/// <summary>
/// Writes <see cref="DataSet"/> to a target file
/// </summary>
/// <param name="dataSet"><see cref="DataSet"/> to write</param>
/// <param name="fileName">Path to a file to write to.</param>
/// <param name="compression">Compression method</param>
/// <param name="formatOptions">Parquet options, optional.</param>
/// <param name="writerOptions">Writer options, optional.</param>
/// <param name="append">When true, assumes that this stream contains existing file and appends data to it, otherwise writes a new Parquet file.</param>
public static void WriteFile(
DataSet dataSet, string fileName, CompressionMethod compression = CompressionMethod.Gzip,
ParquetOptions formatOptions = null, WriterOptions writerOptions = null, bool append = false)
{
using (Stream fs = System.IO.File.Create(fileName))
{
using (var writer = new ParquetWriter(fs, formatOptions, writerOptions))
{
writer.Write(dataSet, compression);
}
}
}
public static Schema Serialize <T>(IEnumerable <T> objectInstances, Stream destination,
Schema schema = null,
WriterOptions writerOptions = null,
CompressionMethod compressionMethod = CompressionMethod.Snappy)
where T : new()
{
if (objectInstances == null)
{
throw new ArgumentNullException(nameof(objectInstances));
}
if (destination == null)
{
throw new ArgumentNullException(nameof(destination));
}
if (!destination.CanWrite)
{
throw new ArgumentException("stream must be writeable", nameof(destination));
}
//if schema is not passed reflect it
if (schema == null)
{
schema = SchemaReflector.Reflect <T>();
}
if (writerOptions == null)
{
writerOptions = new WriterOptions();
}
var extractor = new ColumnExtractor();
using (var writer = new ParquetWriter3(schema, destination, writerOptions: writerOptions))
{
writer.CompressionMethod = compressionMethod;
foreach (IEnumerable <T> batch in objectInstances.Batch(writerOptions.RowGroupsSize))
{
IReadOnlyCollection <DataColumn> columns = extractor.ExtractColumns(batch, schema);
using (ParquetRowGroupWriter groupWriter = writer.CreateRowGroup(batch.Count()))
{
foreach (DataColumn dataColumn in columns)
{
groupWriter.Write(dataColumn);
}
}
}
}
return(schema);
}
/// <summary>
/// Creates an instance of parquet writer on top of a stream
/// </summary>
/// <param name="output">Writeable, seekable stream</param>
/// <param name="formatOptions">Additional options</param>
/// <param name="writerOptions">The writer options.</param>
/// <exception cref="ArgumentNullException">Output is null.</exception>
/// <exception cref="ArgumentException">Output stream is not writeable</exception>
public ParquetWriter(Stream output, ParquetOptions formatOptions = null, WriterOptions writerOptions = null)
: base(new GapStream(output))
{
if (output == null)
{
throw new ArgumentNullException(nameof(output));
}
if (!output.CanWrite)
{
throw new ArgumentException("stream is not writeable", nameof(output));
}
_formatOptions = formatOptions ?? new ParquetOptions();
_writerOptions = writerOptions ?? new WriterOptions();
}
/// <summary>
/// Creates an instance of parquet writer on top of a stream
/// </summary>
/// <param name="output">Writeable, seekable stream</param>
/// <param name="formatOptions">Additional options</param>
/// <param name="writerOptions">The writer options.</param>
/// <exception cref="ArgumentNullException">Output is null.</exception>
/// <exception cref="ArgumentException">Output stream is not writeable</exception>
public ParquetWriter(Stream output, ParquetOptions formatOptions = null, WriterOptions writerOptions = null) : base(output)
{
_output = output ?? throw new ArgumentNullException(nameof(output));
if (!output.CanWrite)
{
throw new ArgumentException("stream is not writeable", nameof(output));
}
_formatOptions = formatOptions ?? new ParquetOptions();
_writerOptions = writerOptions ?? new WriterOptions();
_meta = new FileMetadataBuilder(_writerOptions);
_plainWriter = new PlainValuesWriter(_formatOptions);
_rleWriter = new RunLengthBitPackingHybridValuesWriter();
_dicWriter = new PlainDictionaryValuesWriter(_rleWriter);
}
/// <summary>
/// Creates an instance of parquet writer on top of a stream
/// </summary>
/// <param name="output">Writeable, seekable stream</param>
/// <param name="formatOptions">Additional options</param>
/// <param name="writerOptions">The writer options.</param>
/// <exception cref="ArgumentNullException">Output is null.</exception>
/// <exception cref="ArgumentException">Output stream is not writeable</exception>
public ParquetWriter(Stream output, ParquetOptions formatOptions = null, WriterOptions writerOptions = null)
: base(new PositionTrackingStream(output))
{
if (output == null)
{
throw new ArgumentNullException(nameof(output));
}
if (!output.CanWrite)
{
throw new ArgumentException("stream is not writeable", nameof(output));
}
_formatOptions = formatOptions ?? new ParquetOptions();
_writerOptions = writerOptions ?? new WriterOptions();
_meta = new FileMetadataBuilder();
}
/// <summary>
/// Creates an instance of parquet writer on top of a stream
/// </summary>
/// <param name="schema"></param>
/// <param name="output">Writeable, seekable stream</param>
/// <param name="formatOptions">Additional options</param>
/// <param name="writerOptions">The writer options.</param>
/// <param name="append"></param>
/// <exception cref="ArgumentNullException">Output is null.</exception>
/// <exception cref="ArgumentException">Output stream is not writeable</exception>
public ParquetWriter3(Schema schema, Stream output, ParquetOptions formatOptions = null, WriterOptions writerOptions = null, bool append = false)
: base(new GapStream(output))
{
if (output == null)
{
throw new ArgumentNullException(nameof(output));
}
if (!output.CanWrite)
{
throw new ArgumentException("stream is not writeable", nameof(output));
}
_schema = schema ?? throw new ArgumentNullException(nameof(schema));
_formatOptions = formatOptions ?? new ParquetOptions();
_writerOptions = writerOptions ?? new WriterOptions();
PrepareFile(append);
}
/// <summary>
/// Writes <see cref="DataSet"/> to a target stream
/// </summary>
/// <param name="dataSet"><see cref="DataSet"/> to write</param>
/// <param name="destination">Destination stream</param>
/// <param name="compression">Compression method</param>
/// <param name="formatOptions">Parquet options, optional.</param>
/// <param name="writerOptions">Writer options, optional.</param>
/// <param name="append">When true, assumes that this stream contains existing file and appends data to it, otherwise writes a new Parquet file.</param>
public static void Write(DataSet dataSet, Stream destination, CompressionMethod compression = CompressionMethod.Gzip, ParquetOptions formatOptions = null, WriterOptions writerOptions = null, bool append = false)
{
using (var writer = new ParquetWriter(destination, formatOptions, writerOptions))
{
writer.Write(dataSet, compression, append);
}
}
