Expand description
Local filesystem source + destination, with pluggable file-format codecs
(Parquet now; Csv/Avro later). Built on the arrow-rs parquet crate.
Structs§
- Files
Destination - Local files destination. Writes
part-NNNNN.{extension}files to a directory, or one{dir}.{extension}whensingle_file. Write is atomic via tmp dir + rename. Written paths land inRunReport.written_objects. - Files
Source - Local file source. One or many files, decoded by the supplied
FormatRead. - Parquet
- Parquet file format. Carries encoder knobs; decoding needs none.
Enums§
- Compression
- Compression codec for Parquet column chunks. Default =
Zstd. - Glob
OrPaths - How the source enumerates files: glob or single path, or list of paths.
Traits§
- File
Reader - A readable file handle trait marker for random-access bytes.
- File
Writer - A writable file handle trait.
- Format
Read - Decodes a file’s bytes into Arrow batches.
- Format
Write - Encodes Arrow batches into a file’s bytes.