/// <summary>
/// 尝试获取节点的文本表现形式，对于某些不支持文本表现形式的元素，将直接返回null
/// </summary>
/// <param name="node">要获取文本表现形式的节点</param>
/// <returns></returns>
public static string InnerText(this IHtmlNode node)
{
var textNode = node as IHtmlTextNode;
if (textNode != null)
{
var parent = textNode.Parent();
if (parent == null)
{
throw new InvalidOperationException();
}
if (HtmlSpecification.cdataTags.Contains(parent.Name, StringComparer.OrdinalIgnoreCase))
{
return(textNode.HtmlText);
}
else if (HtmlSpecification.preformatedElements.Contains(parent.Name, StringComparer.OrdinalIgnoreCase))
{
return(HtmlEncoding.HtmlDecode(textNode.HtmlText));
}
else
{
return(HtmlEncoding.HtmlDecode(whitespaceRegex.Replace(textNode.HtmlText, " ")));
}
}
var commentNode = node as IHtmlComment;
if (commentNode != null)
{
return(null);
}
var element = node as IHtmlElement;
if (element != null)
{
if (element.Name.EqualsIgnoreCase("br"))
{
return(Environment.NewLine);
}
else if (HtmlSpecification.nonTextElements.Contains(element.Name, StringComparer.OrdinalIgnoreCase))
{
return(null);
}
}
var container = node as IHtmlContainer;
return(string.Join("", container.Nodes().Select(n => InnerText(n)).ToArray()));
}
/// <summary>
/// 尝试获取节点的文本表现形式，对于某些不支持文本表现形式的元素，将直接返回null
/// </summary>
/// <param name="node">要获取文本表现形式的节点</param>
/// <returns></returns>
public static string InnerText(this IHtmlNode node)
{
if (node == null)
{
throw new ArgumentNullException("node");
}
var specification = node.Document.HtmlSpecification;
var textNode = node as IHtmlTextNode;
if (textNode != null)
{
var parent = textNode.Parent();
if (parent == null)
{
throw new InvalidOperationException();
}
var textMode = specification.ElementTextMode(parent);
if (textMode == TextMode.CData)
{
return(textNode.HtmlText);
}
else if (textMode == TextMode.Preformated)
{
return(HtmlEncoding.HtmlDecode(textNode.HtmlText));
}
else if (textMode == TextMode.Normal)
{
return(HtmlEncoding.HtmlDecode(whitespaceRegex.Replace(textNode.HtmlText, " ")));
}
else
{
return(null);
}
}
var commentNode = node as IHtmlComment;
if (commentNode != null)
{
return(null);
}
var element = node as IHtmlElement;
if (element != null)
{
if (element.Name.EqualsIgnoreCase("br"))
{
return(Environment.NewLine);
}
else if (specification.ElementTextMode(element) == TextMode.NonText)
{
return(null);
}
}
var container = node as IHtmlContainer;
if (container != null)
{
return(string.Join("", container.Nodes().Select(n => InnerText(n)).ToArray()));
}
throw new NotSupportedException();
}