public KnowledgeGraphNode(int index, OriginalGraphType originalGraphType, string label, string htmlName, Uri linkToPage)
{
this.Index = index;
this.OriginalGraphType = originalGraphType;
this.Label = label;
this.HtmlName = htmlName;
this.LinkToPage = linkToPage;
this.Neighbors = new List <KnowledgeGraphNode>();
}
private static void ExtractAndAddUlSubgraphRecursive(KnowledgeGraph graph, int parentIndex, HtmlNode nodeToParse, ref int nodeToParseIndex, OriginalGraphType originalGraphType)
{
foreach (HtmlNode child in nodeToParse.ChildNodes.Where(n => n.Name == "li"))
{
Uri nodeLinkToPage = GetUriFromNode(child);
// and add it as a parent, too
graph.KnGraph.Add(new KnowledgeGraphNode(index: nodeToParseIndex, originalGraphType: originalGraphType, label: GetLiNodeLabel(child), htmlName: "li", linkToPage: nodeLinkToPage));
// 1 li can only have 1 ul in it
HtmlNode ulNode = child.ChildNodes.Where(node => node.Name == "ul").FirstOrDefault();
// if the li has an ul in it, go into recursive
if (ulNode != null)
{
// add the li node to the graph because it will be the parent of its ul's items and increase the index counter
graph.KnGraph[parentIndex].Neighbors.Add(new KnowledgeGraphNode(index: nodeToParseIndex, originalGraphType: originalGraphType, label: GetLiNodeLabel(child), htmlName: "li", linkToPage: nodeLinkToPage));
int newParentNodeIndex = nodeToParseIndex;
nodeToParseIndex++;
ExtractAndAddUlSubgraphRecursive(
graph: graph,
parentIndex: newParentNodeIndex,
nodeToParse: ulNode,
nodeToParseIndex: ref nodeToParseIndex,
originalGraphType: originalGraphType);
}
else
{
graph.KnGraph[parentIndex].Neighbors.Add(new KnowledgeGraphNode(index: nodeToParseIndex, originalGraphType: originalGraphType, label: GetLiNodeLabel(child), htmlName: "li", linkToPage: nodeLinkToPage));
nodeToParseIndex++;
}
}
}
private static void ParseHtmlNodesIntoKnGraph(KnowledgeGraph graph, List <HtmlNode> htmlNodes, OriginalGraphType originalGraphType = OriginalGraphType.Unknown)
{
if (graph == null)
{
graph = new KnowledgeGraph();
}
// resume node index from the given graph. if empty, it starts at 0
int nodeIndex = graph.KnGraph.Count;
// you can rely that the nodes are in order h2 > h3 > h4 however ul can come at any point
// and there are no duplicate pieces of knowledge (I mean you'll see more h2, h3, h4 and ul but they each correspond to a different piece of knowledge)
// list with the last index of h2, h3, and h4 - compute index of this list by taking the h's number - 2
List <int> mostRecentHIndexes = new List <int>()
{
nodeIndex, nodeIndex, nodeIndex
};
// I need this variable to store the last known h for the ul elements since I can't know which h was laste just with the list
int mostRecentHIndex = nodeIndex;
foreach (var node in htmlNodes)
{
if (node.Name == "ul")
{
// ul has more nodes in it which need to be indexed and the subtree added here
// the ul itself is not a node but a list of nodes, ul is just a placeholder
ExtractAndAddUlSubgraphRecursive(graph: graph, parentIndex: mostRecentHIndex, nodeToParse: node, nodeToParseIndex: ref nodeIndex, originalGraphType: originalGraphType);
}
else if (node.Name.StartsWith("h"))
{
string nodeLabel = GetNodeHeadlineText(node);
Uri nodeLinkToPage = GetUriFromNode(node);
// stop processing nodes once you hit the see also node because we don't want any of the info after see also in the graph
// which is "see also"'s child nodes and the references section
if (string.Equals(nodeLabel, "see also", StringComparison.InvariantCultureIgnoreCase))
{
break;
}
// only add h's and li's to the graph, not ul's
graph.KnGraph.Add(new KnowledgeGraphNode(nodeIndex, originalGraphType, nodeLabel, node.Name, nodeLinkToPage));
mostRecentHIndex = nodeIndex;
// turn the h number into int and -2 to get the index for mostRecentHIndexes
int hIndex = int.Parse(node.Name[1].ToString()) - 2;
mostRecentHIndexes[hIndex] = nodeIndex;
// if it's h3 then add it to the most recent h2 and if it's h4 add it to the most recent h3 and so on if hN add it to h(N-1)
if (hIndex > 0)
{
graph.KnGraph[mostRecentHIndexes[hIndex - 1]].Neighbors.Add(new KnowledgeGraphNode(nodeIndex, originalGraphType, nodeLabel, node.Name, nodeLinkToPage));
}
nodeIndex++;
}
}
}