public static string QuerySelectorAttributeOrHtml(this HtmlNode htmlNode, SelectorSettings settings)
{
HtmlNode selectedNode = htmlNode.QuerySelector(settings.Selector);
if (selectedNode == null)
{
throw new ApplicationException($"Error: Selector {settings.Selector} failed to find any elements.");
}
if (string.IsNullOrWhiteSpace(settings.Attribute))
{
return(selectedNode.InnerHtml);
}
return(selectedNode.Attributes[settings.Attribute].Value);
}
public static string QuerySelectorAttributeOrText(this HtmlNode htmlNode, SelectorSettings settings)
{
HtmlNode selectedNode = htmlNode.QuerySelector(settings.Selector);
if (selectedNode == null)
{
throw new ApplicationException($"Error: Selector {settings.Selector} failed to find any elements.");
}
// Hack: Add physical newlines to <br />s to make date parsing easier for LastUpdated / Published.
// Also means that we match Firefox functionality.
foreach (HtmlNode nextNode in htmlNode.QuerySelectorAll("br"))
{
nextNode.InnerHtml = "\n";
}
if (string.IsNullOrWhiteSpace(settings.Attribute))
{
return(selectedNode.InnerText);
}
return(selectedNode.Attributes[settings.Attribute].Value);
}
