public SimilaritiesReport FindForIncident(int incidentId)
{
using (var cmd = _uow.CreateCommand())
{
cmd.CommandText =
@"select Id, Name, Properties from IncidentContextCollections
where IncidentId = @incidentId";
cmd.AddParameter("incidentId", incidentId);
var collections = new List <SimilarityCollection>();
using (var reader = cmd.ExecuteReader())
{
while (reader.Read())
{
var json = (string)reader["Properties"];
var properties = CoderrDtoSerializer.Deserialize <ContextCollectionPropertyDbEntity[]>(json);
var col = new SimilarityCollection(incidentId, reader.GetString(1));
col.GetType().GetProperty("Id").SetValue(col, reader.GetInt32(0));
foreach (var entity in properties)
{
var prop = new Similarity(entity.Name);
prop.LoadValues(
entity.Values.Select(x => new SimilarityValue(x.Value, x.Percentage, x.Count)).ToArray());
col.Properties.Add(prop);
}
collections.Add(col);
}
}
return(collections.Count == 0 ? null : new SimilaritiesReport(incidentId, collections));
}
}
public static Dictionary <List <HtmlNode>, double> GetSimilarTrees(HtmlNode rootNode, int level, double minSimilarity = 0.5)
{
var groups = new List <SimilarityCollection>();
var lastNodeTrees = HtmlHelpers.GetMaxDepthNodes(rootNode, level)
.Select(GenerateTree);
foreach (var nodeTree in lastNodeTrees)
{
var found = false;
foreach (var group in groups)
{
var c = nodeTree.CompareTrees(group.Value[0]);
if (c < minSimilarity)
{
continue;
}
group.Value.Add(nodeTree);
group.SumSimilarity += c;
found = true;
}
if (found)
{
continue;
}
var newPair = new SimilarityCollection();
newPair.Value.Add(nodeTree);
groups.Add(newPair);
}
return(groups
.Where(p => p.Value.Count > 1)
.ToDictionary(
pair => pair.Value.Select(x => x.OrigNode).ToList(),
pair => pair.Average
));
}
