public List <string> ParseList(XpathExtractModel model)
{
if (model == null || string.IsNullOrEmpty(model.XpathRule))
{
return(null);
}
HtmlNodeCollection nodes = this.htmlNode?.SelectNodes(model.XpathRule);
if (nodes == null || nodes.Count <= 0)
{
return(null);
}
if (model.XpathEndAttributes != null && model.XpathEndAttributes.Count > 0)
{
return(nodes.Select(n => n.Attributes.Where(a => model.XpathEndAttributes.Contains(a.Name)).Select(a => a.Value?.Trim()).FirstOrDefault()).
Where(n => !string.IsNullOrEmpty(n))
.ToList());
}
switch (model.ExtractType)
{
case ExtractType.Text:
return(nodes.Select(n => n.InnerText.Trim()).Where(n => !string.IsNullOrEmpty(n)).ToList());
case ExtractType.Html:
default:
return(nodes.Select(n => n.InnerHtml.Trim()).Where(n => !string.IsNullOrEmpty(n)).ToList());
}
}
public string ParseFirst(XpathExtractModel model)
{
List <string> results = this.ParseList(model);
if (results == null || results.Count <= 0)
{
return(null);
}
return(results.FirstOrDefault());
//if (this._rootNode == null || model == null || string.IsNullOrEmpty(model.XpathRule)) return null;
//HtmlNodeCollection nodes = this._rootNode.SelectNodes(model.XpathRule);
//if (nodes == null) return null;
//switch (model.ExtractType)
//{
// case ExtractType.Text:
// return nodes.FirstOrDefault().InnerText.Trim();
// case ExtractType.Html:
// default:
// return nodes.FirstOrDefault().InnerHtml.Trim();
//}
}
public string ParseFirst(XpathExtractModel model)
{
List <string> results = this.ParseList(model);
if (results == null || results.Count <= 0)
{
return(null);
}
return(results.FirstOrDefault());
}
public List <string> ParseList(XpathExtractModel model)
{
if (this._rootNode == null || model == null || string.IsNullOrEmpty(model.XpathRule))
{
return(null);
}
HtmlNodeCollection nodes = this._rootNode.SelectNodes(model.XpathRule);
return(this.GetNodeValues(nodes, model.XpathEndAttributes, model.ExtractType));
}
