Parquet.Reader

Documentation

newtype ParquetSource m

Constructors

ParquetSource (Integer -> ConduitT () ByteString m ())

type Url = String

readMetadata :: (MonadError Text m, MonadIO m) => ParquetSource m -> m FileMetadata

localParquetFile :: MonadResource m => FilePath -> ParquetSource m

remoteParquetFile :: (MonadResource m, MonadThrow m, MonadIO m) => Url -> ParquetSource m

readWholeParquetFile :: (MonadThrow m, MonadIO m, MonadError Text m, MonadResource m, MonadLogger m) => String -> m [ParquetObject]

type Record = [(ColumnValue, [Text])]

sourceParquet :: FilePath -> ConduitT () ParquetObject (ResourceT IO) ()

sourceRowGroupFromRemoteFile :: (MonadResource m, MonadIO m, MonadThrow m, MonadLogger m) => String -> FileMetadata -> RowGroup -> ConduitT () ParquetObject m ()

sourceRowGroup :: forall m. (MonadResource m, MonadIO m, MonadThrow m, MonadLogger m) => ParquetSource m -> FileMetadata -> RowGroup -> ConduitT () ParquetObject m ()

Streams the values for every column chunk and zips them into records.

Illustration:

_____________________ | col1 | col2 | col3 | | 1 | a | x | | 2 | b | y | | 3 | c | z | |____|__|____|

sourceRowGroup yields the following values in a stream:

(1, a, x) (2, b, y) (3, c, z)

sourceColumnChunk :: (MonadIO m, MonadResource m, MonadThrow m, MonadLogger m) => ParquetSource m -> FileMetadata -> ColumnChunk -> ConduitT () ColumnValue m ()