public void TestConvertToDelta()
{
string partitionColumnName = "id_plus_one";
DataFrame data = _spark.Range(0, 5).Select(
Functions.Col("id"),
Functions.Expr($"(`id` + 1) AS `{partitionColumnName}`"));
// Run the same test on the different overloads of DeltaTable.ConvertToDelta().
void testWrapper(
DataFrame dataFrame,
Func <string, DeltaTable> convertToDelta,
string partitionColumn = null)
{
using var tempDirectory = new TemporaryDirectory();
string path = Path.Combine(tempDirectory.Path, "parquet-data");
DataFrameWriter dataWriter = dataFrame.Write();
if (!string.IsNullOrEmpty(partitionColumn))
{
dataWriter = dataWriter.PartitionBy(partitionColumn);
}
dataWriter.Parquet(path);
Assert.False(DeltaTable.IsDeltaTable(path));
string identifier = $"parquet.`{path}`";
DeltaTable convertedDeltaTable = convertToDelta(identifier);
ValidateRangeDataFrame(Enumerable.Range(0, 5), convertedDeltaTable.ToDF());
Assert.True(DeltaTable.IsDeltaTable(path));
}
testWrapper(data, identifier => DeltaTable.ConvertToDelta(_spark, identifier));
testWrapper(
data.Repartition(Functions.Col(partitionColumnName)),
identifier => DeltaTable.ConvertToDelta(
_spark,
identifier,
$"{partitionColumnName} bigint"),
partitionColumnName);
testWrapper(
data.Repartition(Functions.Col(partitionColumnName)),
identifier => DeltaTable.ConvertToDelta(
_spark,
identifier,
new StructType(new[]
{
new StructField(partitionColumnName, new IntegerType())
})),
partitionColumnName);
}
public void TestPartitionBy()
{
// arrange
mockDataFrameWriterProxy.Setup(m => m.PartitionBy(It.IsAny <string[]>()));
var dataFrameWriter = new DataFrameWriter(mockDataFrameWriterProxy.Object);
var colNames = new string[] { "col1", "col2", "col3" };
// Act
dataFrameWriter.PartitionBy(colNames);
// Assert
mockDataFrameWriterProxy.Verify(m => m.PartitionBy(colNames));
}
public void TestSignaturesV2_3_X()
{
{
DataFrameWriter dfw = _spark
.Read()
.Schema("age INT, name STRING")
.Json($"{TestEnvironment.ResourceDirectory}people.json")
.Write();
Assert.IsType <DataFrameWriter>(dfw.Mode(SaveMode.Ignore));
Assert.IsType <DataFrameWriter>(dfw.Mode("overwrite"));
Assert.IsType <DataFrameWriter>(dfw.Format("json"));
Assert.IsType <DataFrameWriter>(dfw.Option("stringOption", "value"));
Assert.IsType <DataFrameWriter>(dfw.Option("boolOption", true));
Assert.IsType <DataFrameWriter>(dfw.Option("longOption", 1L));
Assert.IsType <DataFrameWriter>(dfw.Option("doubleOption", 3D));
Assert.IsType <DataFrameWriter>(
dfw.Options(
new Dictionary <string, string>
{
{ "option1", "value1" },
{ "option2", "value2" }
}));
Assert.IsType <DataFrameWriter>(dfw.PartitionBy("age"));
Assert.IsType <DataFrameWriter>(dfw.PartitionBy("age", "name"));
Assert.IsType <DataFrameWriter>(dfw.BucketBy(3, "age"));
Assert.IsType <DataFrameWriter>(dfw.BucketBy(3, "age", "name"));
Assert.IsType <DataFrameWriter>(dfw.SortBy("name"));
}
using (var tempDir = new TemporaryDirectory())
{
DataFrameWriter dfw = _spark
.Read()
.Csv($"{TestEnvironment.ResourceDirectory}people.csv")
.Write();
// TODO: Test dfw.Jdbc without running a local db.
dfw.Option("path", tempDir.Path).SaveAsTable("TestTable");
dfw.InsertInto("TestTable");
dfw.Option("path", $"{tempDir.Path}TestSavePath1").Save();
dfw.Save($"{tempDir.Path}TestSavePath2");
dfw.Json($"{tempDir.Path}TestJsonPath");
dfw.Parquet($"{tempDir.Path}TestParquetPath");
dfw.Orc($"{tempDir.Path}TestOrcPath");
dfw.Text($"{tempDir.Path}TestTextPath");
dfw.Csv($"{tempDir.Path}TestCsvPath");
}
}
public void TestSignaturesV2_3_X()
{
{
DataFrameWriter dfw = _spark
.Read()
.Schema("age INT, name STRING")
.Json($"{TestEnvironment.ResourceDirectory}people.json")
.Write();
Assert.IsType <DataFrameWriter>(dfw.Mode(SaveMode.Ignore));
Assert.IsType <DataFrameWriter>(dfw.Mode("overwrite"));
Assert.IsType <DataFrameWriter>(dfw.Format("json"));
Assert.IsType <DataFrameWriter>(dfw.Option("stringOption", "value"));
Assert.IsType <DataFrameWriter>(dfw.Option("boolOption", true));
Assert.IsType <DataFrameWriter>(dfw.Option("longOption", 1L));
Assert.IsType <DataFrameWriter>(dfw.Option("doubleOption", 3D));
Assert.IsType <DataFrameWriter>(
dfw.Options(
new Dictionary <string, string>
{
{ "option1", "value1" },
{ "option2", "value2" }
}));
Assert.IsType <DataFrameWriter>(dfw.PartitionBy("age"));
Assert.IsType <DataFrameWriter>(dfw.PartitionBy("age", "name"));
Assert.IsType <DataFrameWriter>(dfw.BucketBy(3, "age"));
Assert.IsType <DataFrameWriter>(dfw.BucketBy(3, "age", "name"));
Assert.IsType <DataFrameWriter>(dfw.SortBy("name"));
}
using (var tempDir = new TemporaryDirectory())
{
DataFrameWriter dfw = _spark
.Read()
.Csv($"{TestEnvironment.ResourceDirectory}people.csv")
.Write();
// TODO: Test dfw.Jdbc without running a local db.
dfw.Save($"{tempDir.Path}TestSavePath1");
dfw.Json($"{tempDir.Path}TestJsonPath");
dfw.Parquet($"{tempDir.Path}TestParquetPath");
dfw.Orc($"{tempDir.Path}TestOrcPath");
dfw.Text($"{tempDir.Path}TestTextPath");
dfw.Csv($"{tempDir.Path}TestCsvPath");
dfw.Option("path", tempDir.Path).SaveAsTable("TestTable");
dfw.InsertInto("TestTable");
// In Spark 3.1.1+ setting the `path` Option and then calling .Save(path) is not
// supported unless `spark.sql.legacy.pathOptionBehavior.enabled` conf is set.
// .Json(path), .Parquet(path), etc follow the same code path so the conf
// needs to be set in these scenarios as well.
dfw.Option("path", $"{tempDir.Path}TestSavePath2").Save();
}
}