Merge pull request #402 from broadinstitute/intel_header_field_length

Added functions in VariantHeader to query field length descriptors and field lengths.
broadinstitute · Dec 2, 2014 · d923b53 · d923b53
2 parents 816bca0 + 3a79ceb
commit d923b53
Show file tree

Hide file tree

Showing 2 changed files with 40 additions and 1 deletion.
diff --git a/gamgee/variant/variant_header.h b/gamgee/variant/variant_header.h
@@ -123,7 +123,30 @@ class VariantHeader {
    * @note must check whether the field exists before calling this function, as it doesn't check for you
    */
   uint8_t field_type(const int32_t index, const int32_t field_category) const { return bcf_hdr_id2type(m_header.get(), field_category, index); }
-
+  /*
+   * returns one of BCF_VL_* values for field with the specified name and category (one of BCF_HL_FMT, BCF_HL_INFO, or BCF_HL_FLT)
+   *
+   * @note must check whether the field exists before calling this function, as it doesn't check for you
+   */
+  uint32_t field_length_descriptor(const std::string& tag, const int32_t field_category) const { return field_length_descriptor(field_index(tag), field_category); }
+  /**
+   * returns one of BCF_VL_* values for the field with the specified index and category (one of BCF_HL_FMT, BCF_HL_INFO, or BCF_HL_FLT)
+   *
+   * @note must check whether the field exists before calling this function, as it doesn't check for you
+   */
+  uint32_t field_length_descriptor(const int32_t index, const int32_t field_category) const { return bcf_hdr_id2length(m_header.get(), field_category, index); }
+  /**
+   * returns number of values for the field with the specified name and category (one of BCF_HL_FMT, BCF_HL_INFO, or BCF_HL_FLT), 0xfffff for variable length fields
+   *
+   * @note must check whether the field exists before calling this function, as it doesn't check for you
+   */
+  uint32_t field_length(const std::string& tag, const int32_t field_category) const { return field_length(field_index(tag), field_category); }
+  /**
+   * returns number of values for the field with the specified index and category (one of BCF_HL_FMT, BCF_HL_INFO, or BCF_HL_FLT), 0xfffff for variable length fields
+   *
+   * @note must check whether the field exists before calling this function, as it doesn't check for you
+   */
+  uint32_t field_length(const int32_t index, const int32_t field_category) const { return bcf_hdr_id2number(m_header.get(), field_category, index); }
   /**
    * @brief checks whether the given filter is present given the filter name
    */

diff --git a/test/variant_test.cpp b/test/variant_test.cpp
@@ -112,3 +112,19 @@ BOOST_AUTO_TEST_CASE( test_missing_string_individual_field_values ) {
   BOOST_CHECK(missing(variant.string_individual_field("AS")[1]));
   BOOST_CHECK(! missing(variant.string_individual_field("AS")[2]));
 }
+
+BOOST_AUTO_TEST_CASE( test_field_lengths ) {
+  auto header = SingleVariantReader{"testdata/test_variants.vcf"}.header();
+  BOOST_CHECK_EQUAL(header.field_length_descriptor("AF", BCF_HL_INFO), static_cast<uint32_t>(BCF_VL_A));
+  BOOST_CHECK_EQUAL(header.field_length_descriptor("AF", BCF_HL_FMT), static_cast<uint32_t>(BCF_VL_FIXED));
+  BOOST_CHECK_EQUAL(header.field_length_descriptor("PL", BCF_HL_FMT), static_cast<uint32_t>(BCF_VL_G));
+  BOOST_CHECK_EQUAL(header.field_length_descriptor("AS", BCF_HL_FMT), static_cast<uint32_t>(BCF_VL_FIXED));
+  BOOST_CHECK_EQUAL(header.field_length_descriptor("VLINT", BCF_HL_FMT), static_cast<uint32_t>(BCF_VL_VAR));
+
+  BOOST_CHECK_EQUAL(header.field_length("AF", BCF_HL_INFO), 0xfffffu);
+  BOOST_CHECK_EQUAL(header.field_length("AF", BCF_HL_FMT), 2u);
+  BOOST_CHECK_EQUAL(header.field_length("PL", BCF_HL_FMT), 0xfffffu);
+  BOOST_CHECK_EQUAL(header.field_length("AS", BCF_HL_FMT), 1u);
+  BOOST_CHECK_EQUAL(header.field_length("VLINT", BCF_HL_FMT), 0xfffffu);
+}
+