gfw.common.beam.transforms.ParquetSink#
- class ParquetSink(schema, codec='snappy')[source]#
A
FileSinkthat buffers rows and flushes as one row group.- Parameters:
schema (pa.Schema) – PyArrow schema for the Parquet file.
codec (str) – Parquet compression codec (e.g.
"snappy","zstd").
Methods
create_metadataWrite all buffered rows as a single row group and close the writer.
Open the Parquet writer and initialise the row buffer.
Buffer one keyed row for writing.