为生产者/消费者实现异步流 [英] Implementing async stream for producer/consumer
问题描述
有一个库将其结果输出到给定的 Stream
对象中.我想在 lib 完成之前开始使用结果.Stream
应该阻塞以简化使用并避免如果生产者跑得太远则过度消耗内存;线程安全,允许生产者和消费者独立存在.
There is a lib that outputs its results into a given Stream
object. I would like to begin consuming the results before the lib is done. The Stream
should be blocking to simplify usage and avoid excessive memory consumption if producer runs ahead too far; thread safe to allow independent existence of producer and consumer.
一旦库完成,生产者线程应该关闭流,从而通知消费者没有更多数据.
Once the lib finishes, the producer thread should close the stream, hence notifying consumer that there is no more data.
我正在考虑使用 NetworkStream
或 PipeStream
(匿名),但两者在通过内核发送数据时可能都很慢.
I was thinking of using NetworkStream
or PipeStream
(anonymous), but both are probably slow as they send data through kernel.
有什么推荐吗?
var stream = new AsyncBlockingBufferedStream();
void ProduceData()
{
// In producer thread
externalLib.GenerateData(stream);
stream.Close();
}
void ConsumeData()
{
// In consumer thread
int read;
while ((read = stream.Read(...)) != 0)
{ ... }
}
推荐答案
基于 Chris Taylor 之前的回答,这里是我自己的,经过修订的,基于块的操作更快,并更正了写入完成通知.它现在被标记为 wiki,因此您可以更改它.
Based on the previous answer by Chris Taylor, here's my own, revised, with much faster block based operations and corrected write completion notifications. It's marked as wiki now, so you can change it.
public class BlockingStream : Stream
{
private readonly BlockingCollection<byte[]> _blocks;
private byte[] _currentBlock;
private int _currentBlockIndex;
public BlockingStream(int streamWriteCountCache)
{
_blocks = new BlockingCollection<byte[]>(streamWriteCountCache);
}
public override bool CanTimeout { get { return false; } }
public override bool CanRead { get { return true; } }
public override bool CanSeek { get { return false; } }
public override bool CanWrite { get { return true; } }
public override long Length { get { throw new NotSupportedException(); } }
public override void Flush() {}
public long TotalBytesWritten { get; private set; }
public int WriteCount { get; private set; }
public override long Position
{
get { throw new NotSupportedException(); }
set { throw new NotSupportedException(); }
}
public override long Seek(long offset, SeekOrigin origin)
{
throw new NotSupportedException();
}
public override void SetLength(long value)
{
throw new NotSupportedException();
}
public override int Read(byte[] buffer, int offset, int count)
{
ValidateBufferArgs(buffer, offset, count);
int bytesRead = 0;
while (true)
{
if (_currentBlock != null)
{
int copy = Math.Min(count - bytesRead, _currentBlock.Length - _currentBlockIndex);
Array.Copy(_currentBlock, _currentBlockIndex, buffer, offset + bytesRead, copy);
_currentBlockIndex += copy;
bytesRead += copy;
if (_currentBlock.Length <= _currentBlockIndex)
{
_currentBlock = null;
_currentBlockIndex = 0;
}
if (bytesRead == count)
return bytesRead;
}
if (!_blocks.TryTake(out _currentBlock, Timeout.Infinite))
return bytesRead;
}
}
public override void Write(byte[] buffer, int offset, int count)
{
ValidateBufferArgs(buffer, offset, count);
var newBuf = new byte[count];
Array.Copy(buffer, offset, newBuf, 0, count);
_blocks.Add(newBuf);
TotalBytesWritten += count;
WriteCount++;
}
protected override void Dispose(bool disposing)
{
base.Dispose(disposing);
if (disposing)
{
_blocks.Dispose();
}
}
public override void Close()
{
CompleteWriting();
base.Close();
}
public void CompleteWriting()
{
_blocks.CompleteAdding();
}
private static void ValidateBufferArgs(byte[] buffer, int offset, int count)
{
if (buffer == null)
throw new ArgumentNullException("buffer");
if (offset < 0)
throw new ArgumentOutOfRangeException("offset");
if (count < 0)
throw new ArgumentOutOfRangeException("count");
if (buffer.Length - offset < count)
throw new ArgumentException("buffer.Length - offset < count");
}
}
这篇关于为生产者/消费者实现异步流的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!