/// <summary>
/// Write out dataset to the output stream
/// </summary>
/// <param name="dataSet">Dataset to write</param>
/// <param name="compression">Compression method</param>
/// <param name="append">When true, appends to the file, otherwise creates a new file.</param>
public void Write(DataSet dataSet, CompressionMethod compression = CompressionMethod.Gzip, bool append = false)
{
PrepareFile(dataSet, append);
int offset = 0;
int count;
do
{
count = Math.Min(_writerOptions.RowGroupsSize, dataSet.Count - offset);
Thrift.RowGroup rg = _meta.AddRowGroup();
long rgStartPos = Stream.Position;
rg.Columns = new List <Thrift.ColumnChunk>();
foreach (SchemaElement se in dataSet.Schema.Flatten())
{
var cw = new ColumnWriter(Stream, ThriftStream, _meta, se, compression, _formatOptions, _writerOptions);
IList values = dataSet.GetColumn(se, offset, count);
Thrift.ColumnChunk chunk = cw.Write(offset, count, values);
rg.Columns.Add(chunk);
}
//row group's size is a sum of _uncompressed_ sizes of all columns in it, including the headers
//luckily ColumnChunk already contains sizes of page+header in it's meta
rg.Total_byte_size = rg.Columns.Sum(c => c.Meta_data.Total_compressed_size);
rg.Num_rows = count;
offset += _writerOptions.RowGroupsSize;
}while (offset < dataSet.Count);
_dataWritten = true;
}
public void Can_traverse_streamed()
{
var model = new BagOfCharsModel();
var tree = model.CreateTree(model, _data);
using (var indexStream = new MemoryStream())
using (var vectorStream = new MemoryStream())
using (var pageStream = new MemoryStream())
{
using (var writer = new ColumnWriter(indexStream, keepStreamOpen: true))
{
writer.CreatePage(tree, vectorStream, new PageIndexWriter(pageStream, keepStreamOpen: true));
}
pageStream.Position = 0;
Assert.DoesNotThrow(() =>
{
using (var reader = new ColumnReader(new PageIndexReader(pageStream), indexStream, vectorStream, _sessionFactory, _loggerFactory.CreateLogger <ColumnReader>()))
{
foreach (var word in _data)
{
foreach (var queryVector in model.Tokenize(word))
{
var hit = reader.ClosestMatch(queryVector, model);
if (hit == null)
{
throw new Exception($"unable to find {word} in tree.");
}
if (hit.Score < model.IdenticalAngle)
{
throw new Exception($"unable to score {word}.");
}
Debug.WriteLine($"{word} matched vector in disk with {hit.Score * 100}% certainty.");
}
}
}
});
}
}
private static void Main(string[] args)
{
string path;
while ((path = CaptureFeedPath(args)) == null)
{
}
const char delimiter = ',';
var stopwatch = new Stopwatch();
stopwatch.Start();
var parser = new DelimiterSeparatedFieldParser(delimiter);
TriangleDimensions largestTriangle;
using( var inputFileReader = new StreamReader(path) )
{
var collector = new ErrorCollector();
WriteProgress("Processing file (this may take some time)...");
var columnReader = new ColumnReader(inputFileReader, parser);
var dimensionReader = new Reader<TriangleFeedYearExtract>(columnReader, collector);
var largestTriangleCalculator = new LargestTriangleCalculator(dimensionReader.Read);
largestTriangle = largestTriangleCalculator.Calculate();
ReportAndQuitIfErrors(collector, path, "triangle_dimension_calculation_errors.txt");
}
using( var inputFileReader = new StreamReader(path) )
{
var collector = new ErrorCollector();
using( var outputFileWriter = new StreamWriter(BuildPathFromInputPath(path, "output.csv")) )
{
var columnReader = new ColumnReader(inputFileReader, parser);
var paymentRecordReader = new Reader<TriangleFeedFullDataExtract>(columnReader, collector);
var triangleBuilder = new TriangleBuilder(paymentRecordReader, collector);
Func<ClaimTriangle, string[]> triangleConverter = triangle => triangle.Accumulate().Flatten(largestTriangle);
var header = string.Format("{0}, {1}", largestTriangle.OriginYear, largestTriangle.DevelopmentYears);
var writer = new ColumnWriter<ClaimTriangle>(triangleBuilder.BuildNext, triangleConverter, header, outputFileWriter, delimiter);
writer.Write();
}
ReportAndQuitIfErrors(collector, path, "payment_record_errors.txt");
}
stopwatch.Stop();
WriteProgress("Processing took: {0}", stopwatch.Elapsed);
System.Console.WriteLine("Press any key to quit...");
System.Console.ReadKey();
}
public Array OnColumnWriter <TValue>(ColumnWriter <TValue> columnWriter)
where TValue : unmanaged
{
columnWriter.WriteBatch(_values.Length, (TValue[])_values);
return(_values);
}
