/// <summary>
/// 将文本解析为产生同样文本流效果的 HTML 片段
/// </summary>
/// <param name="text">要解析的文本</param>
/// <param name="manager">HTML 片段管理器</param>
/// <returns>HTML 片段</returns>
public static IHtmlFragment ParseText(string text, IHtmlFragmentManager manager)
{
if (text == null)
{
throw new ArgumentNullException("text");
}
if (manager == null)
{
throw new ArgumentNullException("manager");
}
var parsed = HtmlEncoding.HtmlEncode(text);
parsed = parsed.Replace(" ", " ");
if (parsed.EndsWith(" "))
{
parsed = parsed.Substring(0, parsed.Length - 1) + " ";//如果末尾多出一个空格，则替换为
}
parsed = parsed.Replace("\r\n", "\n").Replace("\r", "\n");
parsed = parsed.Replace("\n", "<br />");
return(manager.ParseFragment(parsed));
}
/// <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>
/// 使用指定文本替换元素内容（警告，此方法会清除元素所有内容）
/// </summary>
/// <param name="element">要替换内容的元素</param>
/// <param name="text">文本内容</param>
/// <param name="encodeWhiteSpaces">是否编码空白字符</param>
public static T InnerText <T>(this T element, string text, bool encodeWhiteSpaces) where T : IHtmlElement
{
if (element == null)
{
throw new ArgumentNullException("element");
}
var modifier = EnsureModifiable(element);
var mode = element.ElementTextMode();
lock (element.SyncRoot)
{
ClearNodes(element);
if (string.IsNullOrEmpty(text))//对于空输入，则只需要清空元素即可
{
return(element);
}
if (mode == TextMode.CData)
{
modifier.AddTextNode(element, text);
}
else if (mode == TextMode.Preformated || !encodeWhiteSpaces)
{
modifier.AddTextNode(element, HtmlEncoding.HtmlEncode(text));
}
else if (mode == TextMode.Normal)
{
var encoded = HtmlEncoding.HtmlEncode(text);
encoded = encoded.Replace(" ", " ");
if (encoded.EndsWith(" "))
{
encoded = encoded.Substring(0, encoded.Length - 1) + " ";//如果末尾多出一个空格，则替换为
}
encoded = encoded.Replace("\r\n", "\n").Replace("\r", "\n");
encoded = encoded.Replace("\n", "<br />");
element.Document.ParseFragment(encoded).Into(element, 0);
}
else
{
throw new InvalidOperationException("元素不包含任何文本内容，无法设置 InnerText");
}
}
return(element);
}
/*
* /// <summary>
* /// 使用指定文本替换元素内容（警告，此方法会清除元素所有内容）
* /// </summary>
* /// <param name="element">要替换内容的元素</param>
* /// <param name="text">文本内容</param>
* public static void ReplaceChildsWithText( this IHtmlElement element, string text )
* {
* if ( element == null )
* throw new ArgumentNullException( "element" );
*
* var factory = element.Document.GetNodeFactory();
* if ( factory == null )
* throw new NotSupportedException();
*
* lock ( element.SyncRoot )
* {
* ClearNodes( element );
*
* if ( !HtmlSpecification.cdataTags.Contains( element.Name, StringComparer.OrdinalIgnoreCase ) )
* {
* var fragment = HtmlEncode( text, factory );
*
* fragment.InsertTo( element, 0 );
* }
* else
* {
* var textNode = factory.CreateTextNode( text );
* textNode.Into( element, 0 );
* }
* }
*
* }
*
* /// <summary>
* /// 使用指定的HTML文本替换元素内容（警告，此方法会清除元素所有内容）
* /// </summary>
* /// <param name="element">要替换内容的元素</param>
* /// <param name="html">要替换的HTML代码</param>
* public static void ReplaceChildsWithHtml( this IHtmlElement element, string html )
* {
* if ( element == null )
* throw new ArgumentNullException( "element" );
*
* var factory = element.Document.GetNodeFactory();
* if ( factory == null )
* throw new NotSupportedException();
*
* lock ( element.SyncRoot )
* {
* ClearNodes( element );
*
* if ( HtmlSpecification.cdataTags.Contains( element.Name, StringComparer.OrdinalIgnoreCase ) )
* {
* var textNode = factory.CreateTextNode( html );
* textNode.Into( element, 0 );
* }
* else
* {
* var fragment = factory.ParseFragment( html );
* fragment.InsertTo( element, 0 );
* }
* }
* }
*/
private static HtmlFragment HtmlEncode(string text, IHtmlNodeFactory factory)
{
if (text == null)
{
throw new ArgumentNullException("text");
}
if (factory == null)
{
throw new ArgumentNullException("factory");
}
var fragment = new HtmlFragment(factory);
var encoded = HtmlEncoding.HtmlEncode(text);
encoded = encoded.Replace("\r\n", "\n").Replace("\r", "\n");
int index = 0, brIndex = 0;
while (true)
{
brIndex = encoded.IndexOf('\n', index);
if (brIndex == -1)
{
if (index < encoded.Length)
{
fragment.AddNode(factory.CreateTextNode(encoded.Substring(index)));
}
break;
}
if (index != brIndex)
{
fragment.AddNode(factory.CreateTextNode(encoded.Substring(index, brIndex - index)));
}
fragment.AddNode(factory.CreateElement("br"));
index = brIndex + 1;
}
return(fragment);
}
/// <summary>
/// 使用指定文本替换元素内容（警告，此方法会清除元素所有内容）
/// </summary>
/// <param name="element">要替换内容的元素</param>
/// <param name="text">文本内容</param>
/// <param name="encodeWhiteSpaces">是否编码空白字符</param>
public static T InnerText <T>(this T element, string text, bool encodeWhiteSpaces) where T : IHtmlElement
{
if (element == null)
{
throw new ArgumentNullException("element");
}
var modifier = EnsureModifiable(element);
lock (element.SyncRoot)
{
ClearNodes(element);
if (string.IsNullOrEmpty(text))//对于空输入，则只需要清空元素即可
{
return(element);
}
if (HtmlSpecification.cdataTags.Contains(element.Name, StringComparer.OrdinalIgnoreCase))
{
modifier.AddTextNode(element, text);
}
else if (HtmlSpecification.preformatedElements.Contains(element.Name, StringComparer.OrdinalIgnoreCase) || !encodeWhiteSpaces)
{
modifier.AddTextNode(element, HtmlEncoding.HtmlEncode(text));
}
else
{
var encoded = HtmlEncoding.HtmlEncode(text);
encoded = encoded.Replace(" ", " ");
encoded = encoded.Replace("\r\n", "\n").Replace("\r", "\n");
encoded = encoded.Replace("\n", "<br />");
element.Document.ParseFragment(encoded).Into(element, 0);
}
}
return(element);
}
/// <summary>
/// 使用指定文本替换元素内容（警告，此方法会清除元素所有内容）
/// </summary>
/// <param name="element">要替换内容的元素</param>
/// <param name="text">文本内容</param>
/// <param name="encodeWhiteSpaces">是否编码空白字符</param>
public static T InnerText <T>(this T element, string text, bool encodeWhiteSpaces) where T : IHtmlElement
{
if (element == null)
{
throw new ArgumentNullException("element");
}
var modifier = EnsureModifiable(element);
var mode = element.ElementTextMode();
lock (element.SyncRoot)
{
ClearNodes(element);
if (string.IsNullOrEmpty(text))//对于空输入，则只需要清空元素即可
{
return(element);
}
if (mode == TextMode.CData)
{
modifier.AddTextNode(element, text);
}
else if (mode == TextMode.Preformated || !encodeWhiteSpaces)
{
modifier.AddTextNode(element, HtmlEncoding.HtmlEncode(text));
}
else if (mode == TextMode.Normal)
{
ParseText(text, element.Document.FragmentManager).Into(element, 0);
}
else
{
throw new InvalidOperationException("元素不包含任何文本内容，无法设置 InnerText");
}
}
return(element);
}
/// <summary>
/// 尝试生成元素开始标签的HTML形式
/// </summary>
/// <param name="element">要生成HTML的元素</param>
/// <param name="selfClosed">指示是否应产生自结束符号</param>
/// <returns></returns>
public static string GenerateTagHtml(IHtmlElement element, bool selfClosed)
{
if (element == null)
{
throw new ArgumentNullException("element");
}
var builder = new StringBuilder(20);
builder.Append("<");
builder.Append(element.Name);
foreach (var attribute in element.Attributes())
{
builder.Append(" ");
builder.Append(attribute.Name);
if (attribute.AttributeValue != null)
{
var specification = element.Document.HtmlSpecification;
if ((specification.IsUriValue(attribute) || specification.IsScriptValue(attribute)) && !attribute.AttributeValue.Contains('"'))
{
builder.Append("=\"").Append(attribute.AttributeValue).Append("\"");
}
else
{
builder.Append("=\"").Append(HtmlEncoding.HtmlAttributeEncode(attribute.AttributeValue)).Append("\"");
}
}
}
if (selfClosed)
{
builder.Append(" /");
}
builder.Append(">");
return(builder.ToString());
}
public static IHtmlTextNode Add(this IHtmlContainer container, XText text)
{
return(container.AddTextNode(HtmlEncoding.HtmlEncode(text.Value)));
}
/// <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();
}