private void AddLink(LinkItem linkItem)
{
string url = LinkExtractor.NormalizeUrl(linkItem.Url);
if (url != null)
{
linkItem.Url = url;
parsedHtml.Links.Add(linkItem);
}
}
public static void Process(string directory)
{
DirectoryInfo folder = new DirectoryInfo(directory);
FileInfo[] files = folder.GetFiles("*.htm");
foreach (FileInfo file in files)
{
string path = file.FullName;
using (StreamReader streamReader = File.OpenText(path))
{
string html = streamReader.ReadToEnd();
List <HtmlNode> nodes = markupParser.Parse(html);
Dump(path, html, nodes);
LinkExtractor linkExtractor = new LinkExtractor();
ParsedHtml parsedHtml = new ParsedHtml();
parsedHtml.Html = html;
parsedHtml.Nodes = nodes;
linkExtractor.Process(parsedHtml, true);
Dump(parsedHtml);
}
}
}
public HtmlParser()
{
markupParser = new MarkupParser();
linkExtractor = new LinkExtractor();
}
