RefTran(char *ns) {
new_name=NULL;
if (ns!=NULL)
new_name=Gstrdup(ns);
}
int GFastaIndex::buildIndex() {
//this parses the whole fasta file, so it could be slow
if (fa_name==NULL)
GError("Error: GFastaIndex::buildIndex() called with no fasta file!\n");
FILE* fa=fopen(fa_name,"rb");
if (fa==NULL) {
GMessage("Warning: cannot open fasta index file: %s!\n",fa_name);
return 0;
}
records.Clear();
GLineReader fl(fa);
char* s=NULL;
uint seqlen=0;
int line_len=0,line_blen=0;
bool newSeq=false; //set to true after defline
off_t newSeqOffset=0;
int prevOffset=0;
char* seqname=NULL;
int last_len=0;
bool mustbeLastLine=false; //true if the line length decreases
while ((s=fl.nextLine())!=NULL) {
if (s[0]=='>') {
if (seqname!=NULL) {
if (seqlen==0)
GError("Warning: empty FASTA record skipped (%s)!\n",seqname);
else { //seqlen!=0
addRecord(seqname, seqlen,newSeqOffset, line_len, line_blen);
}
}
char *p=s;
while (*p > 32) p++;
*p=0;
GFREE(seqname);
seqname=Gstrdup(&s[1]);
newSeq=true;
newSeqOffset=fl.getfpos();
last_len=0;
line_len=0;
line_blen=0;
seqlen=0;
mustbeLastLine=false;
} //defline parsing
else { //sequence line
int llen=fl.length();
int lblen=fl.getFpos()-prevOffset;
if (newSeq) { //first sequence line after defline
line_len=llen;
line_blen=lblen;
}
else {//next seq lines after first
if (mustbeLastLine || llen>last_len)
GError(ERR_FALINELEN);
if (llen<last_len) mustbeLastLine=true;
}
seqlen+=llen;
last_len=llen;
newSeq=false;
} //sequence line
prevOffset=fl.getfpos();
}//for each line of the fasta file
if (seqlen>0)
addRecord(seqname, seqlen, newSeqOffset, line_len, line_blen);
GFREE(seqname);
fclose(fa);
return records.Count();
}
