public TagCloudNode GetCommentTreeString(children children)
{
Dictionary <string, int> topNWordsRoot = children.GetTopNWordsDictionary(10);
TagCloudNode tgnRoot = new TagCloudNode();
tgnRoot.id = children.id;
tgnRoot.key = children.key;
tgnRoot.text = children.text;
bool isOP = author == children.author;
string citePre = isOP ? "<cite class='op'>" : "<cite>";
string citePost = ":</cite>";
tgnRoot.title = citePre + children.author + citePost + children.text;
GetTagCloudFromDictionary(topNWordsRoot);
tgnRoot.children = new List <TagCloudNode>();
children.Children = children.Children.OrderByDescending(x => x.created_at).ToList();
foreach (children child in children.Children)
{
if (!string.IsNullOrWhiteSpace(child.SubtreeText))
{
TagCloudNode tgnChild = GetCommentTreeString(child);
tgnRoot.children.Add(tgnChild);
}
}
return(tgnRoot);
}
void GetCommentDictionary(children childlist, ref Dictionary <int, string> dict)
{
foreach (children child in childlist.Children)
{
dict[child.id] = child.text;
GetCommentDictionary(child, ref dict);
}
}
private void GetChildIDHashSetList(children node)
{
node.ChildIDList = GetChildIDList(node);
foreach (children child in node.Children)
{
GetChildIDHashSetList(child);
}
}
private void PopulateNodeList(children child)
{
NodeList.Add(child);
foreach (children childnode in child.Children)
{
PopulateNodeList(childnode);
}
}
private string GetSubtree(children node)
{
string subtreeText = WebUtility.HtmlDecode(node.text);
foreach (children child in node.Children)
{
subtreeText += " " + WebUtility.HtmlDecode(GetSubtree(child));
}
return(subtreeText);
}
public int GetChildCount(children childrenlist)
{
int counter = string.IsNullOrWhiteSpace(childrenlist.text)?0:1;
foreach (children child in childrenlist.Children)
{
counter += GetChildCount(child);
}
return(counter);
}
public List <string> GetAnchorWords(children root, int N)
{
List <string> anchorWords = new List <string>();
string[] allWords = GetAllWords(root.SubtreeText);
Dictionary <string, string> stemParentDictionary = GetStemParentDictionary(allWords);
children rootNode = new children();
List <HashSet <int> > rootChildIDs = new List <HashSet <int> >();
foreach (children child in root.Children)
{
GetChildIDHashSetList(child);
HashSet <int> currChildIDs = new HashSet <int>();
currChildIDs.Add(child.id);
foreach (var item in child.ChildIDList)
{
currChildIDs.UnionWith(item);
}
rootChildIDs.Add(currChildIDs);
}
rootNode.ChildIDList = rootChildIDs;
NodeList = new List <children>();
NodeList.Add(rootNode);
foreach (children child in root.Children)
{
PopulateNodeList(child);
}
Dictionary <string, HashSet <int> > wordIDMapping = GetWordIDMapping();
//Dictionary<string, double> WordTreeScore = new Dictionary<string, double>();
Dictionary <string, List <children> > WordLCAList = new Dictionary <string, List <children> >();
foreach (var kvp in wordIDMapping)
{
List <children> currLCAList = new List <children>();
int numLCAs = 0;
foreach (children node in NodeList)
{
int numBranchesWithWord = 0;
foreach (var childIDBranch in node.ChildIDList)
{
if (childIDBranch.Intersect(kvp.Value).Count() > 0)
{
numBranchesWithWord += 1;
}
}
if ((numBranchesWithWord == 1 && node.ChildIDList.Count == 1) || numBranchesWithWord > 1)
{
currLCAList.Add(node);
}
}
WordLCAList[stemParentDictionary.ContainsKey(kvp.Key) ? stemParentDictionary[kvp.Key] : kvp.Key] = currLCAList;
}
anchorWords = WordLCAList.OrderByDescending(x => x.Value.Count).Select(x => x.Key).Where(y => CommonWords.GetFrequency(y) < 20).Take(N).ToList();
return(anchorWords);
}
private List <HashSet <int> > GetChildIDList(children node)
{
List <HashSet <int> > myChildren = new List <HashSet <int> >();
foreach (children child in node.Children)
{
HashSet <int> idList = GetChildIDs(child);
myChildren.Add(idList);
}
return(myChildren);
}
public async Task <object> GetCommentTreeForId(string idTuple)
{
string sStoryId = idTuple.Split(':')[1];
string sNodeId = idTuple.Split(':')[0];
int storyid = Convert.ToInt32(sStoryId);
int nodeid = Convert.ToInt32(sNodeId);
FullStory fs = FullStoryFactory.GetFullStory(storyid);
children child = fs.GetNodeById(nodeid);
TagCloudNode tgn = fs.GetCommentTreeString(child);
return(JsonConvert.SerializeObject(tgn));
}
private HashSet <int> GetChildIDs(children childnode)
{
HashSet <int> childIDs = new HashSet <int>();
childIDs.Add(childnode.id);
foreach (children child in childnode.Children)
{
childIDs.Add(child.id);
childIDs.UnionWith(GetChildIDs(child));
}
return(childIDs);
}
void LoadUserComments(children child)
{
if (!string.IsNullOrWhiteSpace(child.author))
{
if (!UserComments.ContainsKey(child.author))
{
UserComments[child.author] = new List <CommentObj>();
}
UserComments[child.author].Add(new CommentObj()
{
Id = child.id, Text = child.text
});
}
foreach (children childNode in child.Children)
{
LoadUserComments(childNode);
}
}
Dictionary <string, HashSet <int> > GetWordIDMapping(children child)
{
Dictionary <string, HashSet <int> > wordIDMapping = new Dictionary <string, HashSet <int> >();
string[] allWords = GetAllWords(child.text);
foreach (string word in allWords)
{
if (stopWords.Contains(word.ToLower()))
{
continue;
}
if (word.Length < 3 && word.Any(c => char.IsLower(c)))
{
continue;
}
string stem = Stemmer.GetStem(word);
if (!wordIDMapping.ContainsKey(stem))
{
wordIDMapping[stem] = new HashSet <int>();
}
wordIDMapping[stem].Add(child.id);
}
foreach (children childitem in child.Children)
{
Dictionary <string, HashSet <int> > mapping = GetWordIDMapping(childitem);
foreach (var kvp in mapping)
{
if (wordIDMapping.ContainsKey(kvp.Key))
{
wordIDMapping[kvp.Key].UnionWith(kvp.Value);
}
else
{
wordIDMapping[kvp.Key] = kvp.Value;
}
}
}
return(wordIDMapping);
}
public TagCloudNode GetTagCloudTreeString(children children)
{
Dictionary <string, int> topNWordsRoot = children.GetTopNWordsDictionary(10);
string topNWords = GetTagCloudFromDictionary(topNWordsRoot);
TagCloudNode tgnRoot = new TagCloudNode();
tgnRoot.id = children.id;
tgnRoot.key = children.key;
tgnRoot.text = topNWords;
tgnRoot.title = topNWords;
tgnRoot.children = new List <TagCloudNode>();
children.Children = children.Children.OrderByDescending(x => x.created_at).ToList();
foreach (children child in children.Children)
{
if (!string.IsNullOrWhiteSpace(child.SubtreeText))
{
TagCloudNode tgnChild = GetTagCloudTreeString(child);
tgnRoot.children.Add(tgnChild);
}
}
return(tgnRoot);
}
public Dictionary <string, List <CommentObj> > GetNamedObjects(int N)
{
StringBuilder sbAllWords = new StringBuilder();
foreach (children child in children)
{
sbAllWords.Append(child.SubtreeText);
sbAllWords.Append(" ");
}
string[] allWords = GetAllWords(sbAllWords.ToString());
Dictionary <string, string> stemParentDictionary = GetStemParentDictionary(allWords);
List <string> namedObjects = new List <string>();
children rootNode = new children();
List <HashSet <int> > rootChildIDs = new List <HashSet <int> >();
foreach (children child in children)
{
GetChildIDHashSetList(child);
HashSet <int> currChildIDs = new HashSet <int>();
currChildIDs.Add(child.id);
foreach (var item in child.ChildIDList)
{
currChildIDs.UnionWith(item);
}
rootChildIDs.Add(currChildIDs);
}
rootNode.ChildIDList = rootChildIDs;
NodeList = new List <children>();
NodeList.Add(rootNode);
foreach (children child in children)
{
PopulateNodeList(child);
}
Dictionary <string, HashSet <int> > wordIDMapping = GetWordIDMapping();
//Dictionary<string, double> WordTreeScore = new Dictionary<string, double>();
Dictionary <string, List <children> > WordLCAList = new Dictionary <string, List <children> >();
foreach (var kvp in wordIDMapping)
{
List <children> currLCAList = new List <children>();
int numLCAs = 0;
foreach (children node in NodeList)
{
int numBranchesWithWord = 0;
foreach (var childIDBranch in node.ChildIDList)
{
if (childIDBranch.Intersect(kvp.Value).Count() > 0)
{
numBranchesWithWord += 1;
}
}
if ((numBranchesWithWord == 1 && node.ChildIDList.Count == 1) || numBranchesWithWord > 1)
{
currLCAList.Add(node);
}
}
WordLCAList[stemParentDictionary.ContainsKey(kvp.Key) ? stemParentDictionary[kvp.Key] : kvp.Key] = currLCAList;
}
namedObjects = WordLCAList
.OrderByDescending(x => x.Value.Count)
.Select(x => x.Key)
.Where(y => CommonWords.GetFrequency(y) < 1)
.Where(a => char.IsUpper(a[0]))
.Where(b => b.Length > 1)
.Where(z => !(z.EndsWith("n't") || z.EndsWith("'m") || (z.EndsWith("'ll")) || (z.EndsWith("'d")) || z.EndsWith("'ve") || z.EndsWith("'re") || z.EndsWith("'s")))
.Take(N)
.ToList();
//namedObjects.Sort();
Dictionary <string, List <CommentObj> > namedObjectDictionary = new Dictionary <string, List <CommentObj> >();
foreach (string namedObject in namedObjects)
{
List <CommentObj> commentObjsForWord = new List <CommentObj>();
string stem = Stemmer.GetStem(namedObject);
HashSet <int> idsWithWord = wordIDMapping[stem];
foreach (int id in idsWithWord)
{
children child = GetNodeById(id);
CommentObj commentObj = new CommentObj()
{
Id = id, Text = child.text
};
commentObjsForWord.Add(commentObj);
}
namedObjectDictionary[namedObject] = commentObjsForWord;
}
var ordered = namedObjectDictionary.Keys.OrderByDescending(x => namedObjectDictionary[x].Count).ToList().ToDictionary(x => x, x => namedObjectDictionary[x]);
return(ordered);
}
