PARQUET-621: Add flag to indicate if decimalmetadata is set

Author: Uwe L. Korn <uwelk@xhochy.com>

Closes apache#109 from xhochy/parquet-621 and squashes the following commits:

3850a0f [Uwe L. Korn] Add unit test
dc1552f [Uwe L. Korn] Add flag to indicate if decimalmetadata is set

Change-Id: I7cb412aa8165af4cae1b1aca7360a01129cfe456

Author: xhochy

cpp/src/parquet/schema/schema-converter-test.cc

@@ -158,6 +158,27 @@ class TestSchemaFlatten : public ::testing::Test {
   std::vector<format::SchemaElement> elements_;
 };
 
+TEST_F(TestSchemaFlatten, DecimalMetadata) {
+  // Checks that DecimalMetadata is only set for DecimalTypes
+  NodePtr node = PrimitiveNode::Make(
+      "decimal", Repetition::REQUIRED, Type::INT64, LogicalType::DECIMAL, -1, 8, 4);
+  NodePtr group =
+      GroupNode::Make("group", Repetition::REPEATED, {node}, LogicalType::LIST);
+  Flatten(reinterpret_cast<GroupNode*>(group.get()));
+  ASSERT_EQ("decimal", elements_[1].name);
+  ASSERT_TRUE(elements_[1].__isset.precision);
+  ASSERT_TRUE(elements_[1].__isset.scale);
+
+  elements_.clear();
+  // Not for integers with no logical type
+  group =
+      GroupNode::Make("group", Repetition::REPEATED, {Int64("int64")}, LogicalType::LIST);
+  Flatten(reinterpret_cast<GroupNode*>(group.get()));
+  ASSERT_EQ("int64", elements_[1].name);
+  ASSERT_FALSE(elements_[0].__isset.precision);
+  ASSERT_FALSE(elements_[0].__isset.scale);
+}
+
 TEST_F(TestSchemaFlatten, NestedExample) {
   SchemaElement elt;
   std::vector<SchemaElement> elements;

cpp/src/parquet/schema/types.cc

@@ -86,12 +86,14 @@ PrimitiveNode::PrimitiveNode(const std::string& name, Repetition::type repetitio
       physical_type_(type),
       type_length_(length) {
   std::stringstream ss;
+  decimal_metadata_.isset = false;
   // Check if the physical and logical types match
   // Mapping referred from Apache parquet-mr as on 2016-02-22
   switch (logical_type) {
     case LogicalType::NONE:
       // Logical type not set
       // Clients should be able to read these values
+      decimal_metadata_.isset = true;
       decimal_metadata_.precision = precision;
       decimal_metadata_.scale = scale;
       break;
@@ -123,6 +125,7 @@ PrimitiveNode::PrimitiveNode(const std::string& name, Repetition::type repetitio
         ss << " cannot be greater than precision " << precision;
         throw ParquetException(ss.str());
       }
+      decimal_metadata_.isset = true;
       decimal_metadata_.precision = precision;
       decimal_metadata_.scale = scale;
       break;
@@ -299,8 +302,10 @@ void PrimitiveNode::ToParquet(void* opaque_element) const {
   element->__set_type(ToThrift(physical_type_));
   // FIXME: SchemaFlattener does this for us: element->__set_field_id(id_);
   element->__set_type_length(type_length_);
-  element->__set_precision(decimal_metadata_.precision);
-  element->__set_scale(decimal_metadata_.scale);
+  if (decimal_metadata_.isset) {
+    element->__set_precision(decimal_metadata_.precision);
+    element->__set_scale(decimal_metadata_.scale);
+  }
 }
 
 }  // namespace schema

cpp/src/parquet/schema/types.h

@@ -70,6 +70,7 @@ struct ListEncoding {
 };
 
 struct DecimalMetadata {
+  bool isset;
   int32_t scale;
   int32_t precision;
 };