static void flush_buffer(args_t *args, int n)
{
int i, j;
for (i=0; i<n; i++)
{
int k = rbuf_shift(&args->rbuf);
bcf1_t *rec = args->rbuf_lines[k];
int pass = 1;
if ( !args->soft_filter )
{
for (j=0; j<rec->d.n_flt; j++)
{
if ( args->indel_gap && rec->d.flt[j]==args->IndelGap_id ) { pass = 0; break; }
if ( args->snp_gap && rec->d.flt[j]==args->SnpGap_id ) { pass = 0; break; }
}
}
if ( pass ) bcf_write1(args->out_fh, args->hdr, rec);
}
}
bcf1_t *vcfbuf_flush(vcfbuf_t *buf, int flush_all)
{
int i,j;
if ( buf->rbuf.n==0 ) return NULL;
if ( flush_all ) goto ret;
i = rbuf_kth(&buf->rbuf, 0); // first
j = rbuf_last(&buf->rbuf); // last
if ( buf->vcf[i].rec->rid != buf->vcf[j].rec->rid ) goto ret;
if ( buf->overlap.active )
{
int ret = _overlap_can_flush(buf, flush_all);
//printf("can_flush: %d %d - %d\n", ret, buf->vcf[i].rec->pos+1, buf->vcf[j].rec->pos+1);
if ( ret ) goto ret;
}
//if ( buf->overlap.active && _overlap_can_flush(buf, flush_all) ) goto ret;
if ( buf->win > 0 )
{
if ( buf->rbuf.n <= buf->win ) return NULL;
goto ret;
}
else if ( buf->win < 0 )
{
if ( buf->vcf[i].rec->pos - buf->vcf[j].rec->pos > buf->win ) return NULL;
}
else return NULL;
ret:
if ( buf->prune.max_sites && buf->prune.max_sites < buf->rbuf.n ) _prune_sites(buf, flush_all);
i = rbuf_shift(&buf->rbuf);
return buf->vcf[i].rec;
}
static void consensus(args_t *args)
{
htsFile *fasta = hts_open(args->ref_fname, "rb");
if ( !fasta ) error("Error reading %s\n", args->ref_fname);
kstring_t str = {0,0,0};
while ( hts_getline(fasta, KS_SEP_LINE, &str) > 0 )
{
if ( str.s[0]=='>' )
{
// new sequence encountered, apply all chached variants
while ( args->vcf_rbuf.n )
{
if (args->chain) {
print_chain(args);
destroy_chain(args);
}
bcf1_t *rec = args->vcf_buf[args->vcf_rbuf.f];
if ( rec->rid!=args->rid || ( args->fa_end_pos && rec->pos > args->fa_end_pos ) ) break;
int i = rbuf_shift(&args->vcf_rbuf);
apply_variant(args, args->vcf_buf[i]);
}
flush_fa_buffer(args, 0);
init_region(args, str.s+1);
continue;
}
args->fa_length += str.l;
args->fa_src_pos += str.l;
// determine if uppercase or lowercase is used in this fasta file
if ( args->fa_case==-1 ) args->fa_case = toupper(str.s[0])==str.s[0] ? 1 : 0;
if ( args->mask && args->rid>=0) mask_region(args, str.s, str.l);
kputs(str.s, &args->fa_buf);
bcf1_t **rec_ptr = NULL;
while ( args->rid>=0 && (rec_ptr = next_vcf_line(args)) )
{
bcf1_t *rec = *rec_ptr;
// still the same chr and the same region? if not, fasta buf can be flushed
if ( rec->rid!=args->rid || ( args->fa_end_pos && rec->pos > args->fa_end_pos ) )
{
// save the vcf record until next time and flush
unread_vcf_line(args, rec_ptr);
rec_ptr = NULL;
break;
}
// is the vcf record well beyond cached fasta buffer? if yes, the buf can be flushed
if ( args->fa_ori_pos + args->fa_buf.l - args->fa_mod_off <= rec->pos )
{
unread_vcf_line(args, rec_ptr);
rec_ptr = NULL;
break;
}
// is the cached fasta buffer full enough? if not, read more fasta, no flushing
if ( args->fa_ori_pos + args->fa_buf.l - args->fa_mod_off < rec->pos + rec->rlen )
{
unread_vcf_line(args, rec_ptr);
break;
}
apply_variant(args, rec);
}
if ( !rec_ptr ) flush_fa_buffer(args, 60);
}
if (args->chain) {
print_chain(args);
destroy_chain(args);
}
flush_fa_buffer(args, 0);
hts_close(fasta);
free(str.s);
}
