| 1 |
#ifndef _GCDBYANK_H |
| 2 |
#define _GCDBYANK_H |
| 3 |
|
| 4 |
#include "gcdb.h" |
| 5 |
#include <stdio.h> |
| 6 |
#include "GFastaFile.h" |
| 7 |
// FastaSeq class and *charFunc() callback type |
| 8 |
|
| 9 |
#define DEF_CDBREC_DELIM ">" |
| 10 |
|
| 11 |
#ifdef ENABLE_COMPRESSION |
| 12 |
#include <zlib.h> |
| 13 |
#define GCDBZ_SBUF_LEN 8192 |
| 14 |
#define GCDBZ_LBUF_LEN 8192*2 |
| 15 |
|
| 16 |
class GCdbZFasta { |
| 17 |
private: |
| 18 |
char* recdelim; |
| 19 |
char lbuf[GCDBZ_LBUF_LEN]; //larger buffer |
| 20 |
char sbuf[GCDBZ_SBUF_LEN]; //smaller buffer |
| 21 |
char* defline; //defline copy storage -- compression only |
| 22 |
int defline_cap; //currently allocated length of defline |
| 23 |
int defline_len; //currently used length of defline |
| 24 |
z_stream zstream; // de/compression stream |
| 25 |
FILE* zf; //compressed file |
| 26 |
long zpos; //current position in zf |
| 27 |
int zrecsize; // the size of the compressed record |
| 28 |
GFastaCharHandler* chrhandler; |
| 29 |
public: |
| 30 |
GCdbZFasta(FILE* af, int zrsize=0, char* r_delim=DEF_CDBREC_DELIM); |
| 31 |
~GCdbZFasta(); |
| 32 |
FILE* getZFile() { return zf; } |
| 33 |
void decomp_start(int zrsize); |
| 34 |
void decomp_end(); |
| 35 |
int decompress(FastaSeq& rec, int csize=0, int zfofs=-1, charFunc* seqCallBack=NULL); |
| 36 |
// uncompress csize bytes from file zf, from optional file offset zfofs, |
| 37 |
// and send the uncompressed stream to callbackFn |
| 38 |
}; |
| 39 |
|
| 40 |
#endif |
| 41 |
|
| 42 |
class GCdbYank { |
| 43 |
char* idxfile; |
| 44 |
char* dbfile; |
| 45 |
char* recdelim; //record delimiter -- typically ">" |
| 46 |
int warnings; |
| 47 |
bool is_compressed; |
| 48 |
char* dbname; |
| 49 |
char* info_dbname; |
| 50 |
off_t db_size; |
| 51 |
cdbInfo dbstat; |
| 52 |
GCdbRead* cdb; |
| 53 |
#ifdef ENABLE_COMPRESSION |
| 54 |
GCdbZFasta* cdbz; |
| 55 |
#endif |
| 56 |
int fdb; |
| 57 |
int fd; |
| 58 |
FILE* fz; // if compressed |
| 59 |
GFastaCharHandler* fastahandler; |
| 60 |
#ifdef ENABLE_COMPRESSION |
| 61 |
protected: |
| 62 |
GCdbZFasta* openCdbz(char* p); |
| 63 |
#endif |
| 64 |
public: |
| 65 |
GCdbYank(const char* fidx, const char* recsep=DEF_CDBREC_DELIM); |
| 66 |
~GCdbYank(); |
| 67 |
int getRecord(const char* key, FastaSeq& rec, charFunc* seqCallBack=NULL); |
| 68 |
off_t getRecordPos(const char* key); |
| 69 |
char* getDbName() { return dbname; } |
| 70 |
|
| 71 |
}; |
| 72 |
|
| 73 |
#endif |
