PARQUET-968 Implement feedback

Update the proto to parquet schema converter for MAP fields so that it follows the scec: https://github.com/apache/parquet-format/blob/master/LogicalTypes.md#lists This came as feedback from the Amazon Athena team.
apache · Aug 27, 2017 · 8e6f8d1 · 8e6f8d1
1 parent 121c0b7
commit 8e6f8d1
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 10 deletions.
diff --git a/parquet-protobuf/src/main/java/org/apache/parquet/proto/ProtoMessageConverter.java b/parquet-protobuf/src/main/java/org/apache/parquet/proto/ProtoMessageConverter.java
@@ -447,8 +447,8 @@ public MapConverter(Message.Builder parentBuilder, Descriptors.FieldDescriptor f
       }
 
       Type parquetSchema;
-      if (parquetType.asGroupType().containsField("map")){
-        parquetSchema = parquetType.asGroupType().getType("map");
+      if (parquetType.asGroupType().containsField("key_value")){
+        parquetSchema = parquetType.asGroupType().getType("key_value");
       } else {
         throw new ParquetDecodingException("Expected map but got: " + parquetType);
       }

diff --git a/parquet-protobuf/src/main/java/org/apache/parquet/proto/ProtoSchemaConverter.java b/parquet-protobuf/src/main/java/org/apache/parquet/proto/ProtoSchemaConverter.java
@@ -137,12 +137,12 @@ private <T> GroupBuilder<GroupBuilder<T>> addMapField(Descriptors.FieldDescripto
     ParquetType mapKeyParquetType = getParquetType(fields.get(0));
 
     GroupBuilder<GroupBuilder<GroupBuilder<T>>> group = builder
-      .group(getRepetition(descriptor)).as(OriginalType.MAP)
-      .group(Type.Repetition.REPEATED).as(OriginalType.MAP_KEY_VALUE)
+      .group(Type.Repetition.REQUIRED).as(OriginalType.MAP)
+      .group(Type.Repetition.REPEATED) // key_value wrapper
       .primitive(mapKeyParquetType.primitiveType, Type.Repetition.REQUIRED).as(mapKeyParquetType.originalType).named("key");
 
     return addField(fields.get(1), group).named("value")
-      .named("map");
+      .named("key_value");
   }
 
   private ParquetType getParquetType(Descriptors.FieldDescriptor fieldDescriptor) {

diff --git a/parquet-protobuf/src/main/java/org/apache/parquet/proto/ProtoWriteSupport.java b/parquet-protobuf/src/main/java/org/apache/parquet/proto/ProtoWriteSupport.java
@@ -207,7 +207,7 @@ private GroupType getGroupType(Type type) {
       }
 
       if (type.getOriginalType() == OriginalType.MAP) {
-        return type.asGroupType().getType("map").asGroupType().getType("value").asGroupType();
+        return type.asGroupType().getType("key_value").asGroupType().getType("value").asGroupType();
       }
 
       return type.asGroupType();
@@ -369,15 +369,15 @@ public MapWriter(FieldWriter keyWriter, FieldWriter valueWriter) {
     final void writeRawValue(Object value) {
       recordConsumer.startGroup();
 
-      recordConsumer.startField("map", 0); // This is the wrapper group for the map field
+      recordConsumer.startField("key_value", 0); // This is the wrapper group for the map field
       for(MapEntry<?, ?> entry : (Collection<MapEntry<?, ?>>) value) {
         recordConsumer.startGroup();
         keyWriter.writeField(entry.getKey());
         valueWriter.writeField(entry.getValue());
         recordConsumer.endGroup();
       }
 
-      recordConsumer.endField("map", 0);
+      recordConsumer.endField("key_value", 0);
 
       recordConsumer.endGroup();
     }

diff --git a/parquet-protobuf/src/test/java/org/apache/parquet/proto/ProtoSchemaConverterTest.java b/parquet-protobuf/src/test/java/org/apache/parquet/proto/ProtoSchemaConverterTest.java
@@ -103,8 +103,8 @@ public void testProto3ConvertAllDatatypes() throws Exception {
         "  optional binary optionalEnum (ENUM) = 18;" +
         "  optional int32 someInt32 = 19;" +
         "  optional binary someString (UTF8) = 20;" +
-        "  repeated group optionalMap (MAP) = 21 {\n" +
-        "    repeated group map (MAP_KEY_VALUE) {\n" +
+        "  required group optionalMap (MAP) = 21 {\n" +
+        "    repeated group key_value {\n" +
         "      required int64 key;\n" +
         "      optional group value {\n" +
         "        optional int32 someId = 3;\n" +