public static void Main()
{
ArffParser arffParser = new ArffParser();
ArffRecord arffRecord = arffParser.Parse("../../../ArffParser/TestArffFiles/SmallTest.arff");
Console.WriteLine(arffRecord.ToString());
Console.ReadLine();
}
public void ParseTest_SmallTestArff()
{
var arffParser = new ArffParser.ArffParser();
ArffParser.ArffRecord record = arffParser.Parse("../../TestArffFiles/SmallTest.arff");
Assert.IsTrue(record.RelationName == "cars");
Assert.IsTrue(record.Attributes.Count == 5);
// @ATTRIBUTE buying { vhigh, high, med, low }
Assert.IsTrue(record.Attributes[0].Name == "buying");
Assert.IsTrue(record.Attributes[0].Type == ArffParser.AttributeType.CLASS);
Assert.IsTrue(((NominalAttribute)record.Attributes[0]).Values.Count == 4);
Assert.IsTrue(((NominalAttribute)record.Attributes[0]).Values.Contains("vhigh"));
Assert.IsTrue(((NominalAttribute)record.Attributes[0]).Values.Contains("high"));
Assert.IsTrue(((NominalAttribute)record.Attributes[0]).Values.Contains("med"));
Assert.IsTrue(((NominalAttribute)record.Attributes[0]).Values.Contains("low"));
// @ATTRIBUTE price NUMERIC
Assert.IsTrue(record.Attributes[1].Name == "price");
Assert.IsTrue(record.Attributes[1].Type == ArffParser.AttributeType.NUMERIC);
// @ATTRIBUTE miles numeric
Assert.IsTrue(record.Attributes[2].Name == "miles");
Assert.IsTrue(record.Attributes[2].Type == ArffParser.AttributeType.NUMERIC);
// @ATTRIBUTE make STRING
Assert.IsTrue(record.Attributes[3].Name == "make");
Assert.IsTrue(record.Attributes[3].Type == ArffParser.AttributeType.STRING);
// @ATTRIBUTE purchaseDate STRING
Assert.IsTrue(record.Attributes[4].Name == "purchaseDate");
Assert.IsTrue(record.Attributes[4].Type == ArffParser.AttributeType.DATE);
// @DATA
Assert.IsTrue(record.Data.Count == 3);
}
// sets up a state machine and parses according to the expected structure of ARFF
// structure is:
// @relation <identifier>
//
// @attribute <identifier>
// <more attributes ...>
//
// @data
// <data entries ...>
public ArffRecord Parse(string filename)
{
_curParseFileName = filename;
_curParseLineNumber = 0;
string[] rawFileLines = System.IO.File.ReadAllLines(filename);
string[] fileLines = PreprocessFileLines(rawFileLines);
ArffRecord arffRecord = new ArffRecord();
arffRecord.FileName = filename;
ArffState parseState = ArffState.HEADER;
for (_curParseLineNumber = 1; _curParseLineNumber <= fileLines.Length; _curParseLineNumber++)
{
string fileLine = fileLines[_curParseLineNumber - 1];
if (fileLine.Length == 0 || fileLine[0] == COMMENT_TOKEN)
{
continue; // skip comment lines and new lines
}
fileLine = fileLine.Split(COMMENT_TOKEN)[0].Trim(); // remove any trailing comments
switch (parseState)
{
case ArffState.HEADER:
if (fileLine.Contains(RELATION_TOKEN))
{
arffRecord.RelationName = TryParseRelation(fileLine);
parseState = ArffState.ATTRIBUTE;
}
else
{
// non empty line present in header state that did not contain relation token
throw new ArffParseException(_curParseFileName, _curParseLineNumber, "expected @relation declaration");
}
break;
case ArffState.ATTRIBUTE:
if (fileLine.Contains(ATTRIBUTE_TOKEN))
{
var attribute = TryParseAttribute(fileLine);
arffRecord.Attributes.Add(attribute);
}
else if (fileLine.Contains(DATA_TOKEN))
{
parseState = ArffState.DATA;
}
else
{
throw new ArffParseException(_curParseFileName, _curParseLineNumber, "expected @attribute declaration");
}
break;
case ArffState.DATA:
var data = TryParseData(fileLine, arffRecord.Attributes);
arffRecord.Data.Add(data);
break;
}
}
return(arffRecord);
}
