private ThreadAheadReadable(CharReadable actual, int bufferSize) : base(actual)
{
this._actual = actual;
this._theOtherBuffer = new SectionedCharBuffer(bufferSize);
this._sourceDescription = actual.SourceDescription();
start();
}
/// <summary>
/// Extracts the first line, i.e characters until the first newline or end of stream.
/// Reads one character at a time to be sure not to read too far ahead. The stream is left
/// in a state of either exhausted or at the beginning of the next line of data.
/// </summary>
/// <param name="source"> <seealso cref="CharReadable"/> to read from. </param>
/// <returns> char[] containing characters until the first newline character or end of stream. </returns>
/// <exception cref="IOException"> on I/O reading error. </exception>
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
//ORIGINAL LINE: public static char[] extractFirstLineFrom(CharReadable source) throws java.io.IOException
public static char[] ExtractFirstLineFrom(CharReadable source)
{
char[] result = new char[100];
int cursor = 0;
int read;
bool foundEol = false;
do
{
// Grow on demand
if (cursor >= result.Length)
{
result = Arrays.copyOf(result, cursor * 2);
}
// Read one character
read = source.Read(result, cursor, 1);
if (read > 0)
{
foundEol = BufferedCharSeeker.IsEolChar(result[cursor]);
if (!foundEol)
{
cursor++;
}
}
} while (read > 0 && !foundEol);
return(Arrays.copyOf(result, cursor));
}
private void WarnAboutDuplicateSourceFiles(ISet <string> seenSourceFiles, CharReadable source)
{
string sourceDescription = source.SourceDescription();
if (!seenSourceFiles.Add(sourceDescription))
{
_monitor.duplicateSourceFile(sourceDescription);
}
}
/// <summary>
/// Instantiates a <seealso cref="BufferedCharSeeker"/> with optional <seealso cref="ThreadAheadReadable read-ahead"/> capability.
/// </summary>
/// <param name="reader"> the <seealso cref="CharReadable"/> which is the source of data, f.ex. a <seealso cref="System.IO.StreamReader_FileReader"/>. </param>
/// <param name="config"> <seealso cref="Configuration"/> for the resulting <seealso cref="CharSeeker"/>. </param>
/// <param name="readAhead"> whether or not to start a <seealso cref="ThreadAheadReadable read-ahead thread"/>
/// which strives towards always keeping one buffer worth of data read and available from I/O when it's
/// time for the <seealso cref="BufferedCharSeeker"/> to read more data. </param>
/// <returns> a <seealso cref="CharSeeker"/> with optional <seealso cref="ThreadAheadReadable read-ahead"/> capability. </returns>
public static CharSeeker CharSeeker(CharReadable reader, Configuration config, bool readAhead)
{
if (readAhead)
{ // Thread that always has one buffer read ahead
reader = threadAhead(reader, config.BufferSize());
}
// Give the reader to the char seeker
return(new BufferedCharSeeker(new AutoReadingSource(reader, config.BufferSize()), config));
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
//ORIGINAL LINE: private long[] sample(Iterable<DataFactory> dataFactories, Header.Factory headerFactory, System.Func<org.neo4j.values.storable.Value[], int> valueSizeCalculator, System.Func<org.neo4j.unsafe.impl.batchimport.input.InputEntity, int> additionalCalculator) throws java.io.IOException
private long[] Sample(IEnumerable <DataFactory> dataFactories, Header.Factory headerFactory, System.Func <Value[], int> valueSizeCalculator, System.Func <InputEntity, int> additionalCalculator)
{
long[] estimates = new long[4]; // [entity count, property count, property size, labels (for nodes only)]
using (CsvInputChunkProxy chunk = new CsvInputChunkProxy())
{
// One group of input files
int groupId = 0;
foreach (DataFactory dataFactory in dataFactories) // one input group
{
groupId++;
Header header = null;
Data data = dataFactory.Create(_config);
RawIterator <CharReadable, IOException> sources = data.Stream();
while (sources.HasNext())
{
using (CharReadable source = sources.Next())
{
if (header == null)
{
// Extract the header from the first file in this group
header = extractHeader(source, headerFactory, _idType, _config, _groups);
}
using (CsvInputIterator iterator = new CsvInputIterator(source, data.Decorator(), header, _config, _idType, EMPTY, extractors(_config), groupId), InputEntity entity = new InputEntity())
{
int entities = 0;
int properties = 0;
int propertySize = 0;
int additional = 0;
while (iterator.Position() < _estimateSampleSize && iterator.Next(chunk))
{
for ( ; chunk.Next(entity); entities++)
{
properties += entity.PropertyCount();
propertySize += calculatePropertySize(entity, valueSizeCalculator);
additional += additionalCalculator(entity);
}
}
if (entities > 0)
{
long entityCountInSource = ( long )((( double )source.Length() / iterator.Position()) * entities);
estimates[0] += entityCountInSource;
estimates[1] += ( long )((( double )properties / entities) * entityCountInSource);
estimates[2] += ( long )((( double )propertySize / entities) * entityCountInSource);
estimates[3] += ( long )((( double )additional / entities) * entityCountInSource);
}
}
}
}
}
}
return(estimates);
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test public void shouldExtractOnlyLine() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
public virtual void ShouldExtractOnlyLine()
{
// given
string firstLine = "characters of only line";
CharReadable reader = Readables.Wrap(firstLine);
// when
char[] firstLineCharacters = Readables.ExtractFirstLineFrom(reader);
int readAfterwards = reader.Read(new char[1], 0, 1);
// then
assertArrayEquals(firstLine.ToCharArray(), firstLineCharacters);
assertEquals(-1, readAfterwards);
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
//ORIGINAL LINE: private void warnAboutDuplicateSourceFiles(java.util.Set<String> seenSourceFiles, Iterable<DataFactory> dataFactories) throws java.io.IOException
private void WarnAboutDuplicateSourceFiles(ISet <string> seenSourceFiles, IEnumerable <DataFactory> dataFactories)
{
foreach (DataFactory dataFactory in dataFactories)
{
RawIterator <CharReadable, IOException> stream = dataFactory.Create(_config).stream();
while (stream.HasNext())
{
using (CharReadable source = stream.Next())
{
WarnAboutDuplicateSourceFiles(seenSourceFiles, source);
}
}
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
//ORIGINAL LINE: private boolean goToNextSource() throws java.io.IOException
private bool GoToNextSource()
{
if (_actual.hasNext())
{
if (_current != null)
{
_previousPosition += _current.position();
}
CloseCurrent();
_current = _actual.next();
return(true);
}
return(false);
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test void shouldFailIfNoNewlineInChunk() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
internal virtual void ShouldFailIfNoNewlineInChunk()
{
// GIVEN
CharReadable reader = Readables.Wrap("1234567\n89012345678901234");
// (next chunks): ^
// (actual chunks): ^
using (ClosestNewLineChunker source = new ClosestNewLineChunker(reader, 12))
{
// WHEN
Source_Chunk chunk = source.NewChunk();
assertTrue(source.NextChunk(chunk));
assertArrayEquals("1234567\n".ToCharArray(), CharactersOf(chunk));
assertThrows(typeof(System.InvalidOperationException), () => assertFalse(source.NextChunk(chunk)));
}
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test public void shouldExtractFirstLine() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
public virtual void ShouldExtractFirstLine()
{
// given
string firstLine = "characters of first line";
string secondLine = "here on the second line";
CharReadable reader = Readables.Wrap(firstLine + "\n" + secondLine);
// when
char[] firstLineCharacters = Readables.ExtractFirstLineFrom(reader);
char[] secondLineCharacters = new char[secondLine.Length];
reader.Read(secondLineCharacters, 0, secondLineCharacters.Length);
// then
assertArrayEquals(firstLine.ToCharArray(), firstLineCharacters);
assertArrayEquals(secondLine.ToCharArray(), secondLineCharacters);
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test void shouldHandleReadAheadEmptyData() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
internal virtual void ShouldHandleReadAheadEmptyData()
{
// GIVEN
TrackingReader actual = new TrackingReader(0);
int bufferSize = 10;
CharReadable aheadReadable = ThreadAheadReadable.ThreadAhead(actual, bufferSize);
// WHEN
actual.AwaitCompletedReadAttempts(1);
// THEN
SectionedCharBuffer buffer = new SectionedCharBuffer(bufferSize);
buffer = aheadReadable.Read(buffer, buffer.Front());
assertEquals(buffer.Pivot(), buffer.Back());
assertEquals(buffer.Pivot(), buffer.Front());
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test void shouldReadAhead() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
internal virtual void ShouldReadAhead()
{
// GIVEN
TrackingReader actual = new TrackingReader(23);
int bufferSize = 5;
CharReadable aheadReader = ThreadAheadReadable.ThreadAhead(actual, bufferSize);
SectionedCharBuffer buffer = new SectionedCharBuffer(bufferSize);
// WHEN starting it up it should read and fill the buffer to the brim
assertEquals(bufferSize, actual.AwaitCompletedReadAttempts(1));
// WHEN we read one buffer
int read = 0;
buffer = aheadReader.Read(buffer, buffer.Front());
AssertBuffer(Chars(read, bufferSize), buffer, 0, bufferSize);
read += buffer.Available();
// and simulate reading all characters, i.e. back section will be empty in the new buffer
buffer = aheadReader.Read(buffer, buffer.Front());
AssertBuffer(Chars(read, bufferSize), buffer, 0, bufferSize);
read += buffer.Available();
// then simulate reading some characters, i.e. back section will contain some characters
int keep = 2;
buffer = aheadReader.Read(buffer, buffer.Front() - keep);
AssertBuffer(Chars(read - keep, bufferSize + keep), buffer, keep, bufferSize);
read += buffer.Available();
keep = 3;
buffer = aheadReader.Read(buffer, buffer.Front() - keep);
AssertBuffer(Chars(read - keep, bufferSize + keep), buffer, keep, bufferSize);
read += buffer.Available();
keep = 1;
buffer = aheadReader.Read(buffer, buffer.Front() - keep);
assertEquals(3, buffer.Available());
AssertBuffer(Chars(read - keep, buffer.Available() + keep), buffer, keep, 3);
read += buffer.Available();
assertEquals(23, read);
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
//ORIGINAL LINE: private void shouldComplyWithSpecifiedCharset(java.nio.charset.Charset charset) throws Exception
private void ShouldComplyWithSpecifiedCharset(Charset charset)
{
// GIVEN
string data = "abcåäö[]{}";
File file = WriteToFile(data, charset);
// WHEN
CharReadable reader = Readables.Files(charset, file);
SectionedCharBuffer buffer = new SectionedCharBuffer(100);
buffer = reader.Read(buffer, buffer.Front());
// THEN
char[] expected = data.ToCharArray();
char[] array = buffer.Array();
assertEquals(expected.Length, buffer.Available());
for (int i = 0; i < expected.Length; i++)
{
assertEquals(expected[i], array[buffer.Pivot() + i]);
}
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test void shouldBackUpChunkToClosestNewline() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
internal virtual void ShouldBackUpChunkToClosestNewline()
{
// GIVEN
CharReadable reader = Readables.Wrap("1234567\n8901234\n5678901234");
// (next chunks): ^ ^
// (actual chunks): ^ ^
using (ClosestNewLineChunker source = new ClosestNewLineChunker(reader, 12))
{
// WHEN
Source_Chunk chunk = source.NewChunk();
assertTrue(source.NextChunk(chunk));
assertArrayEquals("1234567\n".ToCharArray(), CharactersOf(chunk));
assertTrue(source.NextChunk(chunk));
assertArrayEquals("8901234\n".ToCharArray(), CharactersOf(chunk));
assertTrue(source.NextChunk(chunk));
assertArrayEquals("5678901234".ToCharArray(), CharactersOf(chunk));
// THEN
assertFalse(source.NextChunk(chunk));
}
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Test public void shouldTrackPosition() throws Exception
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
public virtual void ShouldTrackPosition()
{
// GIVEN
string data = "1234567890";
// ^ ^
CharReadable reader = Readables.Wrap(data);
SectionedCharBuffer buffer = new SectionedCharBuffer(4);
// WHEN
int expected = 0;
do
{
buffer = reader.Read(buffer, buffer.Front());
expected += buffer.Available();
// THEN
assertEquals(expected, reader.Position());
} while (buffer.HasAvailable());
// and THEN
assertEquals(data.ToCharArray().length, expected);
}
public static CharReadable ThreadAhead(CharReadable actual, int bufferSize)
{
return(new ThreadAheadReadable(actual, bufferSize));
}
public ClosestNewLineChunker(CharReadable reader, int chunkSize) : base(reader, chunkSize)
{
}
internal TrackingReader(int length)
{
this.Bytes = length * 2;
this.Actual = Readables.Wrap(new CharArrayReader(Chars(0, length)), length * 2);
}
public CharReadableChunker(CharReadable reader, int chunkSize)
{
this.Reader = reader;
this.ChunkSize = chunkSize;
this._backBuffer = new char[chunkSize >> 4];
}
public AutoReadingSource(CharReadable reader, SectionedCharBuffer charBuffer)
{
this._reader = reader;
this._charBuffer = charBuffer;
}
public AutoReadingSource(CharReadable reader, int bufferSize) : this(reader, new SectionedCharBuffer(bufferSize))
{
}
/// <summary>
/// Instantiates a <seealso cref="BufferedCharSeeker"/> with optional <seealso cref="ThreadAheadReadable read-ahead"/> capability.
/// </summary>
/// <param name="reader"> the <seealso cref="CharReadable"/> which is the source of data, f.ex. a <seealso cref="System.IO.StreamReader_FileReader"/>. </param>
/// <param name="bufferSize"> buffer size of the seeker and, if enabled, the read-ahead thread. </param>
/// <param name="readAhead"> whether or not to start a <seealso cref="ThreadAheadReadable read-ahead thread"/>
/// which strives towards always keeping one buffer worth of data read and available from I/O when it's
/// time for the <seealso cref="BufferedCharSeeker"/> to read more data. </param>
/// <param name="quotationCharacter"> character to interpret quotation character. </param>
/// <returns> a <seealso cref="CharSeeker"/> with optional <seealso cref="ThreadAheadReadable read-ahead"/> capability. </returns>
//JAVA TO C# CONVERTER WARNING: 'final' parameters are ignored unless the option to convert to C# 7.2 'in' parameters is selected:
//ORIGINAL LINE: public static CharSeeker charSeeker(CharReadable reader, final int bufferSize, boolean readAhead, final char quotationCharacter)
public static CharSeeker CharSeeker(CharReadable reader, int bufferSize, bool readAhead, char quotationCharacter)
{
return(charSeeker(reader, new Configuration_OverriddenAnonymousInnerClass(DEFAULT, bufferSize, quotationCharacter)
, readAhead));
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
//ORIGINAL LINE: private void assertReadText(CharReadable readable, String text) throws java.io.IOException
private void AssertReadText(CharReadable readable, string text)
{
char[] readText = ReadMethod.read(readable, text.ToCharArray().length);
assertArrayEquals(readText, text.ToCharArray());
}
