/// <inheritdoc />
protected override void ProcessCore(CsvReaderVisitorBase visitor)
{
var tokenizer = new CsvTokenizer(_delimiter);
bool ignoreUTF8ByteOrderMark = _ignoreUTF8ByteOrderMark;
var bytes = _sequence;
if (bytes.IsSingleSegment)
{
CsvReadOnlyMemoryInput.ProcessFullSegment(bytes.First.Span, ignoreUTF8ByteOrderMark, tokenizer, visitor);
return;
}
var enumerator = bytes.GetEnumerator();
if (ignoreUTF8ByteOrderMark && EatUTF8BOM(tokenizer, visitor, ref enumerator))
{
return;
}
while (enumerator.MoveNext())
{
tokenizer.ProcessNextChunk(enumerator.Current.Span, visitor);
}
tokenizer.ProcessEndOfStream(visitor);
}
/// <summary>
/// Describes the contents of this CSV data stream to a <see cref="CsvReaderVisitorBase"/>.
/// </summary>
/// <param name="visitor">
/// The <see cref="CsvReaderVisitorBase"/> to describe this CSV data stream to.
/// </param>
/// <param name="progress">
/// An optional <see cref="IProgress{T}"/> instance that will receive a report of the size
/// of each chunk (in bytes) as processing finishes, followed by one more report with a zero
/// when the last chunk in the stream has been processed.
/// </param>
/// <param name="cancellationToken">
/// An optional <see cref="CancellationToken"/> that may be used to signal cancellation.
/// </param>
/// <returns>
/// A <see cref="ValueTask"/> encapsulating the operation.
/// </returns>
/// <exception cref="InvalidOperationException">
/// Thrown when this stream has already been processed.
/// </exception>
/// <exception cref="OperationCanceledException">
/// Thrown to acknowledge a canceled <paramref name="cancellationToken"/>. Some subclasses
/// may throw an instance of a subclass, such as <see cref="TaskCanceledException"/>.
/// </exception>
public async ValueTask ProcessAsync(CsvReaderVisitorBase visitor, IProgress <int> progress = null, CancellationToken cancellationToken = default)
{
if (!(Interlocked.CompareExchange(ref _processingHasStarted, ProcessingHasStartedSentinel, null) is null))
{
ThrowProcessingHasAlreadyStartedException();
}
await ProcessAsyncCore(visitor, progress, cancellationToken).ConfigureAwait(false);
}
/// <summary>
/// Describes the contents of this CSV data stream to a <see cref="CsvReaderVisitorBase"/>.
/// </summary>
/// <param name="visitor">
/// The <see cref="CsvReaderVisitorBase"/> to describe this CSV data stream to.
/// </param>
/// <exception cref="InvalidOperationException">
/// Thrown when this stream has already been processed.
/// </exception>
public void Process(CsvReaderVisitorBase visitor)
{
if (!(Interlocked.CompareExchange(ref _processingHasStarted, ProcessingHasStartedSentinel, null) is null))
{
ThrowProcessingHasAlreadyStartedException();
}
ProcessCore(visitor);
}
/// <inheritdoc />
protected override unsafe void ProcessCore(CsvReaderVisitorBase visitor)
{
var tokenizer = new CsvTokenizer(_delimiter);
using (var fl = new FileStream(_csvFilePath, FileMode.Open, FileAccess.Read, FileShare.Read, 4096, FileOptions.SequentialScan))
{
long length = fl.Length;
if (length == 0)
{
tokenizer.ProcessEndOfStream(visitor);
return;
}
using (var memoryMappedFile = MemoryMappedFile.CreateFromFile(fl, null, 0, MemoryMappedFileAccess.Read, HandleInheritability.None, leaveOpen: true))
using (var accessor = memoryMappedFile.CreateViewAccessor(0, 0, MemoryMappedFileAccess.Read))
{
var handle = accessor.SafeMemoryMappedViewHandle;
byte *ptr = null;
RuntimeHelpers.PrepareConstrainedRegions();
try
{
handle.AcquirePointer(ref ptr);
if (_ignoreUTF8ByteOrderMark)
{
var head = new ReadOnlySpan <byte>(UTF8BOM, 0, length < 3 ? unchecked ((int)length) : 3);
if (head.SequenceEqual(new ReadOnlySpan <byte>(ptr, head.Length)))
{
length -= head.Length;
ptr += head.Length;
}
}
while (length > int.MaxValue)
{
tokenizer.ProcessNextChunk(new ReadOnlySpan <byte>(ptr, int.MaxValue), visitor);
length -= int.MaxValue;
ptr += int.MaxValue;
}
tokenizer.ProcessNextChunk(new ReadOnlySpan <byte>(ptr, unchecked ((int)length)), visitor);
tokenizer.ProcessEndOfStream(visitor);
}
finally
{
if (ptr != null)
{
handle.ReleasePointer();
}
}
}
}
}
/// <inheritdoc />
protected override async ValueTask ProcessAsyncCore(CsvReaderVisitorBase visitor, IProgress <int> progress, CancellationToken cancellationToken)
{
var tokenizer = new CsvTokenizer(_delimiter);
var reader = _reader;
if (_ignoreUTF8ByteOrderMark && await EatUTF8BOMAsync(tokenizer, visitor, progress, cancellationToken).ConfigureAwait(false))
{
return;
}
while (true)
{
var result = await reader.ReadAsync(cancellationToken).ConfigureAwait(false);
if (result.IsCanceled)
{
throw new OperationCanceledException(cancellationToken);
}
var buffer = result.Buffer;
foreach (var segment in buffer)
{
tokenizer.ProcessNextChunk(segment.Span, visitor);
}
reader.AdvanceTo(buffer.End);
if (progress != null)
{
long totalLength = buffer.Length;
while (totalLength > int.MaxValue)
{
progress.Report(int.MaxValue);
totalLength -= int.MaxValue;
}
if (totalLength != 0)
{
progress.Report(unchecked ((int)totalLength));
}
}
if (result.IsCompleted)
{
break;
}
}
tokenizer.ProcessEndOfStream(visitor);
progress?.Report(0);
}
/// <inheritdoc />
protected override async ValueTask ProcessAsyncCore(CsvReaderVisitorBase visitor, IProgress <int> progress, CancellationToken cancellationToken)
{
// not all streams support cancellation, so we might as well do this ourselves. it
// does involve a volatile read, so don't go overboard.
cancellationToken.ThrowIfCancellationRequested();
var tokenizer = new CsvTokenizer(_delimiter);
var csvStream = _csvStream;
int minReadBufferByteCount = _minReadBufferByteCount;
var readBufferPool = _readBufferPool;
byte[] readBuffer;
if (readBufferPool is null)
{
readBuffer = new byte[minReadBufferByteCount];
}
else
{
readBuffer = readBufferPool.Rent(minReadBufferByteCount);
}
try
{
if (_ignoreUTF8ByteOrderMark && await EatUTF8BOMAsync(tokenizer, visitor, csvStream, readBuffer, progress, cancellationToken).ConfigureAwait(false))
{
return;
}
int cnt;
while ((cnt = await csvStream.ReadAsync(readBuffer, 0, readBuffer.Length, cancellationToken).ConfigureAwait(false)) != 0)
{
// not all streams support cancellation, so we might as well do this ourselves. it
// does involve a volatile read, so don't go overboard.
cancellationToken.ThrowIfCancellationRequested();
tokenizer.ProcessNextChunk(new ReadOnlySpan <byte>(readBuffer, 0, cnt), visitor);
progress?.Report(cnt);
}
}
finally
{
readBufferPool?.Return(readBuffer, clearArray: true);
}
tokenizer.ProcessEndOfStream(visitor);
progress?.Report(0);
}
private static bool EatUTF8BOM(CsvTokenizer tokenizer, CsvReaderVisitorBase visitor, Stream csvStream, byte[] readBuffer)
{
if (readBuffer.Length < 3)
{
// don't bother pooling; nobody should really ever care.
readBuffer = new byte[3];
}
int byteCount = 0;
while (byteCount < 3)
{
int readLength = csvStream.Read(readBuffer, byteCount, readBuffer.Length - byteCount);
if (readLength == 0)
{
if (byteCount != 0)
{
if (!new ReadOnlySpan <byte>(readBuffer, 0, byteCount).SequenceEqual(new ReadOnlySpan <byte>(UTF8BOM, 0, byteCount)))
{
tokenizer.ProcessNextChunk(new ReadOnlySpan <byte>(readBuffer, 0, byteCount), visitor);
}
}
tokenizer.ProcessEndOfStream(visitor);
return(true);
}
byteCount += readLength;
}
var buf = new ReadOnlySpan <byte>(readBuffer, 0, byteCount);
if (buf.StartsWith(UTF8BOM))
{
buf = buf.Slice(3);
}
tokenizer.ProcessNextChunk(buf, visitor);
return(false);
}
/// <inheritdoc />
protected override void ProcessCore(CsvReaderVisitorBase visitor)
{
var tokenizer = new CsvTokenizer(_delimiter);
var csvStream = _csvStream;
int minReadBufferByteCount = _minReadBufferByteCount;
var readBufferPool = _readBufferPool;
byte[] readBuffer;
if (readBufferPool is null)
{
readBuffer = new byte[minReadBufferByteCount];
}
else
{
readBuffer = readBufferPool.Rent(minReadBufferByteCount);
}
try
{
if (_ignoreUTF8ByteOrderMark && EatUTF8BOM(tokenizer, visitor, csvStream, readBuffer))
{
return;
}
int cnt;
while ((cnt = csvStream.Read(readBuffer, 0, readBuffer.Length)) != 0)
{
tokenizer.ProcessNextChunk(new ReadOnlySpan <byte>(readBuffer, 0, cnt), visitor);
}
}
finally
{
readBufferPool?.Return(readBuffer, clearArray: true);
}
tokenizer.ProcessEndOfStream(visitor);
}
private static async ValueTask <bool> EatUTF8BOMAsync(CsvTokenizer tokenizer, CsvReaderVisitorBase visitor, Stream csvStream, byte[] readBuffer, IProgress <int> progress, CancellationToken cancellationToken)
{
if (readBuffer.Length < 3)
{
// don't bother pooling; nobody should really ever care.
readBuffer = new byte[3];
}
int byteCount = 0;
while (byteCount < 3)
{
int readLength = await csvStream.ReadAsync(readBuffer, byteCount, readBuffer.Length - byteCount, cancellationToken).ConfigureAwait(false);
// not all streams support cancellation, so we might as well do this ourselves. it
// does involve a volatile read, so don't go overboard.
cancellationToken.ThrowIfCancellationRequested();
if (readLength == 0)
{
if (byteCount != 0)
{
if (!new ReadOnlySpan <byte>(readBuffer, 0, byteCount).SequenceEqual(new ReadOnlySpan <byte>(UTF8BOM, 0, byteCount)))
{
tokenizer.ProcessNextChunk(new ReadOnlySpan <byte>(readBuffer, 0, byteCount), visitor);
}
progress?.Report(byteCount);
}
tokenizer.ProcessEndOfStream(visitor);
progress?.Report(0);
return(true);
}
byteCount += readLength;
}
var buf = new ReadOnlyMemory <byte>(readBuffer, 0, byteCount);
if (buf.Span.StartsWith(UTF8BOM))
{
buf = buf.Slice(3);
}
tokenizer.ProcessNextChunk(buf.Span, visitor);
progress?.Report(byteCount);
return(false);
}
/// <summary>
/// Implements the inner logic for <see cref="ProcessAsync"/>.
/// </summary>
/// <param name="visitor">
/// The <see cref="CsvReaderVisitorBase"/> to describe this CSV data stream to.
/// </param>
/// <param name="progress">
/// An optional <see cref="IProgress{T}"/> instance that will receive a report of the size
/// of each chunk (in bytes) as processing finishes, followed by one more report with a zero
/// when the last chunk in the stream has been processed.
/// </param>
/// <param name="cancellationToken">
/// An optional <see cref="CancellationToken"/> that may be used to signal cancellation.
/// </param>
/// <returns>
/// A <see cref="ValueTask"/> encapsulating the operation.
/// </returns>
/// <exception cref="InvalidOperationException">
/// Thrown when this stream has already been processed.
/// </exception>
/// <exception cref="OperationCanceledException">
/// Thrown to acknowledge a canceled <paramref name="cancellationToken"/>. Some subclasses
/// may throw an instance of a subclass, such as <see cref="TaskCanceledException"/>.
/// </exception>
protected abstract ValueTask ProcessAsyncCore(CsvReaderVisitorBase visitor, IProgress <int> progress, CancellationToken cancellationToken);
/// <summary> /// Implements the inner logic for <see cref="Process"/>. /// </summary> /// <param name="visitor"> /// The <see cref="CsvReaderVisitorBase"/> to describe this CSV data stream to. /// </param> /// <remarks> /// The base class will call this method at most once per instance. /// </remarks> protected abstract void ProcessCore(CsvReaderVisitorBase visitor);
private static bool EatUTF8BOM(CsvTokenizer tokenizer, CsvReaderVisitorBase visitor, ref ReadOnlySequence <byte> .Enumerator enumerator)
{
ReadOnlyMemory <byte> segment;
while (true)
{
if (!enumerator.MoveNext())
{
tokenizer.ProcessEndOfStream(visitor);
return(true);
}
segment = enumerator.Current;
if (!segment.IsEmpty)
{
break;
}
}
var span = segment.Span;
ReadOnlySpan <byte> head = UTF8BOM;
// this greed should **probably** pay off most of the time.
if (span.Length >= 3)
{
if (span.StartsWith(head))
{
span = span.Slice(3);
}
tokenizer.ProcessNextChunk(span, visitor);
return(false);
}
int alreadyEaten = 0;
while (true)
{
if (span[0] == head[alreadyEaten])
{
span = span.Slice(1);
if (++alreadyEaten == 3)
{
tokenizer.ProcessNextChunk(span, visitor);
return(false);
}
}
else
{
tokenizer.ProcessNextChunk(head.Slice(0, alreadyEaten), visitor);
tokenizer.ProcessNextChunk(span, visitor);
return(false);
}
if (span.IsEmpty)
{
while (true)
{
if (!enumerator.MoveNext())
{
tokenizer.ProcessEndOfStream(visitor);
return(true);
}
segment = enumerator.Current;
if (!segment.IsEmpty)
{
break;
}
}
span = segment.Span;
}
}
}
internal static void ProcessFullSegment(ReadOnlySpan <byte> bytes, bool ignoreUTF8ByteOrderMark, CsvTokenizer tokenizer, CsvReaderVisitorBase visitor)
{
if (ignoreUTF8ByteOrderMark)
{
var head = new ReadOnlySpan <byte>(UTF8BOM, 0, bytes.Length < 3 ? bytes.Length : 3);
if (bytes.StartsWith(head))
{
bytes = bytes.Slice(head.Length);
}
}
tokenizer.ProcessNextChunk(bytes, visitor);
tokenizer.ProcessEndOfStream(visitor);
}
/// <inheritdoc />
protected override void ProcessCore(CsvReaderVisitorBase visitor)
{
ProcessFullSegment(_memory.Span, _ignoreUTF8ByteOrderMark, new CsvTokenizer(_delimiter), visitor);
}
private async ValueTask <bool> EatUTF8BOMAsync(CsvTokenizer tokenizer, CsvReaderVisitorBase visitor, IProgress <int> progress, CancellationToken cancellationToken)
{
var reader = _reader;
ReadOnlySequence <byte> buffer;
// keep asking for more until we've seen either 3+ bytes or the end of the data.
while (true)
{
var result = await reader.ReadAsync(cancellationToken).ConfigureAwait(false);
if (result.IsCanceled)
{
throw new OperationCanceledException(cancellationToken);
}
buffer = result.Buffer;
if (buffer.Length >= 3)
{
// we've seen 3+ bytes.
break;
}
if (result.IsCompleted)
{
// we've seen the end of the data.
Finish();
tokenizer.ProcessEndOfStream(visitor);
reader.AdvanceTo(buffer.End);
progress?.Report(0);
return(true);
}
// tell the reader that we've looked at everything it had to give us, and we weren't
// able to consume any of it, so the next read should have everything we've seen so
// far, plus at least one more byte.
reader.AdvanceTo(buffer.Start, buffer.End);
}
Finish();
return(false);
void Finish()
{
Span <byte> upToFirstThreeBytes = stackalloc byte[3];
int alreadyEaten = 0;
foreach (var segment in buffer)
{
int lengthToCopy = 3 - alreadyEaten;
if (lengthToCopy > segment.Length)
{
lengthToCopy = segment.Length;
}
segment.Slice(0, lengthToCopy).Span.CopyTo(upToFirstThreeBytes.Slice(alreadyEaten, lengthToCopy));
alreadyEaten += lengthToCopy;
if (alreadyEaten == 3)
{
break;
}
}
upToFirstThreeBytes = upToFirstThreeBytes.Slice(0, alreadyEaten);
var head = new ReadOnlySpan <byte>(UTF8BOM, 0, alreadyEaten);
if (!upToFirstThreeBytes.SequenceEqual(head))
{
tokenizer.ProcessNextChunk(upToFirstThreeBytes, visitor);
}
reader.AdvanceTo(buffer.GetPosition(alreadyEaten));
progress?.Report(alreadyEaten);
}
}
