writing avro pcollection to hdfs in parquet format with custom naming strategy

ParquetFileOutput.java

genericRecordPCollection.apply(
      FileIO
          .<GenericRecord>write()
          .via(ParquetIO.sink(schema))
          .to(pipelineOptions.getOutputFilePath())
          .withNumShards(pipelineOptions.getOutputFileShardCount())
          .withNaming((window, pane, numShards, shardIndex, compression) -> String.format(
                  "%s_%d_%d.%s", outputFilePrefix, startTime, endTime, "parq"))
);