Make reheader --fai aware of long contigs

by switching to 64-bit version of the faidx_seq_len() function

Resolves #1959

Author: pd3

NEWS

@@ -84,6 +84,10 @@ Changes affecting specific commands:
 
     - Fix `-H` header output in formatting expressions containing newlines
 
+* bcftools reheader
+
+    - Make `-f, --fai` aware of long contigs not representable by 32-bit integer (#1959)
+
 * bcftools +split-vep
 
     - Prevent a segfault when `-i/-e` use a VEP subfield not included in `-f` or `-c` (#1877)

reheader.c

@@ -68,7 +68,8 @@ static char *copy_and_update_contig_line(faidx_t *fai, char *line, void *chr_see
     kstring_t key = {0,0,0}, val = {0,0,0}, tmp = {0,0,0};
     char *chr_name = NULL, *p, *q = line + 9;   // skip ##contig=
     char *end = q;
-    int nopen = 1, chr_len = 0;
+    int nopen = 1;
+    hts_pos_t chr_len = 0;
     while ( *end && *end!='\n' ) end++;
     while ( *q && *q!='\n' && nopen>0 )
     {
@@ -118,7 +119,7 @@ static char *copy_and_update_contig_line(faidx_t *fai, char *line, void *chr_see
         if ( !strcmp("ID",key.s) )
         {
             if ( khash_str2int_has_key(chr_seen,val.s) ) continue;
-            chr_len = faidx_seq_len(fai, val.s);
+            chr_len = faidx_seq_len64(fai, val.s);
             if ( chr_len==-1 )
             {
                 free(val.s); free(key.s); free(tmp.s);
@@ -136,7 +137,7 @@ static char *copy_and_update_contig_line(faidx_t *fai, char *line, void *chr_see
         if ( quoted ) kputc('"',&tmp);
     }
     if ( !chr_name ) return end;
-    ksprintf(dst,"##contig=<ID=%s,length=%d%s>",chr_name,chr_len,tmp.l ? tmp.s : "");
+    ksprintf(dst,"##contig=<ID=%s,length=%"PRIhts_pos"%s>",chr_name,chr_len,tmp.l ? tmp.s : "");
     free(key.s); free(val.s); free(tmp.s);
     return q;
 }
@@ -211,7 +212,7 @@ static void update_from_fai(args_t *args)
     for (i=0; i<n; i++)
     {
         if ( khash_str2int_has_key(chr_seen,faidx_iseq(fai,i)) ) continue;
-        ksprintf(&hdr_txt_new,"##contig=<ID=%s,length=%d>\n",faidx_iseq(fai,i),faidx_seq_len(fai,faidx_iseq(fai,i)));
+        ksprintf(&hdr_txt_new,"##contig=<ID=%s,length=%"PRIhts_pos">\n",faidx_iseq(fai,i),faidx_seq_len64(fai,faidx_iseq(fai,i)));
     }
     kputs(tmp+1,&hdr_txt_new);
 
@@ -699,7 +700,7 @@ int main_reheader(int argc, char *argv[])
     int c;
     args_t *args  = (args_t*) calloc(1,sizeof(args_t));
     args->argc    = argc; args->argv = argv;
-    
+
     static struct option loptions[] =
     {
         {"temp-prefix",1,0,'T'},

test/reheader.3.fai

@@ -0,0 +1,12 @@
+1	2364278061	6	2364278061	2364278062
+2	2317450362	7159592787	2317450362	2317450363
+3	2291775479	9477043156	2291775479	2291775480
+4	2192534405	11768818642	2192534405	2192534406
+5	2148190925	13961353054	2148190925	2148190926
+6	2107674557	16109543986	2107674557	2107674558
+7	2082167746	18217218550	2082167746	2082167747
+8	2081484518	20299386303	2081484518	2081484519
+9	2024734096	22380870828	2024734096	2024734097
+10	1752849333	2364278075	1752849333	1752849334
+11	1650012615	4117127416	1650012615	1650012616
+12	1392452741	5767140039	1392452741	1392452742

test/reheader.3.vcf

@@ -0,0 +1,4 @@
+##fileformat=VCFv4.3
+##contig=<ID=1,length=1>
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+1	22	.	A	G	.	.	.

test/reheader.6.out

@@ -0,0 +1,16 @@
+##fileformat=VCFv4.3
+##FILTER=<ID=PASS,Description="All filters passed">
+##contig=<ID=1,length=2364278061>
+##contig=<ID=2,length=2317450362>
+##contig=<ID=3,length=2291775479>
+##contig=<ID=4,length=2192534405>
+##contig=<ID=5,length=2148190925>
+##contig=<ID=6,length=2107674557>
+##contig=<ID=7,length=2082167746>
+##contig=<ID=8,length=2081484518>
+##contig=<ID=9,length=2024734096>
+##contig=<ID=10,length=1752849333>
+##contig=<ID=11,length=1650012615>
+##contig=<ID=12,length=1392452741>
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+1	22	.	A	G	.	.	.

test/test.pl

@@ -729,6 +729,7 @@
 run_test(\&test_vcf_reheader,$opts,in=>'empty',out=>'reheader.empty.out',header=>'reheader.empty.hdr');
 run_test(\&test_vcf_reheader,$opts,in=>'reheader.2',out=>'reheader.5.out',args=>'-f {PATH}/reheader.fai',nostdin=>1);
 run_test(\&test_vcf_reheader,$opts,in=>'reheader.2',out=>'reheader.5.out',args=>'-h {PATH}/reheader.2.hdr -f {PATH}/reheader.fai',nostdin=>1);
+run_test(\&test_vcf_reheader,$opts,in=>'reheader.3',out=>'reheader.6.out',args=>'-f {PATH}/reheader.3.fai',nostdin=>1);
 run_test(\&test_rename_chrs,$opts,in=>'annotate');
 run_test(\&test_vcf_convert,$opts,in=>'convert',out=>'convert.gs.gt.gen',args=>'-g -,.');
 run_test(\&test_vcf_convert,$opts,in=>'convert',out=>'convert.gs.gt.ids.gen',args=>'-g -,. --vcf-ids');