public List<Tuple<string, double>> ParseNGrams(string xmlData, bool sortByFrequency)
{
HtmlAgilityPack.XmlDocument xmlDoc = new HtmlAgilityPack.XmlDocument();
xmlDoc.LoadXml(xmlData);
List<Tuple<string, double>> ngrams = new List<Tuple<string, double>>();
HtmlAgilityPack.HtmlNodeCollection ngramNodes = xmlDoc.DocumentNode.SelectNodes("/results/ngrams//ngram");
if (ngramNodes == null) return ngrams;
foreach (var ngramNode in ngramNodes)
{
string text = GenLib.Text.Text.DecodeXMLString(ngramNode.GetAttributeValue("text", "").ToLower());
if (string.IsNullOrEmpty(text))
continue;
string fqStr = ngramNode.GetAttributeValue("fq", "0");
int fq = int.Parse(fqStr);
ngrams.Add(new Tuple<string, double>(text, fq));
}
if (sortByFrequency)
ngrams.Sort((x, y) =>
{
if (x.Item2 > y.Item2) return -1;
if (x.Item2 < y.Item2) return 1;
return 0;
});
return ngrams;
}
void KEUIResponse_OnMessageReceived(string message)
{
try
{
HtmlAgilityPack.XmlDocument eventDoc = new HtmlAgilityPack.XmlDocument();
eventDoc.LoadXml(message);
var sequenceIdNode = eventDoc.DocumentNode.SelectSingleNode("//ns1:head/ns1:sequencenumber");
string sequenceId = sequenceIdNode.InnerText;
Trace.WriteLine("Recieved response to " + sequenceId);
var responseNode = eventDoc.DocumentNode.SelectSingleNode("//ns1:eventData/s1:keuiResponse");
string responseData = responseNode.InnerHtml;
var requestNode = eventDoc.DocumentNode.SelectSingleNode("//ns1:eventData/s1:keuiRequest");
var requestTypeNode = eventDoc.DocumentNode.SelectSingleNode("//ns1:eventData/s1:keuiRequest/s1:requestType");
var requestDataNode = eventDoc.DocumentNode.SelectSingleNode("//ns1:eventData/s1:keuiRequest/s1:requestData");
string requestType = requestTypeNode.InnerText;
if (requestType == "GetTagInfo")
{
var paramNode = requestDataNode.SelectSingleNode("./params");
int startIndex = paramNode.GetAttributeValue("startIndex", 0);
int count = paramNode.GetAttributeValue("count", 1000);
HtmlAgilityPack.HtmlNodeCollection tags = responseNode.SelectNodes("./TagInfoBase") ?? new HtmlAgilityPack.HtmlNodeCollection(null);
foreach (var tag in tags)
{
string tagData = tag.OuterHtml;
TagInfoBase newTag = TagInfoBase.FromXML(tagData);
_tagIdToTagInfo[newTag.TagId] = newTag;
if (!string.IsNullOrEmpty(newTag.TagIdStr))
_tagIdStrToTagInfo[newTag.TagIdStr] = newTag;
if (!_tagIdToChildrenInfo.ContainsKey(newTag.ParentTagId))
_tagIdToChildrenInfo[newTag.ParentTagId] = new List<TagInfoBase>();
_tagIdToChildrenInfo[newTag.ParentTagId].Add(newTag);
}
if (tags.Count() == count)
{
string newSequenceId;
string request = Defaults.BuildRequest(PublisherName, "GetTagInfo", String.Format("<params startIndex=\"{0}\" count=\"{1}\" />", startIndex + count, count), out newSequenceId);
//this.BeginInvoke((Action)(() => { SendRequest(request, newSequenceId, "GetTagInfo"); }));
SendRequest(request, newSequenceId, "GetTagInfo");
}
Invoke((Action)(() =>
{
if (tags.Count == count)
StatusText.Text = String.Format("Loading tags... ({0} tags loaded)", _tagIdToTagInfo.Count);
else
StatusText.Text = "Tags successfully loaded.";
}));
}
else if (requestType == "GetAnnotationOntologyRDF")
{
LoadOntology(responseData);
}
else if (requestType == "GetSimilarConcepts")
{
foreach (HtmlAgilityPack.HtmlNode conceptNode in responseNode.SelectNodes(".//item") ?? new HtmlAgilityPack.HtmlNodeCollection(null))
{
string uri = conceptNode.GetAttributeValue("uri", "");
string label = conceptNode.GetAttributeValue("label", "");
AddRelatedConcept(label, uri, _conceptRelations[0]);
}
}
else if (requestType == "ExecuteCommand")
Response_ExecuteCommand(responseData, sequenceId, message);
else if (requestType == "SetData")
Response_SetData(responseData, sequenceId, message);
else if (requestType == "Query")
Response_Query(responseData, sequenceId, message);
else
MessageBox.Show("Don't know how to process request type " + requestType + ".");
// remove the id of the processed event
if (_sequenceIdToType.ContainsKey(sequenceId))
_sequenceIdToType.Remove(sequenceId);
}
catch (Exception ex)
{
Trace.WriteLine("Exception while processing response: " + ex.Message);
}
}
public List<Tuple<string, double>> ParseKeywords(string xmlData)
{
HtmlAgilityPack.XmlDocument xmlDoc = new HtmlAgilityPack.XmlDocument();
xmlDoc.LoadXml(xmlData);
List<Tuple<string, double>> keywords = new List<Tuple<string, double>>();
HtmlAgilityPack.HtmlNodeCollection kwNodes = xmlDoc.DocumentNode.SelectNodes("/results/keywords//kw");
if (kwNodes == null) return keywords;
foreach (var kwNode in kwNodes)
{
string text = GenLib.Text.Text.DecodeXMLString(kwNode.GetAttributeValue("str", "").ToLower());
text = text.Replace("-", "").Replace("_", "");
if (string.IsNullOrEmpty(text))
continue;
if (text.Contains(':'))
text = text.Substring(text.IndexOf(':') + 1);
string wgtStr = kwNode.GetAttributeValue("wgt", "0.0");
double wgt = double.Parse(wgtStr, System.Globalization.NumberFormatInfo.InvariantInfo);
bool existed = false;
for (int i = 0; i < keywords.Count; i++)
{
var kw = keywords[i];
if (kw.Item1 == text)
{
if (kw.Item2 < wgt)
keywords[i] = new Tuple<string, double>(text, wgt);
existed = true;
}
}
if (!existed)
keywords.Add(new Tuple<string, double>(text, wgt));
}
return keywords;
}
