public async Task RunsPivotStep()
{
using (var writer = new TestBulkWriter <PipelineTestsMyTestClass>())
{
var idCounter = 0;
var items = Enumerable.Range(1, 10).ToList();
var pipeline = EtlPipeline.StartWith(items)
.Pivot(i =>
{
var result = new List <PipelineTestsMyTestClass>();
for (var j = 0; j < i; j++)
{
++idCounter;
result.Add(new PipelineTestsMyTestClass {
Id = idCounter, Name = $"Bob {idCounter}"
});
}
return(result);
})
.WriteTo(writer);
await pipeline.ExecuteAsync();
var expectedSum = items.Sum();
Assert.Equal(expectedSum, writer.ItemsWritten.Count);
Assert.Equal(1, writer.ItemsWritten[0].Id);
Assert.Equal("Bob 1", writer.ItemsWritten[0].Name);
Assert.Equal(expectedSum, writer.ItemsWritten[expectedSum - 1].Id);
Assert.Equal($"Bob {expectedSum}", writer.ItemsWritten[expectedSum - 1].Name);
}
}
public async Task RunsTransformerStep()
{
using (var writer = new TestBulkWriter <PipelineTestsMyTestClass>())
{
var items = Enumerable.Range(1, 1000).Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
});
var pipeline = EtlPipeline
.StartWith(items)
.TransformInPlace(i =>
{
i.Id -= 1;
i.Name = $"Alice {i.Id}";
})
.WriteTo(writer);
await pipeline.ExecuteAsync();
Assert.Equal(1000, writer.ItemsWritten.Count);
Assert.Equal(0, writer.ItemsWritten[0].Id);
Assert.Equal("Alice 0", writer.ItemsWritten[0].Name);
Assert.Equal(999, writer.ItemsWritten[999].Id);
Assert.Equal("Alice 999", writer.ItemsWritten[999].Name);
}
}
public async Task LogsExceptionInPipelineStep()
{
using (var writer = new TestBulkWriter <PipelineTestsMyTestClass>())
{
var items = Enumerable
.Range(1, 1000)
.Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
})
.ToAsyncEnumerable();
var loggerFactory = new FakeLoggerFactory();
var pipeline = EtlPipeline
.StartWith(items)
.Project <PipelineTestsMyTestClass>(i => throw new Exception("This is my exception"))
.LogWith(loggerFactory)
.WriteTo(writer);
var pipelineTask = pipeline.ExecuteAsync();
await Assert.ThrowsAsync <Exception>(() => pipelineTask);
var errorMessages = loggerFactory.LoggedMessages.Where(m => m.LogLevel == LogLevel.Error).ToList();
Assert.Single(errorMessages);
Assert.Contains("Error", errorMessages[0].Message);
Assert.Equal("This is my exception", errorMessages[0].Exception.Message);
}
}
public async Task IgnoresNullLogger()
{
using (var writer = new TestBulkWriter <PipelineTestsMyTestClass>())
{
var items = Enumerable
.Range(1, 1000)
.Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
})
.ToAsyncEnumerable();
var pipeline = EtlPipeline
.StartWith(items)
.TransformInPlace(i =>
{
i.Id -= 1;
i.Name = $"Alice {i.Id}";
})
.LogWith(null)
.WriteTo(writer);
await pipeline.ExecuteAsync();
Assert.Equal(1000, writer.ItemsWritten.Count);
}
}
public async Task RunsProjectorStep()
{
using (var writer = new TestBulkWriter <PipelineTestsMyTestClass>())
{
var items = Enumerable
.Range(1, 1000)
.Select(i => new PipelineTestsOtherTestClass {
Id = i, FirstName = "Bob", LastName = $"{i}"
})
.ToAsyncEnumerable();
var pipeline = EtlPipeline
.StartWith(items)
.Project(i => new PipelineTestsMyTestClass {
Id = i.Id, Name = $"{i.FirstName} {i.LastName}"
})
.WriteTo(writer);
await pipeline.ExecuteAsync();
Assert.Equal(1000, writer.ItemsWritten.Count);
Assert.Equal(1, writer.ItemsWritten[0].Id);
Assert.Equal("Bob 1", writer.ItemsWritten[0].Name);
Assert.Equal(1000, writer.ItemsWritten[999].Id);
Assert.Equal("Bob 1000", writer.ItemsWritten[999].Name);
}
}
public async Task LogsStartAndFinishOfPipelineSteps()
{
using (var writer = new TestBulkWriter <PipelineTestsOtherTestClass>())
{
var items = Enumerable.Range(1, 1000).Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
});
var loggerFactory = new FakeLoggerFactory();
var pipeline = EtlPipeline
.StartWith(items)
.LogWith(loggerFactory)
.Aggregate(f =>
{
Thread.Sleep(1);
return(f.Max(c => c.Id));
})
.Pivot(i =>
{
var result = new List <PipelineTestsMyTestClass>();
for (var j = 1; j <= i; j++)
{
result.Add(new PipelineTestsMyTestClass {
Id = j, Name = $"Bob {j}"
});
}
return(result);
})
.Project(i =>
{
var nameParts = i.Name.Split(' ');
return(new PipelineTestsOtherTestClass {
Id = i.Id, FirstName = nameParts[0], LastName = nameParts[1]
});
})
.TransformInPlace(i =>
{
i.Id -= 1;
i.FirstName = "Alice";
i.LastName = $"{i.Id}";
})
.WriteTo(writer);
await pipeline.ExecuteAsync();
var totalStepsExpected = 6;
Assert.True(2 * totalStepsExpected == loggerFactory.LoggedMessages.Count, string.Join("\r\n", loggerFactory.LoggedMessages.Select(m => m.Message))); //2 log messages for each step in the pipeline
for (var i = 1; i <= totalStepsExpected; i++)
{
var messagesForStep = loggerFactory.LoggedMessages.Where(m => m.Message.Contains($"step {i} of {totalStepsExpected}")).ToList();
Assert.True(messagesForStep.Count == 2, $"Found {messagesForStep.Count} messages for step {i}");
Assert.Equal(1, messagesForStep.Count(m => m.Message.Contains("Starting")));
Assert.Equal(1, messagesForStep.Count(m => m.Message.Contains("Completing")));
}
}
}
public async Task RunsAllStepsInMultiStagePipeline()
{
using (var writer = new TestBulkWriter <PipelineTestsOtherTestClass>())
{
var items = Enumerable
.Range(1, 1000)
.Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
})
.ToAsyncEnumerable();
var pipeline = EtlPipeline
.StartWith(items)
.Aggregate(f => f.Max(c => c.Id))
.Pivot(i =>
{
var result = new List <PipelineTestsMyTestClass>();
for (var j = 1; j <= i; j++)
{
result.Add(new PipelineTestsMyTestClass {
Id = j, Name = $"Bob {j}"
});
}
return(result);
})
.Project(i =>
{
var nameParts = i.Name.Split(' ');
return(new PipelineTestsOtherTestClass {
Id = i.Id, FirstName = nameParts[0], LastName = nameParts[1]
});
})
.TransformInPlace(i =>
{
i.Id -= 1;
i.FirstName = "Alice";
i.LastName = $"{i.Id}";
})
.WriteTo(writer);
await pipeline.ExecuteAsync();
Assert.Equal(1000, writer.ItemsWritten.Count);
Assert.Equal(0, writer.ItemsWritten[0].Id);
Assert.Equal("Alice", writer.ItemsWritten[0].FirstName);
Assert.Equal("0", writer.ItemsWritten[0].LastName);
Assert.Equal(999, writer.ItemsWritten[999].Id);
Assert.Equal("Alice", writer.ItemsWritten[999].FirstName);
Assert.Equal("999", writer.ItemsWritten[999].LastName);
}
}
public async Task RunsAggregatorStep()
{
using (var writer = new TestBulkWriter <int>())
{
var items = Enumerable.Range(1, 1000).Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
});
var pipeline = EtlPipeline.StartWith(items)
.Aggregate(f => f.Sum(c => c.Id))
.WriteTo(writer);
await pipeline.ExecuteAsync();
Assert.Single(writer.ItemsWritten);
Assert.Equal(Enumerable.Range(1, 1000).Sum(), writer.ItemsWritten[0]);
}
}
