public static void TestPropertiesBuilder()
{
var p = new WriterPropertiesBuilder()
.Compression(Compression.Snappy)
.CompressionLevel(3)
.CreatedBy("Meeeee!!!")
.DataPagesize(123)
.DictionaryPagesizeLimit(456)
.Encoding(Encoding.DeltaByteArray)
.MaxRowGroupLength(789)
.Version(ParquetVersion.PARQUET_1_0)
.WriteBatchSize(666)
.Build();
Assert.AreEqual("Meeeee!!!", p.CreatedBy);
Assert.AreEqual(Compression.Snappy, p.Compression(new ColumnPath("anypath")));
Assert.AreEqual(3, p.CompressionLevel(new ColumnPath("anypath")));
Assert.AreEqual(123, p.DataPageSize);
Assert.AreEqual(Encoding.PlainDictionary, p.DictionaryIndexEncoding);
Assert.AreEqual(Encoding.PlainDictionary, p.DictionaryPageEncoding);
Assert.AreEqual(456, p.DictionaryPagesizeLimit);
Assert.AreEqual(789, p.MaxRowGroupLength);
Assert.AreEqual(ParquetVersion.PARQUET_1_0, p.Version);
Assert.AreEqual(666, p.WriteBatchSize);
}
private static WriterProperties CreateWriterProperties()
{
var builder = new WriterPropertiesBuilder();
builder.Compression(Compression.Snappy);
return(builder.Build());
}
private static WriterProperties CreateWriterProperties(FileEncryptionProperties fileEncryptionProperties)
{
using var builder = new WriterPropertiesBuilder();
return(builder
.Compression(Compression.Lz4)
.Encryption(fileEncryptionProperties)
.Build());
}
public static void TestPropertiesBuilder()
{
var builder = new WriterPropertiesBuilder();
builder
.Compression(Compression.Snappy)
.CreatedBy("Meeeee!!!")
.DataPagesize(123)
.DictionaryPagesizeLimit(456)
.Encoding(Encoding.DeltaByteArray)
.MaxRowGroupLength(789)
.Version(ParquetVersion.PARQUET_1_0)
.WriteBatchSize(666)
;
var p = builder.Build();
Assert.AreEqual("Meeeee!!!", p.CreatedBy);
Assert.AreEqual(123, p.DataPageSize);
Assert.AreEqual(Encoding.PlainDictionary, p.DictionaryIndexEncoding);
Assert.AreEqual(Encoding.PlainDictionary, p.DictionaryPageEncoding);
Assert.AreEqual(456, p.DictionaryPagesizeLimit);
Assert.AreEqual(789, p.MaxRowGroupLength);
Assert.AreEqual(ParquetVersion.PARQUET_1_0, p.Version);
Assert.AreEqual(666, p.WriteBatchSize);
/*
* public WriterPropertiesBuilder DisableDictionary()
* public WriterPropertiesBuilder DisableDictionary(string path)
* public WriterPropertiesBuilder EnableDictionary()
* public WriterPropertiesBuilder EnableDictionary(string path)
*
* // Statistics enable/disable
*
* public WriterPropertiesBuilder DisableStatistics()
* public WriterPropertiesBuilder DisableStatistics(string path)
* public WriterPropertiesBuilder EnableStatistics()
* public WriterPropertiesBuilder EnableStatistics(string path)
*
* // Other properties
*
* public WriterPropertiesBuilder Compression(Compression codec)
* public WriterPropertiesBuilder Compression(string path, Compression codec)
* public WriterPropertiesBuilder CreatedBy(string createdBy)
* public WriterPropertiesBuilder DataPagesize(long pageSize)
* public WriterPropertiesBuilder DictionaryPagesizeLimit(long dictionaryPagesizeLimit)
* public WriterPropertiesBuilder Encoding(Encoding encoding)
* public WriterPropertiesBuilder Encoding(string path, Encoding encoding)
* public WriterPropertiesBuilder MaxRowGroupLength(long maxRowGroupLength)
* public WriterPropertiesBuilder Version(ParquetVersion version)
* public WriterPropertiesBuilder WriteBatchSize(long writeBatchSize)
*/
}
private static WriterProperties CreateWriterProperties(ExpectedColumn[] expectedColumns, bool useDictionaryEncoding)
{
var builder = new WriterPropertiesBuilder();
builder.Compression(Compression.Lz4);
if (!useDictionaryEncoding)
{
foreach (var column in expectedColumns)
{
builder.DisableDictionary(column.Name);
}
}
return(builder.Build());
}
private static WriterProperties CreateWriterProperties(bool enableDictionary)
{
using var builder = new WriterPropertiesBuilder();
builder.Compression(Compression.Snappy);
return((enableDictionary ? builder : builder.DisableDictionary("value")).Build());
}
public static void TestBigFileBufferedRowGroup()
{
// Test a large amount of rows with a buffered row group to uncover any particular issue.
const int numBatches = 64;
const int batchSize = 8192;
using var buffer = new ResizableBuffer();
using (var output = new BufferOutputStream(buffer))
{
var columns = new Column[]
{
new Column <int>("int"),
new Column <double>("double"),
new Column <string>("string"),
new Column <bool>("bool")
};
using var builder = new WriterPropertiesBuilder();
using var writerProperties = builder.Compression(Compression.Snappy).DisableDictionary("double").Build();
using var fileWriter = new ParquetFileWriter(output, columns, writerProperties);
using var rowGroupWriter = fileWriter.AppendBufferedRowGroup();
using var col0 = rowGroupWriter.Column(0).LogicalWriter <int>();
using var col1 = rowGroupWriter.Column(1).LogicalWriter <double>();
using var col2 = rowGroupWriter.Column(2).LogicalWriter <string>();
using var col3 = rowGroupWriter.Column(3).LogicalWriter <bool>();
for (var batchIndex = 0; batchIndex < numBatches; ++batchIndex)
{
var startIndex = batchSize * batchIndex;
col0.WriteBatch(Enumerable.Range(startIndex, batchSize).ToArray());
col1.WriteBatch(Enumerable.Range(startIndex, batchSize).Select(i => (double)i).ToArray());
col2.WriteBatch(Enumerable.Range(startIndex, batchSize).Select(i => i.ToString()).ToArray());
col3.WriteBatch(Enumerable.Range(startIndex, batchSize).Select(i => i % 2 == 0).ToArray());
}
fileWriter.Close();
}
using (var input = new BufferReader(buffer))
{
using var fileReader = new ParquetFileReader(input);
using var rowGroupReader = fileReader.RowGroup(0);
using var col0 = rowGroupReader.Column(0).LogicalReader <int>();
using var col1 = rowGroupReader.Column(1).LogicalReader <double>();
using var col2 = rowGroupReader.Column(2).LogicalReader <string>();
using var col3 = rowGroupReader.Column(3).LogicalReader <bool>();
for (var batchIndex = 0; batchIndex < numBatches; ++batchIndex)
{
var startIndex = batchSize * batchIndex;
Assert.AreEqual(Enumerable.Range(startIndex, batchSize).ToArray(), col0.ReadAll(batchSize));
Assert.AreEqual(Enumerable.Range(startIndex, batchSize).Select(i => (double)i).ToArray(), col1.ReadAll(batchSize));
Assert.AreEqual(Enumerable.Range(startIndex, batchSize).Select(i => i.ToString()).ToArray(), col2.ReadAll(batchSize));
Assert.AreEqual(Enumerable.Range(startIndex, batchSize).Select(i => i % 2 == 0).ToArray(), col3.ReadAll(batchSize));
}
fileReader.Close();
}
}
public static void TestProperties()
{
// Test the various properties exposed by ParquetFileWriter.
using var writerPropertiesBuilder = new WriterPropertiesBuilder();
using var writerProperties = writerPropertiesBuilder
.Compression(Compression.Zstd)
.DisableDictionary()
.CreatedBy("Some crazy unit test")
.Build();
var columns = new Column[]
{
new Column <int>("Index"),
new Column <float>("Value")
};
var kvm = (IReadOnlyDictionary <string, string>) new Dictionary <string, string>
{
{ "some key", "some value" },
{ "α ∧ ¬β", "2H₂ + O₂ ⇌ 2H₂O, R = 4.7 kΩ, ⌀ 200 mm" }
};
using var buffer = new ResizableBuffer();
using var outStream = new BufferOutputStream(buffer);
using var fileWriter = new ParquetFileWriter(outStream, columns, writerProperties, kvm);
Assert.AreEqual(2, fileWriter.NumColumns);
Assert.AreEqual(0, fileWriter.NumRows);
Assert.AreEqual(0, fileWriter.NumRowGroups);
Assert.IsNull(fileWriter.FileMetaData);
Assert.AreEqual(Column.CreateSchemaNode(columns), fileWriter.Schema.GroupNode);
Assert.AreEqual(Column.CreateSchemaNode(columns), fileWriter.Schema.SchemaRoot);
Assert.AreEqual(columns[0].Name, fileWriter.ColumnDescriptor(0).Name);
Assert.AreEqual(columns[1].Name, fileWriter.ColumnDescriptor(1).Name);
Assert.AreEqual(kvm, fileWriter.KeyValueMetadata);
Assert.AreEqual(Compression.Zstd, fileWriter.WriterProperties.Compression(new ColumnPath("")));
Assert.AreEqual(false, fileWriter.WriterProperties.DictionaryEnabled(new ColumnPath("")));
Assert.AreEqual("Some crazy unit test", fileWriter.WriterProperties.CreatedBy);
using (var groupWriter = fileWriter.AppendRowGroup())
{
Assert.AreEqual(0, fileWriter.NumRows);
Assert.AreEqual(1, fileWriter.NumRowGroups);
Assert.IsNull(fileWriter.FileMetaData);
using (var writer = groupWriter.NextColumn().LogicalWriter <int>())
{
writer.WriteBatch(new[] { 1, 2, 3, 4, 5, 6 });
}
Assert.AreEqual(0, fileWriter.NumRows);
Assert.AreEqual(1, fileWriter.NumRowGroups);
Assert.IsNull(fileWriter.FileMetaData);
using (var writer = groupWriter.NextColumn().LogicalWriter <float>())
{
writer.WriteBatch(new[] { 1f, 2f, 3f, 4f, 5f, 6f });
}
Assert.AreEqual(0, fileWriter.NumRows);
Assert.AreEqual(1, fileWriter.NumRowGroups);
Assert.IsNull(fileWriter.FileMetaData);
}
Assert.AreEqual(0, fileWriter.NumRows);
Assert.AreEqual(1, fileWriter.NumRowGroups);
Assert.IsNull(fileWriter.FileMetaData);
fileWriter.Close();
//Assert.AreEqual(0, fileWriter.NumRows); // 2021-04-08: calling this results in a segfault when the writer has been closed
//Assert.AreEqual(1, fileWriter.NumRowGroups); // 2021-04-08: calling this results in a segfault when the writer has been closed
Assert.IsNotNull(fileWriter.FileMetaData);
Assert.AreEqual(2, fileWriter.FileMetaData?.NumColumns);
Assert.AreEqual(6, fileWriter.FileMetaData?.NumRows);
Assert.AreEqual(1, fileWriter.FileMetaData?.NumRowGroups);
Assert.AreEqual(kvm, fileWriter.FileMetaData?.KeyValueMetadata);
}