
|
484
|
+
if (count == 0) {
|
485
|
+
fp->block_length = 0;
|
486
|
+
return 0;
|
487
|
+
}
|
488
|
+
size = count;
|
489
|
+
if (count != sizeof(header)) {
|
490
|
+
report_error(fp, "read failed");
|
491
|
+
return -1;
|
492
|
+
}
|
493
|
+
if (!check_header(header)) {
|
494
|
+
report_error(fp, "invalid block header");
|
495
|
+
return -1;
|
496
|
+
}
|
497
|
+
block_length = unpackInt16((uint8_t*)&header[16]) + 1;
|
498
|
+
bgzf_byte_t* compressed_block = (bgzf_byte_t*) fp->compressed_block;
|
499
|
+
memcpy(compressed_block, header, BLOCK_HEADER_LENGTH);
|
500
|
+
remaining = block_length - BLOCK_HEADER_LENGTH;
|
501
|
+
#ifdef _USE_KNETFILE
|
502
|
+
count = knet_read(fp->x.fpr, &compressed_block[BLOCK_HEADER_LENGTH], remaining);
|
503
|
+
#else
|
504
|
+
count = fread(&compressed_block[BLOCK_HEADER_LENGTH], 1, remaining, fp->file);
|
505
|
+
#endif
|
506
|
+
if (count != remaining) {
|
507
|
+
report_error(fp, "read failed");
|
508
|
+
return -1;
|
509
|
+
}
|
510
|
+
size += count;
|
511
|
+
count = inflate_block(fp, block_length);
|
512
|
+
if (count < 0) return -1;
|
513
|
+
if (fp->block_length != 0) {
|
514
|
+
// Do not reset offset if this read follows a seek.
|
515
|
+
fp->block_offset = 0;
|
516
|
+
}
|
517
|
+
fp->block_address = block_address;
|
518
|
+
fp->block_length = count;
|
519
|
+
cache_block(fp, size);
|
520
|
+
return 0;
|
521
|
+
}
|
522
|
+
|
523
|
+
int
|
524
|
+
bgzf_read(BGZF* fp, void* data, int length)
|
525
|
+
{
|
526
|
+
if (length <= 0) {
|
527
|
+
return 0;
|
528
|
+
}
|
529
|
+
if (fp->open_mode != 'r') {
|
530
|
+
report_error(fp, "file not open for reading");
|
531
|
+
return -1;
|
532
|
+
}
|
533
|
+
|
534
|
+
int bytes_read = 0;
|
535
|
+
bgzf_byte_t* output = data;
|
536
|
+
while (bytes_read < length) {
|
537
|
+
int copy_length, available = fp->block_length - fp->block_offset;
|
538
|
+
bgzf_byte_t *buffer;
|
539
|
+
if (available <= 0) {
|
540
|
+
if (bgzf_read_block(fp) != 0) {
|
541
|
+
return -1;
|
542
|
+
}
|
543
|
+
available = fp->block_length - fp->block_offset;
|
544
|
+
if (available <= 0) {
|
545
|
+
break;
|
546
|
+
}
|
547
|
+
}
|
548
|
+
copy_length = bgzf_min(length-bytes_read, available);
|
549
|
+
buffer = fp->uncompressed_block;
|
550
|
+
memcpy(output, buffer + fp->block_offset, copy_length);
|
551
|
+
fp->block_offset += copy_length;
|
552
|
+
output += copy_length;
|
553
|
+
bytes_read += copy_length;
|
554
|
+
}
|
555
|
+
if (fp->block_offset == fp->block_length) {
|
556
|
+
#ifdef _USE_KNETFILE
|
557
|
+
fp->block_address = knet_tell(fp->x.fpr);
|
558
|
+
#else
|
559
|
+
fp->block_address = ftello(fp->file);
|
560
|
+
#endif
|
561
|
+
fp->block_offset = 0;
|
562
|
+
fp->block_length = 0;
|
563
|
+
}
|
564
|
+
return bytes_read;
|
565
|
+
}
|
566
|
+
|
567
|
+
int bgzf_flush(BGZF* fp)
|
568
|
+
{
|
569
|
+
while (fp->block_offset > 0) {
|
570
|
+
int count, block_length;
|
571
|
+
block_length = deflate_block(fp, fp->block_offset);
|
572
|
+
if (block_length < 0) return -1;
|
573
|
+
#ifdef _USE_KNETFILE
|
574
|
+
count = fwrite(fp->compressed_block, 1, block_length, fp->x.fpw);
|
575
|
+
#else
|
576
|
+
count = fwrite(fp->compressed_block, 1, block_length, fp->file);
|
577
|
+
#endif
|
578
|
+
if (count != block_length) {
|
579
|
+
report_error(fp, "write failed");
|
580
|
+
return -1;
|
581
|
+
}
|
582
|
+
fp->block_address += block_length;
|
583
|
+
}
|
584
|
+
return 0;
|
585
|
+
}
|
586
|
+
|
587
|
+
int bgzf_flush_try(BGZF *fp, int size)
|
588
|
+
{
|
589
|
+
if (fp->block_offset + size > fp->uncompressed_block_size)
|
590
|
+
return bgzf_flush(fp);
|
591
|
+
return -1;
|
592
|
+
}
|
593
|
+
|
594
|
+
int bgzf_write(BGZF* fp, const void* data, int length)
|
595
|
+
{
|
596
|
+
const bgzf_byte_t *input = data;
|
597
|
+
int block_length, bytes_written;
|
598
|
+
if (fp->open_mode != 'w') {
|
599
|
+
report_error(fp, "file not open for writing");
|
600
|
+
return -1;
|
601
|
+
}
|
602
|
+
|
603
|
+
if (fp->uncompressed_block == NULL)
|
604
|
+
fp->uncompressed_block = malloc(fp->uncompressed_block_size);
|
605
|
+
|
606
|
+
input = data;
|
607
|
+
block_length = fp->uncompressed_block_size;
|
608
|
+
bytes_written = 0;
|
609
|
+
while (bytes_written < length) {
|
610
|
+
int copy_length = bgzf_min(block_length - fp->block_offset, length - bytes_written);
|
611
|
+
bgzf_byte_t* buffer = fp->uncompressed_block;
|
612
|
+
memcpy(buffer + fp->block_offset, input, copy_length);
|
613
|
+
fp->block_offset += copy_length;
|
614
|
+
input += copy_length;
|
615
|
+
bytes_written += copy_length;
|
616
|
+
if (fp->block_offset == block_length) {
|
617
|
+
if (bgzf_flush(fp) != 0) {
|
618
|
+
break;
|
619
|
+
}
|
620
|
+
}
|
621
|
+
}
|
622
|
+
return bytes_written;
|
623
|
+
}
|
624
|
+
|
625
|
+
int bgzf_close(BGZF* fp)
|
626
|
+
{
|
627
|
+
if (fp->open_mode == 'w') {
|
628
|
+
if (bgzf_flush(fp) != 0) return -1;
|
629
|
+
{ // add an empty block
|
630
|
+
int count, block_length = deflate_block(fp, 0);
|
631
|
+
#ifdef _USE_KNETFILE
|
632
|
+
count = fwrite(fp->compressed_block, 1, block_length, fp->x.fpw);
|
633
|
+
#else
|
634
|
+
count = fwrite(fp->compressed_block, 1, block_length, fp->file);
|
635
|
+
#endif
|
636
|
+
}
|
637
|
+
#ifdef _USE_KNETFILE
|
638
|
+
if (fflush(fp->x.fpw) != 0) {
|
639
|
+
#else
|
640
|
+
if (fflush(fp->file) != 0) {
|
641
|
+
#endif
|
642
|
+
report_error(fp, "flush failed");
|
643
|
+
return -1;
|
644
|
+
}
|
645
|
+
}
|
646
|
+
if (fp->owned_file) {
|
647
|
+
#ifdef _USE_KNETFILE
|
648
|
+
int ret;
|
649
|
+
if (fp->open_mode == 'w') ret = fclose(fp->x.fpw);
|
650
|
+
else ret = knet_close(fp->x.fpr);
|
651
|
+
if (ret != 0) return -1;
|
652
|
+
#else
|
653
|
+
if (fclose(fp->file) != 0) return -1;
|
654
|
+
#endif
|
655
|
+
}
|
656
|
+
free(fp->uncompressed_block);
|
657
|
+
free(fp->compressed_block);
|
658
|
+
free_cache(fp);
|
659
|
+
free(fp);
|
660
|
+
return 0;
|
661
|
+
}
|
662
|
+
|
663
|
+
void bgzf_set_cache_size(BGZF *fp, int cache_size)
|
664
|
+
{
|
665
|
+
if (fp) fp->cache_size = cache_size;
|
666
|
+
}
|
667
|
+
|
668
|
+
int bgzf_check_EOF(BGZF *fp)
|
669
|
+
{
|
670
|
+
static uint8_t magic[28] = "\037\213\010\4\0\0\0\0\0\377\6\0\102\103\2\0\033\0\3\0\0\0\0\0\0\0\0\0";
|
671
|
+
uint8_t buf[28];
|
672
|
+
off_t offset;
|
673
|
+
#ifdef _USE_KNETFILE
|
674
|
+
offset = knet_tell(fp->x.fpr);
|
675
|
+
if (knet_seek(fp->x.fpr, -28, SEEK_END) != 0) return -1;
|
676
|
+
knet_read(fp->x.fpr, buf, 28);
|
677
|
+
knet_seek(fp->x.fpr, offset, SEEK_SET);
|
678
|
+
#else
|
679
|
+
offset = ftello(fp->file);
|
680
|
+
if (fseeko(fp->file, -28, SEEK_END) != 0) return -1;
|
681
|
+
fread(buf, 1, 28, fp->file);
|
682
|
+
fseeko(fp->file, offset, SEEK_SET);
|
683
|
+
#endif
|
684
|
+
return (memcmp(magic, buf, 28) == 0)? 1 : 0;
|
685
|
+
}
|
686
|
+
|
687
|
+
int64_t bgzf_seek(BGZF* fp, int64_t pos, int where)
|
688
|
+
{
|
689
|
+
int block_offset;
|
690
|
+
int64_t block_address;
|
691
|
+
|
692
|
+
if (fp->open_mode != 'r') {
|
693
|
+
report_error(fp, "file not open for read");
|
694
|
+
return -1;
|
695
|
+
}
|
696
|
+
if (where != SEEK_SET) {
|
697
|
+
report_error(fp, "unimplemented seek option");
|
698
|
+
return -1;
|
699
|
+
}
|
700
|
+
block_offset = pos & 0xFFFF;
|
701
|
+
block_address = (pos >> 16) & 0xFFFFFFFFFFFFLL;
|
702
|
+
#ifdef _USE_KNETFILE
|
703
|
+
if (knet_seek(fp->x.fpr, block_address, SEEK_SET) != 0) {
|
704
|
+
#else
|
705
|
+
if (fseeko(fp->file, block_address, SEEK_SET) != 0) {
|
706
|
+
#endif
|
707
|
+
report_error(fp, "seek failed");
|
708
|
+
return -1;
|
709
|
+
}
|
710
|
+
fp->block_length = 0; // indicates current block is not loaded
|
711
|
+
fp->block_address = block_address;
|
712
|
+
fp->block_offset = block_offset;
|
713
|
+
return 0;
|
714
|
+
}
|