Get an SCollection for a Parquet file as Avro records.
Get an SCollection for a Parquet file as Avro records. Since Avro records produced by Parquet column projection may be incomplete and may fail serialization, you must map the result to extract projected fields from the Avro records.
Note that due to limitations of the underlying HadoopInputFormatIO
,
Avro GenericRecord and dynamic work rebalancing
are not supported. Without the latter, pipelines may not autoscale up or down during the
initial read and subsequent fused transforms.
Enhanced version of ScioContext with Parquet Avro methods.