Remove CNV:TR-related INFO tags when empty

If all the alleles referenced by CNV:TR related INFO tags are
removed, the INFO tags will no longer have any values and
so also need to be removed.

Author: daviesrob

test/test-vcf-api.c

@@ -982,6 +982,7 @@ void test_bcf_remove_allele_set(void)
         "5\t112000\t.\tC\t<CNV:TR>,<CNV:TR>\t.\tPASS\tRN=2,1;RUS=CAG,TTG,CA;RUL=3,3,2;RB=12,6,6;RUC=4,2,3;RUB=3,3,3,3,3,3,2,2,2;SVLEN=18,6",
         "5\t113000\t.\tT\tC,A\t.\tPASS\tAC=90,1;AD=6,5,6;AF=0.009,0.0001;VL_A_STR_INFO=alt_c,alt_a;VL_R_STR_INFO=ref,alt_c,alt_a\tGT:LAA:LAD:LEC:LPL:VL_LA_STR_FMT:VL_LG_STR_FMT:VL_LR_STR_FMT\t0/0:.:3:.:0:.:gt_00:ref\t0/1:1,2:3,2,0:44,27:114,0,15,35,73,113:alt_c,alt_a:gt_00,gt_01,gt_11,gt_02,gt_12,gt_22:ref,alt_c,alt_a\t1/1:1:0,3:46:110,15,0:alt_c:gt_00,gt_01,gt_11:ref,alt_c",
         "5\t114000\t.\tT\tC,A\t.\tPASS\tAC=90,1;AD=6,5,6;AF=0.009,0.0001;VL_A_STR_INFO=alt_c,alt_a;VL_R_STR_INFO=ref,alt_c,alt_a\tGT:LAA:LAD:LEC:LPL:VL_LA_STR_FMT:VL_LG_STR_FMT:VL_LR_STR_FMT\t0/0:.:3:.:0:.:gt_00:ref\t0/1:1,2:3,2,0:44,27:114,0,15,35,73,113:alt_c,alt_a:gt_00,gt_01,gt_11,gt_02,gt_12,gt_22:ref,alt_c,alt_a\t1/1:1:0,3:46:110,15,0:alt_c:gt_00,gt_01,gt_11:ref,alt_c",
+        "5\t115000\t.\tC\t<CNV:TR>,<CNV:TR>\t.\tPASS\tRN=2,1;RUS=CAG,TTG,CA;RUL=3,3,2;RB=12,6,6;RUC=4,2,3;RUB=3,3,3,3,3,3,2,2,2;SVLEN=18,6",
     };
     const char * expected[] = {
         // 2nd allele removed
@@ -995,6 +996,7 @@ void test_bcf_remove_allele_set(void)
         "5\t113000\t.\tT\tA\t.\tPASS\tAC=1;AD=6,6;AF=0.0001;VL_A_STR_INFO=alt_a;VL_R_STR_INFO=ref,alt_a\tGT:LAA:LAD:LEC:LPL:VL_LA_STR_FMT:VL_LG_STR_FMT:VL_LR_STR_FMT\t0/0:.:3:.:0:.:gt_00:ref\t0/.:1:3,0:27:114,35,113:alt_a:gt_00,gt_02,gt_22:ref,alt_a\t./.:.:0:.:110:.:gt_00:ref",
         // Both alleles removed
         "5\t114000\t.\tT\t.\t.\tPASS\tAD=6;VL_R_STR_INFO=ref\tGT:LAA:LAD:LEC:LPL:VL_LA_STR_FMT:VL_LG_STR_FMT:VL_LR_STR_FMT\t0/0:.:3:.:0:.:gt_00:ref\t0/.:.:3:.:114:.:gt_00:ref\t./.:.:0:.:110:.:gt_00:ref",
+        "5\t115000\t.\tC\t.\t.\tPASS\t.",
     };
 
     kstring_t kstr = KS_INITIALIZE;
@@ -1021,7 +1023,7 @@ void test_bcf_remove_allele_set(void)
         kbs_clear(rm_set);
         if (rec->pos == 113000 - 1) {
             kbs_insert(rm_set, 1);
-        } else if (rec->pos == 114000 - 1) {
+        } else if (rec->pos >= 114000 - 1) {
             kbs_insert(rm_set, 1);
             kbs_insert(rm_set, 2);
         } else {

vcfutils.c

@@ -404,6 +404,18 @@ static int fixup_info_length_code(bcf_info_t *info)
     return 0;
 }
 
+static int mark_for_removal(bcf_info_t *info)
+{
+    if (info->vptr_free)
+    {
+        free(info->vptr - info->vptr_off);
+        info->vptr_free = 0;
+    }
+    info->vptr = NULL;
+    info->vptr_off = info->vptr_len = 0;
+    return 0;
+}
+
 // Remove integer CNV:TR-related tag data
 // Returns 0 on success
 //         1 if the tag had an unexpected type or number of elements
@@ -468,6 +480,10 @@ static int trim_int_cnv_tr_int_tags(bcf_info_t *info,
         new_pos += byte_len;
         new_total += n_items;
     }
+
+    if (new_total == 0)
+        return mark_for_removal(info);
+
     info->vptr_len = new_pos;
     info->len = new_total;
     if (info->len == 1)
@@ -525,6 +541,9 @@ static int trim_int_cnv_tr_str_tags(bcf_info_t *info,
         orig_pos += end - start;
         new_pos += end - start;
     }
+    if (new_pos == 0)
+        return mark_for_removal(info);
+
     if (new_pos < orig_pos)
     {
         info->vptr[new_pos] = '\0';
@@ -620,10 +639,10 @@ static int fixup_cnv_tr_info_tags(const bcf_hdr_t *header, bcf1_t *line,
                 return res;
         }
 
-        // Check if storage had to be reallocated.  This can only happen if
-        // the length code stored before info->vptr was too small, which
-        // should hopefully never be the case.
-        if (info->vptr - info->vptr_off != orig_ptr)
+        // Check if the info tag was removed, or storage had to be reallocated.
+        // The latter can only happen if the length code stored before info->vptr
+        // was too small, which should hopefully never be the case.
+        if (!info->vptr || info->vptr - info->vptr_off != orig_ptr)
             line->d.shared_dirty |= BCF1_DIRTY_INF;
     }
     // Now do RUC, if present