public void TestProbabilisticSqlInsert()
{
string sentences = "INSERT INTO socialData VALUES (351,PROBABLY 785 25% ,Smith,Single) probably 50%";
SqlInsertQuery query = new SqlInsertQuery(sentences);
query.ProcessAndPopulateEachField();
Assert.IsTrue(query.TupleP == 50.0);
Assert.IsTrue(query.TableName == "socialData");
Assert.IsTrue(query.Attributes.Count == 4);
ProbabilisticSingleAttribute p = (ProbabilisticSingleAttribute)query.Attributes[1];
Assert.IsTrue(p.Values[0] == "785");
Assert.IsTrue(p.Probs[0] == 25.0);
}
public static ProbabilisticAttribute processProbabilisticValueClause(string probabilisticValues)
{
// pattern here is: Word DecimalNumber% / Word DecimalNumber% / ... (repeats)
string sPattern = @"\s*((?<value>[\w\s\.]+)|(?<values>\[.+?\]))\s+(?<prob>\d+)%\s*";
MatchCollection matchs = Regex.Matches(probabilisticValues, sPattern, RegexOptions.IgnoreCase);
if (matchs.Count == 0)
{
throw new Exception("no matching on Probabilistic Value Clause");
}
List <String> values = new List <String>();
List <Double> pValues = new List <Double>();
List <List <string> > multiAttrbutes = new List <List <string> >();
for (int i = 0; i < matchs.Count; i++)
{
String value = matchs[i].Groups["value"].Value;
String valuesCapture = matchs[i].Groups["values"].Value;
String probability = matchs[i].Groups["prob"].Value;
double attributeP = 0;
bool isSuccess = Double.TryParse(probability, out attributeP);
if (!isSuccess)
{
throw new Exception("probability value invalid");
}
pValues.Add(attributeP);
if (!string.IsNullOrEmpty(valuesCapture))
{
List <String> ListOfValue = ProcessListOfValues(valuesCapture);
multiAttrbutes.Add(ListOfValue);
}
else if (!string.IsNullOrEmpty(value))
{
values.Add(value);
}
else
{
throw new Exception("both value and values are empty !");
}
}
if (values.Count > 0 && multiAttrbutes.Count == 0)
{
var pSingleAttribute = new ProbabilisticSingleAttribute(values, pValues);
return(pSingleAttribute);
}
if (multiAttrbutes.Count > 0 && values.Count == 0)
{
bool consistency = true;
int multiAttributesSize = multiAttrbutes[0].Count;
foreach (var multiAttrbute in multiAttrbutes)
{
if (multiAttributesSize != multiAttrbute.Count)
{
// multi attributes sould be of the same size.
consistency = false;
}
}
if (consistency)
{
var multiAttribute = new ProbabilisticMultiAttribute(multiAttrbutes, pValues);
return(multiAttribute);
}
}
throw new Exception("fail to parse probabilistic multi attributes");
}
