protected ColRegExpExtractProcessingResult ExtractRegExpValues(ColRegExp regExp, double documentID, string text)
{
ColRegExpExtractProcessingResult result = null;
try
{
var matches = regExp.GetFilteredMatches(text);
var matchValues = matches.Select(match => match.Value)
.ToList();
if (matchValues.Any())
{
var returnValue = String.Empty;
switch (regExp.ExtractOptions.InstanceNo)
{
case 1:
returnValue = matchValues[0];
break;
case 2:
returnValue = matchValues[matches.Count - 1];
break;
case 3:
returnValue = matchValues.ElementAtOrDefault(regExp.ExtractOptions.NthInstaceNumber.Value - 1);
break;
case 4:
returnValue = matchValues.Count > 1 ? Newtonsoft.Json.JsonConvert.SerializeObject(matchValues.ToArray()) : matchValues.First();
break;
}
///////////////////////////////////////////////////////////////////////////////
if (!String.IsNullOrEmpty(returnValue))
{
result = new ColRegExpExtractProcessingResult()
{
DocumentID = documentID,
ColumnID = regExp.ColumnID,
Value = returnValue,
ExtractOptions = regExp.ExtractOptions
};
}
}
}
catch (Exception ex)
{
Logger.HandleException(ex);
}
return(result);
}
protected void CalcRegExpScore(ColRegExp regExp, string text, int score)
{
try
{
var matches = regExp.GetFilteredMatches(text);
if (matches.Any())
{
regExp.IncrementTotalDocuments();
if (score > 0)
{
regExp.IncrementPositiveDocuments();
}
regExp.AddTotalMatches(matches.Count);
}
}
catch (Exception ex)
{
Logger.HandleException(ex);
}
}