update vcfreader

Zilong-Li · Jun 28, 2024 · 99ca8df · 99ca8df
1 parent 4973704
commit 99ca8df
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 19 deletions.
diff --git a/src/vcf-reader.cpp b/src/vcf-reader.cpp
@@ -155,26 +155,38 @@ class vcfreader {
 
     int infoInt(std::string tag) {
         int i;
-        var.getINFO(tag, i);
-        return i;
+        if (var.getINFO(tag, i)) {
+            return i;
+        } else {
+            return NA_INTEGER;
+        }
     }
     double infoFloat(std::string tag) {
         float f;
-        var.getINFO(tag, f);
-        return (double)f;
+        if (var.getINFO(tag, f)) {
+            return (double)f;
+        } else {
+            return NA_REAL;
+        }
     }
     std::string infoStr(std::string tag) {
-        std::string s;
+        std::string s{""};
         var.getINFO(tag, s);
         return s;
     }
     vector<int> infoIntVec(std::string tag) {
-        var.getINFO(tag, v_int);
-        return v_int;
+        if (var.getINFO(tag, v_int)) {
+            return v_int;
+        } else {
+            return vector<int>();
+        }
     }
     vector<double> infoFloatVec(std::string tag) {
-        var.getINFO(tag, v_float);
-        return vector<double>(v_float.begin(), v_float.end());
+        if (var.getINFO(tag, v_float)) {
+            return vector<double>(v_float.begin(), v_float.end());
+        } else {
+            return vector<double>();
+        }
     }
 
     vector<int> genotypes(bool collapse) {
@@ -197,10 +209,7 @@ class vcfreader {
     }
 
     vector<int> formatInt(std::string tag) {
-        if (!var.getFORMAT(tag, v_int)) {
-            vector<int> vna;
-            return vna;
-        }
+        if (!var.getFORMAT(tag, v_int)) { return vector<int>(); }
         int nvals = v_int.size() / br.nsamples;  // how many values per sample
         for (int i = 0; i < br.nsamples; i++) {
             for (int j = 0; j < nvals; j++)
@@ -229,11 +238,11 @@ class vcfreader {
     }
 
     vector<std::string> formatStr(std::string tag) {
-        if (!var.getFORMAT(tag, v_str)) {
-            vector<std::string> vstr;
-            return vstr;
+        if (var.getFORMAT(tag, v_str)) {
+            return v_str;
+        } else {
+            return vector<std::string>();
         }
-        return v_str;
     }
 
     inline bool isSNP() const { return var.isSNP(); }

diff --git a/src/vcfpp.h b/src/vcfpp.h
@@ -672,7 +672,7 @@ class BcfRecord
         fmt = bcf_get_fmt(header->hdr, line.get(), tag.c_str());
         if(!fmt)
         {
-            throw std::invalid_argument("no FORMAT=" + tag + " in the VCF header.\n");
+            throw std::invalid_argument("invalid FORMAT=" + tag + " for current variant.\n");
         }
         nvalues = fmt->n;
         // if ndst < (fmt->n+1)*nsmpl; then realloc is involved

diff --git a/tests/testthat/test-vcf-reader.R b/tests/testthat/test-vcf-reader.R
@@ -193,7 +193,7 @@ test_that("vcfreader: remove tag from FORMAT", {
   expect_identical(s[9], "GT:DP:GQ:PL")
   expect_identical(s[10], "1/1:2:6:64,6,0")
   ## AD was removed, so we get integer(0)
-  ad <- br$formatInt("AD") 
+  ad <- br$formatInt("AD")
   expect_identical(length(ad),0L)
   ## output current variant to another vcf
   outvcf <- file.path(tempdir(), "test.vcf.gz")
@@ -206,5 +206,8 @@ test_that("vcfreader: remove tag from FORMAT", {
   br$variant()
   s <- unlist(strsplit(br$line(), "\t"))
   expect_identical(s[9], "GT:DP:GQ:PL")
+  expect_identical(s[10], "1/1:2:6:64,6,0")
+  ## AD doesn't exist. so we get error, is this a bad design?
   expect_error(br$formatInt("AD"))
 })
+