static int set_sample_pairs(char **samples, int nsamples, kstring_t *hdr, int idx)
{
int i, j, n;
// Are these samples "old-name new-name" pairs?
void *hash = khash_str2str_init();
for (i=0; i<nsamples; i++)
{
char *key, *value;
key = value = samples[i];
while ( *value && !isspace(*value) ) value++;
if ( !*value ) break;
*value = 0; value++;
while ( isspace(*value) ) value++;
khash_str2str_set(hash,key,value);
}
if ( i!=nsamples ) // not "old-name new-name" pairs
{
khash_str2str_destroy(hash);
return 0;
}
while ( hdr->l>0 && isspace(hdr->s[hdr->l-1]) ) hdr->l--; // remove trailing newlines
hdr->s[hdr->l] = 0;
kstring_t tmp = {0,0,0};
i = j = n = 0;
while ( hdr->s[idx+i] && hdr->s[idx+i])
{
if ( hdr->s[idx+i]=='\t' )
{
hdr->s[idx+i] = 0;
if ( ++n>9 )
{
char *ori = khash_str2str_get(hash,hdr->s+idx+j);
kputs(ori ? ori : hdr->s+idx+j, &tmp);
}
else
kputs(hdr->s+idx+j, &tmp);
kputc('\t',&tmp);
j = ++i;
continue;
}
i++;
}
char *ori = khash_str2str_get(hash,hdr->s+idx+j);
kputs(ori ? ori : hdr->s+idx+j, &tmp);
if ( hash ) khash_str2str_destroy(hash);
hdr->l = idx;
kputs(tmp.s, hdr);
kputc('\n', hdr);
free(tmp.s);
return 1;
}
int bam_smpl_add_samples(bam_smpl_t *bsmpl, char *list, int is_file)
{
if ( list[0]!='^' ) bsmpl->sample_logic = 1;
else list++;
int i, nsamples = 0;
char **samples = hts_readlist(list, is_file, &nsamples);
if ( !nsamples ) return 0;
kstring_t ori = {0,0,0};
kstring_t ren = {0,0,0};
bsmpl->sample_list = khash_str2str_init();
for (i=0; i<nsamples; i++)
{
char *ptr = samples[i];
ori.l = ren.l = 0;
int escaped = 0;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &ori);
escaped = 0;
ptr++;
}
if ( *ptr )
{
while ( *ptr && isspace(*ptr) ) ptr++;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &ren);
escaped = 0;
ptr++;
}
}
khash_str2str_set(bsmpl->sample_list,strdup(ori.s),strdup(ren.l?ren.s:ori.s));
free(samples[i]);
}
free(samples);
free(ori.s);
free(ren.s);
return nsamples;
}
int bam_smpl_add_readgroups(bam_smpl_t *bsmpl, char *list, int is_file)
{
if ( list[0]!='^' ) bsmpl->rg_logic = 1;
else list++;
int i, nrows = 0;
char **rows = hts_readlist(list, is_file, &nrows);
if ( !nrows ) return 0;
kstring_t fld1 = {0,0,0};
kstring_t fld2 = {0,0,0};
kstring_t fld3 = {0,0,0};
bsmpl->rg_list = khash_str2str_init();
for (i=0; i<nrows; i++)
{
char *ptr = rows[i];
fld1.l = fld2.l = fld3.l = 0;
int escaped = 0;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &fld1);
escaped = 0;
ptr++;
}
if ( *ptr )
{
while ( *ptr && isspace(*ptr) ) ptr++;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &fld2);
escaped = 0;
ptr++;
}
}
if ( *ptr )
{
while ( *ptr && isspace(*ptr) ) ptr++;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &fld3);
escaped = 0;
ptr++;
}
}
if ( fld3.l )
{
// ID FILE SAMPLE
kputc('\t',&fld1);
kputs(fld2.s,&fld1);
fld2.l = 0;
kputs(fld3.s,&fld2);
}
// fld2.s now contains a new sample name. If NULL, use \t to keep the bam header name
char *value = khash_str2str_get(bsmpl->rg_list,fld1.s);
if ( !value )
khash_str2str_set(bsmpl->rg_list,strdup(fld1.s),strdup(fld2.l?fld2.s:"\t"));
else if ( strcmp(value,fld2.l?fld2.s:"\t") )
error("Error: The read group \"%s\" was assigned to two different samples: \"%s\" and \"%s\"\n", fld1.s,value,fld2.l?fld2.s:"\t");
free(rows[i]);
}
free(rows);
free(fld1.s);
free(fld2.s);
free(fld3.s);
return nrows;
}
