public HtmlTextElementIterator(HtmlPlainText HtmlPlainText)
{
}
/// <summary>
/// This function replaces the Html-Tag-Names from the input, while it conserves the structure of the html and possible Html-Attributes.
/// </summary>
/// <param name="HtmlInput">The Html-String you want to replace tags in.</param>
/// <param name="TagReplacementMap">Dictionary in which the replacement rules are defined. (Form: Key-OldTagName, Value-NewTagName)</param>
/// <param name="ReplacementType">Whether or not unmatched Tags (and their content!) should remain in the string.</param>
/// <returns></returns>
public string ApplyTagReplacementMap(string HtmlInput, Dictionary<string, string> TagReplacementMap, TagReplacementTypes ReplacementType)
{
HtmlTag root = new HtmlTag(null, false, "root");
HtmlConstruct ObjectConstruct = Analyze(HtmlInput, root);
root.AddChildren(ObjectConstruct.elements);
IHtmlTagIterator i = ObjectConstruct.getIterator();
while (i.hasNext())
{
HtmlElement ele = i.next();
if (ele.GetType() == typeof(HtmlPlainText)) // Current element is Plain Text
{
// ((HtmlPlainText)ele)
// ignore
}
else if (ele.GetType() == typeof(HtmlTag)) // Current element is tag
{
// ((HtmlTag)ele)
if (TagReplacementMap.Keys.Contains(((HtmlTag)ele).tagName)) // tag found in ReplacementMap
{
// Replace
HtmlTag replacement = new HtmlTag(((HtmlTag)ele).getParent(), ((HtmlTag)ele).isSingleTag, TagReplacementMap[((HtmlTag)ele).tagName], ((HtmlTag)ele).children, ((HtmlTag)ele).tagAttributes);
int index = ((HtmlTag)ele).getParent().children.IndexOf(ele);
((HtmlTag)ele).getParent().children[index] = replacement;
}
else // tag not found in ReplacementMap
{
switch (ReplacementType)
{
case TagReplacementTypes.RemoveNotMatchedTags:
((HtmlTag)ele).getParent().children.Remove(ele);
break;
case TagReplacementTypes.IgnoreNotMatchedTags:
// ignore
break;
case TagReplacementTypes.ConvertNotMatchedTagsToPlainText:
string plaintext = "";
IHtmlTagIterator it = ((HtmlTag)ele).getIterator();
while (it.hasNext())
{
HtmlElement e = it.next();
if (e.GetType() == typeof(HtmlPlainText)) plaintext += ((HtmlPlainText)e).content;
}
HtmlPlainText replacement = new HtmlPlainText(((HtmlTag)ele).getParent(), plaintext);
int index = ((HtmlTag)ele).getParent().children.IndexOf(ele);
((HtmlTag)ele).getParent().children[index] = replacement;
break;
default:
throw new NotImplementedException("TagReplacementType not implemented!"); //break;
}
}
}
else throw new NotImplementedException("HtmlElement Type not implemented!");
}
return new HtmlConstruct(root.children).ToString();
}