/// <summary>
/// Read the next character in the stream (more expensive)
/// </summary>
private bool TryReadNextCharInternal(out char ch)
{
while (true)
{
if (this.charsConsumed < this.charsProduced)
{
ch = this.charBuffer[this.charsConsumed];
this.charsConsumed++;
return(true);
}
// adjust stream position to where it should be for reading..
// this may matter
if (this.stream.Position != this.startPosition + this.bytesConverted)
{
this.stream.Position = this.startPosition + this.bytesConverted;
}
this.startPosition = this.stream.Position;
this.bytesRead = stream.Read(this.byteBuffer, 0, this.byteBuffer.Length);
this.charsConsumed = 0;
this.charsProduced = 0;
if (this.bytesRead == 0)
{
ch = '\0';
return(false);
}
int conversionStart = 0;
if (this.adjustFirstBlock)
{
conversionStart = EncodingUtil.GetNextCodeStart(this.byteBuffer, 0, this.bytesRead, this.encoding);
this.startPosition += conversionStart;
this.adjustFirstBlock = false;
}
bool completed;
this.decoder.Convert(this.byteBuffer, conversionStart, this.bytesRead, this.charBuffer, 0, this.charBuffer.Length, true, out this.bytesConverted, out this.charsProduced, out completed);
this.bytesConverted += conversionStart;
}
}
/// <summary>
/// Divides the stream into separate partitions.
/// </summary>
/// <param name="stream">A stream over the entire file.</param>
/// <param name="partitions">The number of similar size partitions requested.</param>
/// <param name="encoding">An optional encoding for the file. If unspecified it will be inferred from the stream itself.</param>
/// <returns>Returns an array of starting positions, one for each partition.</returns>
public static long[] GetPartitionStarts(Stream stream, int partitions, Encoding encoding = null)
{
encoding = encoding ?? GetEncoding(stream);
// give each stream initial starting point.
var size = stream.Length / partitions;
// position each stream at initial computed start position
long[] starts = new long[partitions];
long start = 0;
for (int i = 0; i < partitions; i++, start += size)
{
starts[i] = start + EncodingUtil.GetAlignment(start, encoding);
}
// adjust starting points by finding next actual line start
for (int i = 1; i < partitions; i++)
{
// move ahead if the prior stream is already overlapping initial starting point
if (starts[i - 1] > starts[i])
{
starts[i] = starts[i - 1];
}
stream.Position = starts[i];
var nextLineStart = FindNextLineStart(stream, encoding);
if (nextLineStart > 0)
{
starts[i] = nextLineStart;
}
else
{
starts[i] = stream.Length;
}
}
stream.Position = 0;
return(starts);
}
