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 = ProcessingBlock <DataBatch> .Create("Validate",
DataBatch.IdGetter,
x => x.Data.Where(item => item.IsValid).ForEach(_personValidator.Validate),
cancellationSource.Token,
Settings.ProcessInParallel?Settings.MaxDegreeOfParallelism : 1);
var computeFieldsBlock = ProcessingBlock <DataBatch> .Create("ComputeFields",
DataBatch.IdGetter,
x => x.Data.Where(item => item.IsValid).ForEach(_personFieldsComputer.Compute),
cancellationSource.Token,
Settings.ProcessInParallel?Settings.MaxDegreeOfParallelism : 1);
var extraProcessingBlocks = CreateExtraProcessingBlocks(cancellationSource);
var writeBlock = _writingBlockFactory.Create(targetFilePath, cancellationSource.Token);
var progressBlock = _progressReportingBlockFactory.Create(DataBatch.IdGetter, progress, readBlock.EstimatedOutputCount, 1, cancellationSource.Token);
var disposeBlock = ProcessingBlock <DataBatch> .Create("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));
}
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));
}