private IPropagatorBlock <int, Data> UseLinesReaderAndParser(StreamReader peopleJsonStream, DataPool dataPool, CancellationToken cancellation)
{
// Create blocks
// NOTE:
// - extract part which must be single-thread
// - ability to replace just reading (e.g. from db)
var readLinesBlock = DataflowFacade.TransformManyBlock <int, Data>(
"ReadLines",
x => - 1,
x => _streamLinesReader.Read(peopleJsonStream, x)
.Select(line =>
{
var data = dataPool.Rent();
data.PersonJson = line;
return(data);
}),
cancellation);
// NOTE: can be multi-thread
var parseDataBlock = DataflowFacade.TransformBlock(
"ParseData",
Data.IdGetter,
x => _dataParser.Parse(x),
cancellation);
// Link blocks
readLinesBlock.LinkWithCompletion(parseDataBlock);
return(DataflowBlock.Encapsulate(readLinesBlock, parseDataBlock));
}
public ProcessingBlock <TData> Create <TData>(string name,
Func <TData, int> dataIdGetter,
IProgress <PipelineProgress> progress,
Lazy <int> estimatedInputCount,
int inputPerReport,
CancellationToken cancellation)
{
var state = new State();
// Create blocks
var reportBlock = DataflowFacade.TransformBlock(
name,
dataIdGetter,
x => TryReport(state, estimatedInputCount.Value, inputPerReport, progress),
cancellation);
// Handle completion
var completion = reportBlock.Completion.ContinueWithStatusPropagation(
x =>
{
if (!x.IsFaulted && !x.IsCanceled && !state.Reported100)
{
progress.Report(new PipelineProgress {
Percentage = 100
});
}
});
return(new ProcessingBlock <TData>(reportBlock, completion));
}
public ProcessingBlock <DataBatch> Create(string targetFilePath, CancellationToken cancellation)
{
var targetDirectoryPath = Path.GetDirectoryName(targetFilePath);
if (!Directory.Exists(targetDirectoryPath))
{
Directory.CreateDirectory(targetDirectoryPath);
}
var writer = new StreamWriter(targetFilePath);
// Create blocks
var writeBlock = DataflowFacade.TransformBlock(
"WriteData",
DataBatch.IdGetter,
x => x.Data.ForEach(item => _dataWriter.Write(writer, item)),
cancellation);
// Handle completion
var completion = writeBlock.Completion.ContinueWithStatusPropagation(_ => writer.Dispose());
return(new ProcessingBlock <DataBatch>
{
Processor = writeBlock,
Completion = completion
});
}
private StartableBlock <DataBatch <Person> > CreatePeoplePipeline(string targetFilePath, PeopleSummary builtSummary, IProgress <PipelineProgress> progress, CancellationToken cancellation)
{
var cancellationSource = CancellationTokenSource.CreateLinkedTokenSource(cancellation);
// Create blocks
var readBlock = _readPeopleBlockFactory.Create(cancellationSource.Token);
var writeBlock = _writeCsvBlockFactory.Create <Person>(targetFilePath, true, cancellationSource.Token);
var buildSummaryBlock = new ProcessingBlock <DataBatch <Person> >(DataflowFacade.TransformBlock("BuildSummary",
DataBatch <Person> .IdGetter,
x => x.Data.ForEach(person => _peopleSummaryBuilder.Include(person, builtSummary)),
cancellationSource.Token));
var progressBlock = _progressReportingBlockFactory.Create("PersonProgress",
DataBatch <Person> .IdGetter,
progress,
readBlock.EstimatedOutputCount,
1,
cancellationSource.Token);
// Create pipeline
var pipeline = _straightPipelineFactory.Create(readBlock,
new[] { writeBlock, buildSummaryBlock, progressBlock },
cancellationSource);
pipeline.ContinueCompletionWith(_ => cancellationSource.Dispose());
return(pipeline);
}
private IEnumerable <ProcessingBlock <DataBatch> > CreateExtraProcessingBlocks(CancellationTokenSource cancellationSource)
{
return(Enumerable.Range(1, Settings.ExtraProcessingBlocksCount)
.Select(x => new ProcessingBlock <DataBatch>(DataflowFacade.TransformBlock($"ExtraProcessing {x}",
DataBatch.IdGetter,
batch => batch.Data.ForEach(_ => ComputationsHelper.PerformTimeConsumingOperation()),
cancellationSource.Token,
Settings.DegreeOfParallelism))));
}
private IPropagatorBlock <DataBatch, DataBatch> UseDataReader(StreamReader peopleJsonStream, DataPool dataPool, CancellationToken cancellation)
{
return(DataflowFacade.TransformBlock <DataBatch, DataBatch>(
"ReadData",
DataBatch.IdGetter,
x =>
{
x.Data = _dataReader.Read(peopleJsonStream, x.IntendedSize, dataPool).ToList();
return x;
},
cancellation));
}
public static ProcessingBlock <TData> Create(string name,
Func <TData, int> dataIdGetter,
Func <TData, TData> process,
CancellationToken cancellation,
int maxDegreeOfParallelism = 1)
{
var processor = DataflowFacade.TransformBlock <TData>(name, dataIdGetter, process, cancellation, maxDegreeOfParallelism);
return(new ProcessingBlock <TData>
{
Processor = processor,
Completion = processor.Completion
});
}
public ProcessingBlock <TData> Create <TData>(Func <TData, int> dataIdGetter, CancellationToken cancellation)
{
// Create blocks
var throwBlock = DataflowFacade.TransformBlock(
"Throw",
dataIdGetter,
x =>
{
throw new InvalidOperationException();
},
cancellation);
return(new ProcessingBlock <TData>(throwBlock));
}
public ProcessingBlock <DataBatch <TItem> > Create <TItem>(string targetFilePath, bool appendTargetFile, CancellationToken cancellation)
{
var csvWriter = new Lazy <CsvWriter>(() => CreateCsvWriter(targetFilePath, appendTargetFile));
// Create blocks
var writeBlock = DataflowFacade.TransformBlock($"Write{typeof(TItem).Name}",
DataBatch <TItem> .IdGetter,
x => csvWriter.Value.WriteRecords(x.Data),
cancellation);
// Handle completion
var completion = writeBlock.Completion.ContinueWithStatusPropagation(_ => csvWriter.ValueIfCreated()?.Dispose());
return(new ProcessingBlock <DataBatch <TItem> >(writeBlock, completion));
}
public StartableBlock <DataBatch <Person> > Create(CancellationToken cancellation)
{
var batchSize = Settings.ReadingBatchSize;
var batchesCount = new Lazy <int>(() => GetBatchesCount(batchSize));
// Create blocks
var bufferBlock = DataflowFacade.BufferBlock <DataBatch <Person> >(cancellation);
var readBlock = DataflowFacade.TransformBlock <DataBatch <Person> >("ReadPerson",
DataBatch <Person> .IdGetter,
x => x.Data = _peopleBatchReader.Read(x.DataOffset, x.IntendedSize),
cancellation);
// Link blocks
bufferBlock.LinkWithCompletion(readBlock);
return(new StartableBlock <DataBatch <Person> >(
() => Start(batchSize, batchesCount.Value, bufferBlock),
readBlock,
batchesCount));
}
public StartableBlock <Data> Create(string peopleJsonFilePath,
string targetFilePath,
string errorsFilePath,
IProgress <PipelineProgress> progress,
CancellationTokenSource cancellationSource)
{
var dataPool = new DataPool();
// Create blocks
// TODO: Progress reporting approach 1: before anything
var readBlock = _readingBlockFactory.Create(peopleJsonFilePath, dataPool, cancellationSource.Token);
var validateBlock = new ProcessingBlock <Data>(DataflowFacade.TransformBlock("Validate",
Data.IdGetter,
x => _personValidator.Validate(x),
cancellationSource.Token,
Settings.DegreeOfParallelism));
var computeFieldsBlock = new ProcessingBlock <Data>(DataflowFacade.TransformBlock("ComputeFields",
Data.IdGetter,
x => _personFieldsComputer.Compute(x),
cancellationSource.Token,
Settings.DegreeOfParallelism));
var writeBlock = _writingBlockFactory.Create(targetFilePath, cancellationSource.Token);
var throwBlock = Settings.ThrowTest
? _throwingBlockFactory.Create(Data.IdGetter, cancellationSource.Token)
: _emptyBlockFactory.Create <Data>(cancellationSource.Token);
var handleErrorBlock = _writingBlockFactory.Create(errorsFilePath, cancellationSource.Token);
var progressBlock = _progressReportingBlockFactory.Create("ReportProgress",
Data.IdGetter,
progress,
readBlock.EstimatedOutputCount,
Settings.ProgressBatchSize,
cancellationSource.Token);
return(_railroadPipelineFactory.Create(readBlock,
new[] { validateBlock, computeFieldsBlock, writeBlock, throwBlock },
handleErrorBlock,
progressBlock,
x => x.IsValid,
cancellationSource));
}
public StartableBlock <DataBatch> Create(string peopleJsonFilePath,
string targetFilePath,
IProgress <PipelineProgress> progress,
CancellationTokenSource cancellationSource)
{
var dataPool = new DataPool();
// Create blocks
var readBlock = _readingBlockFactory.Create(peopleJsonFilePath, dataPool, cancellationSource.Token);
var processBlock = new ProcessingBlock <DataBatch>(DataflowFacade.TransformBlock("Process",
DataBatch.IdGetter,
x =>
{
x.Data.Where(item => item.IsValid).ForEach(_personValidator.Validate);
x.Data.Where(item => item.IsValid).ForEach(_personFieldsComputer.Compute);
for (var i = 0; i < Settings.ExtraProcessingBlocksCount; i++)
{
x.Data.ForEach(_ => ComputationsHelper.PerformTimeConsumingOperation());
}
},
cancellationSource.Token,
Settings.DegreeOfParallelism));
var writeBlock = _writingBlockFactory.Create(targetFilePath, cancellationSource.Token);
var progressBlock = _progressReportingBlockFactory.Create("ReportProgress",
DataBatch.IdGetter,
progress,
readBlock.EstimatedOutputCount,
1,
cancellationSource.Token);
var disposeBlock = new ProcessingBlock <DataBatch>(DataflowFacade.TransformBlock("DisposeData",
DataBatch.IdGetter,
x => x.Data.ForEach(dataPool.Return),
cancellationSource.Token));
return(_straightPipelineFactory.Create(readBlock,
new[] { processBlock, writeBlock, progressBlock, disposeBlock },
cancellationSource));
}
public StartableBlock <DataBatch <PeopleSummary> > Create(CancellationToken cancellation)
{
var readBlock = DataflowFacade.TransformBlock <DataBatch <PeopleSummary> >("ReadPeopleSummary",
DataBatch <PeopleSummary> .IdGetter,
x => x.Data = new[] { _peopleSummaryReader.Read() },
cancellation);
return(new StartableBlock <DataBatch <PeopleSummary> >(
() =>
{
readBlock.Post(new DataBatch <PeopleSummary>
{
Number = -1,
DataOffset = 0,
IntendedSize = 1
});
readBlock.Complete();
},
readBlock,
1));
}
public StartableBlock <DataBatch> Create(string peopleJsonFilePath,
string targetFilePath,
IProgress <PipelineProgress> progress,
CancellationTokenSource cancellationSource)
{
var dataPool = new DataPool();
// Create blocks
var readBlock = _readingBlockFactory.Create(peopleJsonFilePath, dataPool, cancellationSource.Token);
var validateBlock = new ProcessingBlock <DataBatch>(DataflowFacade.TransformBlock("Validate",
DataBatch.IdGetter,
x => x.Data.Where(item => item.IsValid).ForEach(_personValidator.Validate),
cancellationSource.Token,
Settings.DegreeOfParallelism));
var computeFieldsBlock = new ProcessingBlock <DataBatch>(DataflowFacade.TransformBlock("ComputeFields",
DataBatch.IdGetter,
x => x.Data.Where(item => item.IsValid).ForEach(_personFieldsComputer.Compute),
cancellationSource.Token,
Settings.DegreeOfParallelism));
var extraProcessingBlocks = CreateExtraProcessingBlocks(cancellationSource);
var writeBlock = _writingBlockFactory.Create(targetFilePath, cancellationSource.Token);
var progressBlock = _progressReportingBlockFactory.Create("ReportProgress",
DataBatch.IdGetter,
progress,
readBlock.EstimatedOutputCount,
1,
cancellationSource.Token);
var disposeBlock = new ProcessingBlock <DataBatch>(DataflowFacade.TransformBlock("DisposeData",
DataBatch.IdGetter,
x => x.Data.ForEach(dataPool.Return),
cancellationSource.Token));
return(_straightPipelineFactory.Create(readBlock,
new[] { validateBlock, computeFieldsBlock }.Concat(extraProcessingBlocks)
.Concat(new[] { writeBlock, progressBlock, disposeBlock })
.ToArray(),
cancellationSource));
}