private string IndentationFor(HtmlNode node)
{
int length = node.Ancestors("ol").Count() + node.Ancestors("ul").Count();
return node.ParentNode.Name.ToLowerInvariant() == "li" && node.ParentNode.FirstChild != node
? new string(' ', length * 4)
: Environment.NewLine + Environment.NewLine;
}
private string AnalyzeNode(HtmlNode node)
{
string curText = node.InnerText;
int wordLength = Extractor.WordLength(curText);
int wordCount = Extractor.WordCount(curText);
int upperCaseCount = Extractor.UpperCaseCount(curText);
bool finishedSentence = Extractor.ContainsPunctuation(curText);
if (wordCount == 0)
{
return "";
}
bool isLink = node.Ancestors("a").ToList().Count > 0;
double linkRatio = 0;
int averageSentenceLength = Extractor.SentenceLength(curText);
int nrOfSentences = Extractor.NrOfSentences(curText);
if (isLink)
{
HtmlNode parentNode = node.ParentNode;
string clazz = parentNode.Attributes["class"] == null ? null : parentNode.Attributes["class"].Value;
while (parentNode.Name == "a" || (clazz != null && clazz.IndexOf("x-nc-sel") != -1))
{
parentNode = parentNode.ParentNode;
clazz = parentNode.Attributes["class"] == null ? null : parentNode.Attributes["class"].Value;
}
string parentText = parentNode.InnerText;
int totalWords = Extractor.WordCount(parentText);
linkRatio = ((double)wordCount) / ((double)totalWords);
}
return String.Format("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\t{7}", wordLength, wordCount,
nrOfSentences, averageSentenceLength, isLink ? "Yes" : "No", linkRatio.ToString("##0.00###", CultureInfo.CreateSpecificCulture("en-GB")), upperCaseCount, finishedSentence ? "Yes" : "No" );
}
// try to find the city of hall by xPath
private static string TryToGetCity(HtmlNode htmlNode)
{
var td = htmlNode.Ancestors("td").Where(t => t.GetAttributeValue("hallslistwrap", "") != null).ToList();
if (td.First() == null) return null;
const string xPath1 = @"../../../../preceding::tr[1]//h5"; // xpath to City Name
var h5 = td.First().SelectSingleNode(xPath1);
return h5?.InnerHtml;
}
private string IndentationFor(HtmlNode node)
{
int length = node.Ancestors("ol").Count() + node.Ancestors("ul").Count();
return new string(' ', Math.Max(length-1,0));
}
private bool AlreadyItalic(HtmlNode node)
{
return node.Ancestors("i").Count() > 0 || node.Ancestors("em").Count() > 0;
}
private void ProcessNode(HtmlNode node)
{
Node n = new Node();
n.text = node.InnerText;
n.wordCount = WordCount(node.InnerText);
HtmlNodeCollection linkNodes = node.SelectNodes("../a");
List<HtmlNode> ancestors = node.Ancestors("a").ToList();
n.isLink = ancestors.Count > 0; // represents the element itself being inside some hierarchy that contains some number of a tags
double linkCount = n.isLink ? 1 : 0;
if (linkNodes != null)
{
linkCount += linkNodes.Count;
}
n.linkRatio = linkCount / n.wordCount;
nodes.Add(n);
}
private bool AlreadyBold(HtmlNode node)
{
return node.Ancestors("strong").Count() > 0 || node.Ancestors("b").Count() > 0;
}
