[HIVE-21492] VectorizedParquetRecordReader can't to read parquet file generated using thrift/custom tool

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.0.0-alpha-1
• Component/s: Parquet
• Assignee: Ganesha Shreedhara
• Reporter: Ganesha Shreedhara
• Created: 22/Mar/19 15:59
• Updated: 17/Nov/22 08:51
• Resolved: 07/Apr/20 15:26

Description

Taking an example of a parquet table having array of integers as below. 

CREATE EXTERNAL TABLE ( list_of_ints` array<int>)
STORED AS PARQUET 
LOCATION '{location}';

Parquet file generated using hive will have schema for Type as below:

group list_of_ints (LIST) { repeated group bag { optional int32 array;\n};\n} 

Parquet file generated using thrift or any custom tool (using org.apache.parquet.io.api.RecordConsumer)

may have schema for Type as below:

required group list_of_ints (LIST) { repeated int32 list_of_tuple} 

VectorizedParquetRecordReader handles only parquet file generated using hive. It throws the following exception when parquet file generated using thrift is read because of the changes done as part of HIVE-18553 .

Caused by: java.lang.ClassCastException: repeated int32 list_of_ints_tuple is not a group
 at org.apache.parquet.schema.Type.asGroupType(Type.java:207)
 at org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader.getElementType(VectorizedParquetRecordReader.java:479)
 at org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader.buildVectorizedParquetReader(VectorizedParquetRecordReader.java:532)
 at org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader.checkEndOfRowGroup(VectorizedParquetRecordReader.java:440)
 at org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader.nextBatch(VectorizedParquetRecordReader.java:401)
 at org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader.next(VectorizedParquetRecordReader.java:353)
 at org.apache.hadoop.hive.ql.io.parquet.vector.VectorizedParquetRecordReader.next(VectorizedParquetRecordReader.java:92)
 at org.apache.hadoop.hive.ql.io.HiveContextAwareRecordReader.doNext(HiveContextAwareRecordReader.java:365)

 

 I have done a small change to handle the case where the child type of group type can be PrimitiveType.