| 1 |
gpertea |
29 |
/* |
| 2 |
|
|
* GFaIdx.h |
| 3 |
|
|
* |
| 4 |
|
|
* Created on: Aug 25, 2010 |
| 5 |
|
|
* Author: gpertea |
| 6 |
|
|
*/ |
| 7 |
|
|
|
| 8 |
|
|
#ifndef GFAIDX_H_ |
| 9 |
|
|
#define GFAIDX_H_ |
| 10 |
|
|
|
| 11 |
|
|
#include "GHash.hh" |
| 12 |
|
|
#include "GList.hh" |
| 13 |
|
|
|
| 14 |
|
|
class GFastaRec { |
| 15 |
|
|
public: |
| 16 |
|
|
char* seqname; |
| 17 |
|
|
uint seqlen; |
| 18 |
|
|
off_t fpos; |
| 19 |
|
|
int line_len; //effective line length (without EoL) |
| 20 |
|
|
int line_blen; //length of line including EoL characters |
| 21 |
|
|
GFastaRec(uint slen=0, off_t fp=0, int llen=0, int llenb=0) { |
| 22 |
|
|
seqname=NULL; //only a pointer copy |
| 23 |
|
|
seqlen=slen; |
| 24 |
|
|
fpos=fp; |
| 25 |
|
|
line_len=llen; |
| 26 |
|
|
line_blen=llenb; |
| 27 |
|
|
} |
| 28 |
|
|
bool operator==(GFastaRec& d){ |
| 29 |
|
|
return (fpos==d.fpos); |
| 30 |
|
|
} |
| 31 |
|
|
bool operator>(GFastaRec& d){ |
| 32 |
|
|
return (fpos>d.fpos); |
| 33 |
|
|
} |
| 34 |
|
|
bool operator<(GFastaRec& d){ |
| 35 |
|
|
return (fpos<d.fpos); |
| 36 |
|
|
} |
| 37 |
|
|
|
| 38 |
|
|
}; |
| 39 |
|
|
|
| 40 |
|
|
class GFastaIndex { |
| 41 |
|
|
char* fa_name; |
| 42 |
|
|
char* fai_name; |
| 43 |
|
|
bool haveFai; |
| 44 |
|
|
public: |
| 45 |
|
|
GHash<GFastaRec> records; |
| 46 |
|
|
void addRecord(const char* seqname, uint seqlen, |
| 47 |
|
|
off_t foffs, int llen, int llen_full); |
| 48 |
|
|
|
| 49 |
|
|
GFastaRec* getRecord(const char* seqname) { |
| 50 |
|
|
return records.Find(seqname); |
| 51 |
|
|
} |
| 52 |
|
|
bool hasIndex() { return haveFai; } |
| 53 |
|
|
int loadIndex(const char* finame); |
| 54 |
|
|
int buildIndex(); //build index in memory by parsing the whole fasta file |
| 55 |
|
|
int storeIndex(const char* finame); |
| 56 |
|
|
int storeIndex(FILE* fai); |
| 57 |
|
|
int getCount() { return records.Count(); } |
| 58 |
|
|
GFastaIndex(const char* fname, const char* finame=NULL):records() { |
| 59 |
|
|
if (fileExists(fname)!=2) GError("Error: fasta file %s not found!\n",fname); |
| 60 |
|
|
if (fileSize(fname)<=0) GError("Error: invalid fasta file %s !\n",fname); |
| 61 |
|
|
fa_name=Gstrdup(fname); |
| 62 |
|
|
fai_name=finame!=NULL ? Gstrdup(finame) : NULL; |
| 63 |
|
|
if (fileSize(fa_name)==0) { |
| 64 |
|
|
GError("Error creating GFastaIndex(%s): invalid fasta file!\n",fa_name); |
| 65 |
|
|
} |
| 66 |
|
|
haveFai=false; |
| 67 |
|
|
if (fai_name!=NULL && fileSize(fai_name)>0) { |
| 68 |
|
|
//try to load the index file if it exists |
| 69 |
|
|
loadIndex(fai_name); |
| 70 |
|
|
haveFai=(records.Count()>0); |
| 71 |
|
|
} |
| 72 |
|
|
} |
| 73 |
|
|
~GFastaIndex() { |
| 74 |
|
|
GFREE(fa_name); |
| 75 |
|
|
GFREE(fai_name); |
| 76 |
|
|
} |
| 77 |
|
|
}; |
| 78 |
|
|
|
| 79 |
|
|
#endif /* GFAIDX_H_ */ |
