public void Index(IRecord record)
{
var doc = new Document();
foreach (string propname in record.GetProperties())
{
Property prop = _config.GetPropertyByName(propname);
if (prop == null)
{
throw new Exception(String.Format("Record has property {0} for which there is no configuration.",
propname));
}
Field.Index ix; //TODO: could cache this. or get it from property
ix = prop.IsIdProperty ? Field.Index.NOT_ANALYZED : Field.Index.ANALYZED;
foreach (string v in record.GetValues(propname))
{
if (v.Equals(""))
{
continue; //FIXME: not sure if this is necessary
}
doc.Add(new Field(propname, v, Field.Store.YES, ix));
}
}
try
{
_iwriter.AddDocument(doc);
}
catch (Exception ex)
{
logger.Error("Error adding document to index writer: {0}", ex.Message);
}
}
public double Compare(IRecord r1, IRecord r2)
{
double prob = 0.5;
foreach (string propname in r1.GetProperties())
{
Property prop = _config.GetPropertyByName(propname);
if (prop.IsIdProperty || prop.IsIgnoreProperty())
{
continue;
}
List <string> vs1 = r1.GetValues(propname);
List <string> vs2 = r2.GetValues(propname);
if ((vs1.Count == 0) || (vs2.Count == 0))
{
continue; // no values to compare, so skip
}
double high = 0.0;
foreach (string v1 in vs1)
{
if (v1.Equals("")) //TODO: These values shouldn't be here at all.
{
continue;
}
foreach (string v2 in vs2)
{
if (v2.Equals("")) //TODO: These values shouldn't be here at all.
{
continue;
}
try
{
double p = prop.Compare(v1, v2);
high = Math.Max(high, p);
}
catch (Exception e)
{
throw new DukeException(String.Format("Comparison of values {0} and {1} failed. {2}", v1, v2,
e.Message));
}
}
}
prob = StandardUtils.ComputeBayes(prob, high);
}
return(prob);
}
