public static void ParseHtmlDom(TextSource snapSource, IHtmlDocument htmldoc, WebDom.DomElement parentElement)
{
var parser = GetHtmlParser();
//------------------------
parser.Parse(snapSource, (LayoutFarm.WebDom.Impl.HtmlDocument)htmldoc, parentElement);
FreeHtmlParser(parser);
}
/// <summary>
/// Parses the source html to css boxes tree structure.
/// </summary>
/// <param name="source">the html source to parse</param>
public static HtmlDocument ParseDocument(TextSource snapSource)
{
var parser = GetHtmlParser();
//------------------------
var blankHtmlDoc = new HtmlDocument();
parser.Parse(snapSource, blankHtmlDoc, blankHtmlDoc.RootNode);
FreeHtmlParser(parser);
return blankHtmlDoc;
}
private void button1_Click(object sender, EventArgs e)
{
//test web parser
var parser = HtmlParser.CreateHtmlParser(ParseEngineKind.MyHtmlParser);
var blankHtmlDoc = new LayoutFarm.WebDom.Impl.HtmlDocument();
var snapSource = new TextSource(this.richTextBox1.Text.ToCharArray());
parser.Parse(snapSource, blankHtmlDoc, blankHtmlDoc.RootNode);
this.treeView2.Nodes.Clear();
var rootNode = new TreeNode("root");
DescibeNode(blankHtmlDoc.RootNode, rootNode);
treeView2.Nodes.Add(rootNode);
this.treeView2.ExpandAll();
}
public override void Parse(TextSource textSnapshot, WebDocument htmldoc, DomElement currentNode)
{
this._resultHtmlDoc = htmldoc;
char[] copyBuffer = textSnapshot.ActualSnapshot.Copy(0, textSnapshot.ActualSnapshot.Length);
using (var ms = new System.IO.MemoryStream(System.Text.Encoding.UTF8.GetBytes(copyBuffer)))
using (var textReader = new System.IO.StreamReader(ms))
{
var tokenizer = new HtmlTokenizer(textReader);
HtmlToken token;
while (tokenizer.ReadNextToken(out token))
{
switch (token.Kind)
{
case HtmlTokenKind.Data:
{
var text = (HtmlDataToken)token;
currentNode.AddChild(_resultHtmlDoc.CreateTextNode(text.Data.ToCharArray()));
}
break;
case HtmlTokenKind.Tag:
{
var tag = (HtmlTagToken)token;
if (!tag.IsEndTag)
{
//open tag
DomElement elem = this._resultHtmlDoc.CreateElement(null, tag.Name);
currentNode.AddChild(elem);
foreach (var attribute in tag.Attributes)
{
var attr = this._resultHtmlDoc.CreateAttribute(null, attribute.Name);
if (attribute.Value != null)
{
attr.Value = attribute.Value;
}
elem.AddAttribute(attr);
}
if (!tag.IsEmptyElement)
{
openEltStack.Push(currentNode);
currentNode = elem;
}
}
else
{
//this is end tag
//check end tag match or not
int tagNameIndex = _resultHtmlDoc.AddStringIfNotExists(tag.Name);
if (currentNode.Name == tag.Name)
{
currentNode = openEltStack.Pop();
}
else
{
//if not equal then check if current node need close tag or not
int count = 3;//?
bool ok = false;
while (count > 0)
{
if (HtmlTagMatching.IsSingleTag(currentNode.LocalNameIndex))
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
}
if (currentNode.LocalName == tag.Name)
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
ok = true;
break;
}
}
}
else if (HtmlTagMatching.CanAutoClose(currentNode.LocalNameIndex))
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
}
if (currentNode.LocalName == tag.Name)
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
ok = true;
break;
}
}
}
else
{
//implement err handling here!
throw new NotSupportedException();
}
count--;
}
if (!ok)
{
throw new NotSupportedException();
}
}
}
}
break;
case HtmlTokenKind.Comment:
break;
case HtmlTokenKind.DocType:
break;
default:
{
}
break;
}
}
}
}
public void Parse(TextSource textSnapshot, WebDocument htmldoc, DomElement currentNode)
{
this.Parse(textSnapshot.ActualSnapshot, htmldoc, currentNode);
}
public abstract void Parse(TextSource textSnapshot, WebDocument htmldoc, DomElement currentNode);
public override void Parse(TextSource textSnapshot, WebDocument htmldoc, DomElement currentNode)
{
this._resultHtmlDoc = htmldoc;
char[] copyBuffer = textSnapshot.ActualSnapshot.Copy(0, textSnapshot.ActualSnapshot.Length);
using (var ms = new System.IO.MemoryStream(System.Text.Encoding.UTF8.GetBytes(copyBuffer)))
using (var textReader = new System.IO.StreamReader(ms))
{
var tokenizer = new HtmlTokenizer(textReader);
HtmlToken token;
while (tokenizer.ReadNextToken(out token))
{
switch (token.Kind)
{
case HtmlTokenKind.Data:
{
var text = (HtmlDataToken)token;
currentNode.AddChild(_resultHtmlDoc.CreateTextNode(text.Data.ToCharArray()));
}
break;
case HtmlTokenKind.Tag:
{
var tag = (HtmlTagToken)token;
if (!tag.IsEndTag)
{
//open tag
DomElement elem = this._resultHtmlDoc.CreateElement(null, tag.Name);
currentNode.AddChild(elem);
foreach (var attribute in tag.Attributes)
{
var attr = this._resultHtmlDoc.CreateAttribute(null, attribute.Name);
if (attribute.Value != null)
{
attr.Value = attribute.Value;
}
elem.AddAttribute(attr);
}
if (!tag.IsEmptyElement)
{
openEltStack.Push(currentNode);
currentNode = elem;
}
}
else
{
//this is end tag
//check end tag match or not
int tagNameIndex = _resultHtmlDoc.AddStringIfNotExists(tag.Name);
if (currentNode.Name == tag.Name)
{
currentNode = openEltStack.Pop();
}
else
{
//if not equal then check if current node need close tag or not
int count = 3; //?
bool ok = false;
while (count > 0)
{
if (HtmlTagMatching.IsSingleTag(currentNode.LocalNameIndex))
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
}
if (currentNode.LocalName == tag.Name)
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
ok = true;
break;
}
}
}
else if (HtmlTagMatching.CanAutoClose(currentNode.LocalNameIndex))
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
}
if (currentNode.LocalName == tag.Name)
{
if (openEltStack.Count > 0)
{
currentNode = openEltStack.Pop();
ok = true;
break;
}
}
}
else
{
//implement err handling here!
throw new NotSupportedException();
}
count--;
}
if (!ok)
{
throw new NotSupportedException();
}
}
}
}
break;
case HtmlTokenKind.Comment:
break;
case HtmlTokenKind.DocType:
break;
default:
{
}
break;
}
}
}
}
public override void Parse(TextSource textSnapshot, WebDocument htmldoc, DomElement currentNode)
{
this.Parse(textSnapshot.ActualSnapshot, htmldoc, currentNode);
}
public static string GetInternalText(TextSource textsource)
{
return new string(TextSnapshot.UnsafeGetInternalBuffer(textsource.actualSnapshot));
}
public static string GetInternalText(TextSource textsource)
{
return(new string(TextSnapshot.UnsafeGetInternalBuffer(textsource.actualSnapshot)));
}
