public override bool Process(Token t, HtmlTreeBuilder tb)
{
if (IsWhitespace(t))
{
tb.Insert(t.AsCharacter());
return(true);
}
switch (t.Type)
{
case TokenType.Comment:
tb.Insert(t.AsComment());
break;
case TokenType.Doctype:
tb.Error(this);
break;
case TokenType.StartTag:
Token.StartTag startTag = t.AsStartTag();
string name = startTag.Name;
if (name.Equals("html"))
{
return(tb.Process(t, InBody));
}
else if (name.Equals("col"))
{
tb.InsertEmpty(startTag);
}
else
{
return(AnythingElse(t, tb));
}
break;
case TokenType.EndTag:
Token.EndTag endTag = t.AsEndTag();
name = endTag.Name;
if (name.Equals("colgroup"))
{
if (tb.CurrentElement.NodeName.Equals("html")) // frag case
{
tb.Error(this);
return(false);
}
else
{
tb.Pop();
tb.Transition(InTable);
}
}
else
{
return(AnythingElse(t, tb));
}
break;
case TokenType.EOF:
if (tb.CurrentElement.NodeName.Equals("html"))
{
return(true); // stop parsing; frag case
}
else
{
return(AnythingElse(t, tb));
}
default:
return(AnythingElse(t, tb));
}
return(true);
}
public override bool Process(Token t, HtmlTreeBuilder tb)
{
if (t.IsCharacter)
{
tb.NewPendingTableCharacters();
tb.MarkInsertionMode();
tb.Transition(InTableText);
return(tb.Process(t));
}
else if (t.IsComment)
{
tb.Insert(t.AsComment());
return(true);
}
else if (t.IsDoctype)
{
tb.Error(this);
return(false);
}
else if (t.IsStartTag)
{
Token.StartTag startTag = t.AsStartTag();
string name = startTag.Name;
switch (name)
{
case "caption":
tb.ClearStackToTableContext();
tb.InsertMarkerToFormattingElements();
tb.Insert(startTag);
tb.Transition(InCaption);
break;
case "colgroup":
tb.ClearStackToTableContext();
tb.Insert(startTag);
tb.Transition(InColumnGroup);
break;
case "col":
tb.Process(new Token.StartTag("colgroup"));
return(tb.Process(t));
case "table":
tb.Error(this);
bool processed = tb.Process(new Token.EndTag("table"));
if (processed) // only ignored if in fragment
{
return(tb.Process(t));
}
break;
case "tbody":
case "tfoot":
case "thead":
tb.ClearStackToTableContext();
tb.Insert(startTag);
tb.Transition(InTableBody);
break;
case "td":
case "th":
case "tr":
tb.Process(new Token.StartTag("tbody"));
return(tb.Process(t));
case "style":
case "script":
return(tb.Process(t, InHead));
case "input":
if (!startTag.Attributes["type"]
.Equals("hidden", StringComparison.OrdinalIgnoreCase))
{
return(AnythingElse(t, tb));
}
else
{
tb.InsertEmpty(startTag);
}
break;
case "form":
tb.Error(this);
if (tb.FormElement != null)
{
return(false);
}
else
{
HtmlElement form = tb.InsertEmpty(startTag);
tb.FormElement = form;
}
break;
default:
return(AnythingElse(t, tb));
}
}
else if (t.IsEndTag)
{
Token.EndTag endTag = t.AsEndTag();
string name = endTag.Name;
switch (name)
{
case "table":
if (!tb.InTableScope(name))
{
tb.Error(this);
return(false);
}
else
{
tb.PopStackToClose("table");
}
tb.ResetInsertionMode();
break;
case "body":
case "caption":
case "col":
case "colgroup":
case "html":
case "tbody":
case "td":
case "tfoot":
case "th":
case "thead":
case "tr":
tb.Error(this);
return(false);
default:
return(AnythingElse(t, tb));
}
}
else if (t.IsEOF)
{
if (tb.CurrentElement.NodeName.Equals("html"))
{
tb.Error(this);
}
return(true); // stops parsing
}
return(AnythingElse(t, tb));
}
private Nullable <bool> HandleStartTag(Token t, HtmlTreeBuilder tb)
{
Token.StartTag startTag = t.AsStartTag();
string name = startTag.Name;
if (name.Equals("html"))
{
tb.Error(this);
// merge attributes onto real html
var html = tb.Stack.First();
foreach (HtmlAttribute attribute in startTag.Attributes)
{
if (!html.HasAttribute(attribute.Name))
{
html.Attributes.Add(attribute);
}
}
}
else if (StringSet.Create("base basefont bgsound command link meta noframes script style title").Contains(name))
{
return(tb.Process(t, InHead));
}
else if (name.Equals("body"))
{
tb.Error(this);
DescendableLinkedList <DomContainer> stack = tb.Stack;
if (stack.Count == 1 || (stack.Count > 2 && !stack.ElementAt(1).NodeName.Equals("body")))
{
// only in fragment case
return(false); // ignore
}
else
{
tb.FramesetOK = false;
var body = stack.First();
foreach (HtmlAttribute attribute in startTag.Attributes)
{
if (!body.HasAttribute(attribute.Name))
{
body.Attributes.Add(attribute);
}
}
}
}
else if (name.Equals("frameset"))
{
tb.Error(this);
DescendableLinkedList <DomContainer> stack = tb.Stack;
if (stack.Count == 1 || (stack.Count > 2 && !stack.ElementAt(1).NodeName.Equals("body")))
{
// only in fragment case
return(false); // ignore
}
else if (!tb.FramesetOK)
{
return(false); // ignore frameset
}
else
{
var second = stack.ElementAt(1);
if (second.Parent != null)
{
second.Remove();
}
// pop up to html element
while (stack.Count > 1)
{
stack.RemoveLast();
}
tb.Insert(startTag);
tb.Transition(InFrameset);
}
}
else if (StringSet.Create(
@"address article aside blockquote center details dir div dl
fieldset figcaption figure footer header hgroup menu nav ol
p section summary ul").Contains(name))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
}
else if (HeadingTags.Contains(name))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
if (HeadingTags.Contains(tb.CurrentElement.NodeName))
{
tb.Error(this);
tb.Pop();
}
tb.Insert(startTag);
}
else if (StringUtil.In(name, "pre", "listing"))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
// TODO: ignore LF if next token
tb.FramesetOK = false;
}
else if (name.Equals("form"))
{
if (tb.FormElement != null)
{
tb.Error(this);
return(false);
}
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
HtmlElement form = tb.Insert(startTag);
tb.FormElement = form;
}
else if (name.Equals("li"))
{
tb.FramesetOK = false;
DescendableLinkedList <DomContainer> stack = tb.Stack;
for (int i = stack.Count - 1; i > 0; i--)
{
var el = stack.ElementAt(i); // TODO Performance of this?
if (el.NodeName.Equals("li"))
{
tb.Process(new Token.EndTag("li"));
break;
}
if (tb.IsSpecial(el) && !StringUtil.In(el.NodeName, "address", "div", "p"))
{
break;
}
}
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
}
else if (StringUtil.In(name, "dd", "dt"))
{
tb.FramesetOK = false;
DescendableLinkedList <DomContainer> stack = tb.Stack;
for (int i = stack.Count - 1; i > 0; i--)
{
var el = stack.ElementAt(i);
if (StringUtil.In(el.NodeName, "dd", "dt"))
{
tb.Process(new Token.EndTag(el.NodeName));
break;
}
if (tb.IsSpecial(el) && !StringUtil.In(el.NodeName, "address", "div", "p"))
{
break;
}
}
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
}
else if (name.Equals("plaintext"))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
tb.tokeniser.Transition(TokeniserState.PLAINTEXT); // once in, never gets out
}
else if (name.Equals("button"))
{
if (tb.InButtonScope("button"))
{
// close and reprocess
tb.Error(this);
tb.Process(new Token.EndTag("button"));
tb.Process(startTag);
}
else
{
tb.ReconstructFormattingElements();
tb.Insert(startTag);
tb.FramesetOK = false;
}
}
else if (name.Equals("a"))
{
if (tb.GetActiveFormattingElement("a") != null)
{
tb.Error(this);
tb.Process(new Token.EndTag("a"));
// still on stack?
var remainingA = tb.GetFromStack("a");
if (remainingA != null)
{
tb.RemoveFromActiveFormattingElements(remainingA);
tb.RemoveFromStack(remainingA);
}
}
tb.ReconstructFormattingElements();
HtmlElement a = tb.Insert(startTag);
tb.PushActiveFormattingElements(a);
}
else if (StringSet.Create("b big code em font i s small strike strong tt u").Contains(name))
{
tb.ReconstructFormattingElements();
HtmlElement el = tb.Insert(startTag);
tb.PushActiveFormattingElements(el);
}
else if (name.Equals("nobr"))
{
tb.ReconstructFormattingElements();
if (tb.InScope("nobr"))
{
tb.Error(this);
tb.Process(new Token.EndTag("nobr"));
tb.ReconstructFormattingElements();
}
HtmlElement el = tb.Insert(startTag);
tb.PushActiveFormattingElements(el);
}
else if (StringUtil.In(name, "applet", "marquee", "object"))
{
tb.ReconstructFormattingElements();
tb.Insert(startTag);
tb.InsertMarkerToFormattingElements();
tb.FramesetOK = false;
}
else if (name.Equals("table"))
{
if (tb.Document.QuirksMode != QuirksMode.Quirks && tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
tb.FramesetOK = false;
tb.Transition(InTable);
}
else if (StringSet.Create("area br embed img keygen wbr").Contains(name))
{
tb.ReconstructFormattingElements();
tb.InsertEmpty(startTag);
tb.FramesetOK = false;
}
else if (name.Equals("input"))
{
tb.ReconstructFormattingElements();
HtmlElement el = tb.InsertEmpty(startTag);
if (!el.Attribute("type").Equals("hidden", StringComparison.OrdinalIgnoreCase))
{
tb.FramesetOK = false;
}
}
else if (StringUtil.In(name, "param", "source", "track"))
{
tb.InsertEmpty(startTag);
}
else if (name.Equals("hr"))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.InsertEmpty(startTag);
tb.FramesetOK = false;
}
else if (name.Equals("image"))
{
// we're not supposed to ask.
startTag.Name = "img";
return(tb.Process(startTag));
}
else if (name.Equals("isindex"))
{
// how much do we care about the early 90s?
tb.Error(this);
if (tb.FormElement != null)
{
return(false);
}
tb.tokeniser.AcknowledgeSelfClosingFlag();
tb.Process(new Token.StartTag("form"));
if (startTag.Attributes.Contains("action"))
{
HtmlElement form = tb.FormElement;
form.Attribute("action", startTag.Attributes["action"]);
}
tb.Process(new Token.StartTag("hr"));
tb.Process(new Token.StartTag("label"));
// hope you like english.
string prompt = startTag.Attributes.Contains("prompt") ?
startTag.Attributes["prompt"] :
"This is a searchable index. Enter search keywords: ";
tb.Process(new Token.Character(prompt));
// input
var inputStToken = new Token.StartTag("input");
HtmlAttributeCollection inputAttribs = inputStToken.Attributes;
inputAttribs["name"] = "isindex";
foreach (HtmlAttribute attr in startTag.Attributes)
{
if (!StringUtil.In(attr.Name.LocalName, "name", "action", "prompt"))
{
inputAttribs.Add(attr);
}
}
tb.Process(inputStToken);
tb.Process(new Token.EndTag("label"));
tb.Process(new Token.StartTag("hr"));
tb.Process(new Token.EndTag("form"));
}
else if (name.Equals("textarea"))
{
tb.Insert(startTag);
// TODO: If the next token is a U+000A LINE FEED (LF) char token, then ignore that token and move on to the next one. (Newlines at the start of textarea elements are ignored as an authoring convenience.)
tb.tokeniser.Transition(TokeniserState.Rcdata);
tb.MarkInsertionMode();
tb.FramesetOK = false;
tb.Transition(Text);
}
else if (name.Equals("xmp"))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.ReconstructFormattingElements();
tb.FramesetOK = false;
HandleRawtext(startTag, tb);
}
else if (name.Equals("iframe"))
{
tb.FramesetOK = false;
HandleRawtext(startTag, tb);
}
else if (name.Equals("noembed"))
{
// also handle noscript if script enabled
HandleRawtext(startTag, tb);
}
else if (name.Equals("select"))
{
tb.ReconstructFormattingElements();
tb.Insert(startTag);
tb.FramesetOK = false;
HtmlTreeBuilderState state = tb.State;
if (state.Equals(InTable) || state.Equals(InCaption) || state.Equals(InTableBody) || state.Equals(InRow) || state.Equals(InCell))
{
tb.Transition(InSelectInTable);
}
else
{
tb.Transition(InSelect);
}
}
else if (StringUtil.In(name, "optgroup", "option"))
{
if (tb.CurrentElement.NodeName.Equals("option"))
{
tb.Process(new Token.EndTag("option"));
}
tb.ReconstructFormattingElements();
tb.Insert(startTag);
}
else if (StringUtil.In(name, "rp", "rt"))
{
if (tb.InScope("ruby"))
{
tb.GenerateImpliedEndTags();
if (!tb.CurrentElement.NodeName.Equals("ruby"))
{
tb.Error(this);
tb.PopStackToBefore("ruby"); // i.e. close up to but not include name
}
tb.Insert(startTag);
}
}
else if (name.Equals("math"))
{
tb.ReconstructFormattingElements();
// TODO: handle A start tag whose tag name is "math" (i.e. foreign, mathml)
tb.Insert(startTag);
tb.tokeniser.AcknowledgeSelfClosingFlag();
}
else if (name.Equals("svg"))
{
tb.ReconstructFormattingElements();
// TODO: handle A start tag whose tag name is "svg" (xlink, svg)
tb.Insert(startTag);
tb.tokeniser.AcknowledgeSelfClosingFlag();
}
else if (StringSet.Create("caption col colgroup frame head tbody td tfoot th thead tr").Contains(name))
{
tb.Error(this);
return(false);
}
else
{
tb.ReconstructFormattingElements();
tb.Insert(startTag);
}
return(null);
}
public override bool Process(Token t, HtmlTreeBuilder tb)
{
if (IsWhitespace(t))
{
tb.Insert(t.AsCharacter());
}
else if (t.IsComment)
{
tb.Insert(t.AsComment());
}
else if (t.IsDoctype)
{
tb.Error(this);
return(false);
}
else if (t.IsStartTag)
{
Token.StartTag start = t.AsStartTag();
switch (start.Name)
{
case "html":
return(tb.Process(start, InBody));
case "frameset":
tb.Insert(start);
break;
case "frame":
tb.InsertEmpty(start);
break;
case "noframes":
return(tb.Process(start, InHead));
default:
tb.Error(this);
return(false);
}
}
else if (t.IsEndTag && t.AsEndTag().Name.Equals("frameset"))
{
if (tb.CurrentElement.NodeName.Equals("html")) // frag
{
tb.Error(this);
return(false);
}
else
{
tb.Pop();
if (!tb.IsFragmentParsing() && !tb.CurrentElement.NodeName.Equals("frameset"))
{
tb.Transition(AfterFrameset);
}
}
}
else if (t.IsEOF)
{
if (!tb.CurrentElement.NodeName.Equals("html"))
{
tb.Error(this);
return(true);
}
}
else
{
tb.Error(this);
return(false);
}
return(true);
}
public override bool Process(Token t, HtmlTreeBuilder tb)
{
if (IsWhitespace(t))
{
tb.Insert(t.AsCharacter());
return(true);
}
switch (t.Type)
{
case TokenType.Comment:
tb.Insert(t.AsComment());
break;
case TokenType.Doctype:
tb.Error(this);
return(false);
case TokenType.StartTag:
Token.StartTag start = t.AsStartTag();
string name = start.Name;
if (name.Equals("html"))
{
return(InBody.Process(t, tb));
}
else if (StringUtil.In(name, "base", "basefont", "bgsound", "command", "link"))
{
HtmlElement el = tb.InsertEmpty(start);
// jsoup special: update base the frist time it is seen
if (name.Equals("base") && el.HasAttribute("href"))
{
tb.MaybeSetBaseUri(el);
}
}
else if (name.Equals("meta"))
{
HtmlElement meta = tb.InsertEmpty(start);
// TODO: charset switches
}
else if (name.Equals("title"))
{
HandleRcData(start, tb);
}
else if (StringUtil.In(name, "noframes", "style"))
{
HandleRawtext(start, tb);
}
else if (name.Equals("noscript"))
{
// else if noscript && scripting flag = true: rawtext (jsoup doesn't run script, to handle as noscript)
tb.Insert(start);
tb.Transition(InHeadNoscript);
}
else if (name.Equals("script"))
{
// skips some script rules as won't execute them
tb.Insert(start);
tb.tokeniser.Transition(TokeniserState.ScriptData);
tb.MarkInsertionMode();
tb.Transition(Text);
}
else if (name.Equals("head"))
{
tb.Error(this);
return(false);
}
else
{
return(AnythingElse(t, tb));
}
break;
case TokenType.EndTag:
Token.EndTag end = t.AsEndTag();
name = end.Name;
if (name.Equals("head"))
{
tb.Pop();
tb.Transition(AfterHead);
}
else if (StringUtil.In(name, "body", "html", "br"))
{
return(AnythingElse(t, tb));
}
else
{
tb.Error(this);
return(false);
}
break;
default:
return(AnythingElse(t, tb));
}
return(true);
}