public override bool Process(Token t, TreeBuilder tb)
{
switch (t.Type)
{
case Token.TokenType.Character:
Token.Character c = t.AsCharacter();
if (c.Data.Equals(_nullString))
{
// todo confirm that check
tb.Error(this);
return false;
}
else if (IsWhitespace(c))
{
tb.ReconstructFormattingElements();
tb.Insert(c);
}
else
{
tb.ReconstructFormattingElements();
tb.Insert(c);
tb.FramesetOk(false);
}
break;
case Token.TokenType.Comment:
tb.Insert(t.AsComment());
break;
case Token.TokenType.Doctype:
tb.Error(this);
return false;
case Token.TokenType.StartTag:
Token.StartTag startTag = t.AsStartTag();
string name = startTag.Name();
if (name.Equals("html"))
{
tb.Error(this);
// merge attributes onto real html
Element html = tb.Stack.First.Value;
foreach (NSoup.Nodes.Attribute attribute in startTag.Attributes)
{
if (!html.HasAttr(attribute.Key))
{
html.Attributes.Add(attribute);
}
}
}
else if (StringUtil.In(name, "base", "basefont", "bgsound", "command", "link", "meta", "noframes", "script", "style", "title"))
{
return tb.Process(t, InHead);
}
else if (name.Equals("body"))
{
tb.Error(this);
LinkedList<Element> 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);
Element body = stack.ElementAt(1);
foreach (NSoup.Nodes.Attribute attribute in startTag.Attributes)
{
if (!body.HasAttr(attribute.Key))
{
body.Attributes.Add(attribute);
}
}
}
}
else if (name.Equals("frameset"))
{
tb.Error(this);
LinkedList<Element> 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
{
Element 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 (StringUtil.In(name,
"address", "article", "aside", "blockquote", "center", "details", "dir", "div", "dl",
"fieldset", "figcaption", "figure", "footer", "header", "hgroup", "menu", "nav", "ol",
"p", "section", "summary", "ul"))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
}
else if (StringUtil.In(name, "h1", "h2", "h3", "h4", "h5", "h6"))
{
if (tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
if (StringUtil.In(tb.CurrentElement.NodeName, "h1", "h2", "h3", "h4", "h5", "h6"))
{
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"));
}
Element form = tb.Insert(startTag);
tb.FormElement = form;
}
else if (name.Equals("li"))
{
tb.FramesetOk(false);
LinkedList<Element> stack = tb.Stack;
for (int i = stack.Count - 1; i > 0; i--)
{
Element el = stack.ElementAt(i);
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);
LinkedList<Element> stack = tb.Stack;
for (int i = stack.Count - 1; i > 0; i--)
{
Element 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?
Element remainingA = tb.GetFromStack("a");
if (remainingA != null)
{
tb.RemoveFromActiveFormattingElements(remainingA);
tb.RemoveFromStack(remainingA);
}
}
tb.ReconstructFormattingElements();
Element a = tb.Insert(startTag);
tb.PushActiveFormattingElements(a);
}
else if (StringUtil.In(name,
"b", "big", "code", "em", "font", "i", "s", "small", "strike", "strong", "tt", "u"))
{
tb.ReconstructFormattingElements();
Element 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();
}
Element 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() != Document.QuirksModeEnum.Quirks && tb.InButtonScope("p"))
{
tb.Process(new Token.EndTag("p"));
}
tb.Insert(startTag);
tb.FramesetOk(false);
tb.Transition(InTable);
}
else if (StringUtil.In(name, "area", "br", "embed", "img", "keygen", "wbr"))
{
tb.ReconstructFormattingElements();
tb.InsertEmpty(startTag);
tb.FramesetOk(false);
}
else if (name.Equals("input"))
{
tb.ReconstructFormattingElements();
Element el = tb.InsertEmpty(startTag);
if (!el.Attr("type").Equals("hidden", StringComparison.InvariantCultureIgnoreCase))
{
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.ContainsKey("action"))
{
Element form = tb.FormElement;
form.Attr("action", startTag.Attributes["action"]);
}
tb.Process(new Token.StartTag("hr"));
tb.Process(new Token.StartTag("label"));
// hope you like english.
string prompt = startTag.Attributes.ContainsKey("prompt") ?
startTag.Attributes["prompt"] :
"This is a searchable index. Enter search keywords: ";
tb.Process(new Token.Character(prompt));
// input
Attributes inputAttribs = new Attributes();
foreach (NSoup.Nodes.Attribute attr in startTag.Attributes)
{
if (!StringUtil.In(attr.Key, "name", "action", "prompt"))
{
inputAttribs.Add(attr);
}
}
inputAttribs["name"] = "isindex";
tb.Process(new Token.StartTag("input", inputAttribs));
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) character 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);
TreeBuilderState 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("optgroup", "option"))
{
if (tb.CurrentElement.NodeName.Equals("option"))
{
tb.Process(new Token.EndTag("option"));
}
tb.ReconstructFormattingElements();
tb.Insert(startTag);
}
else if (StringUtil.In("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 (StringUtil.In(name,
"caption", "col", "colgroup", "frame", "head", "tbody", "td", "tfoot", "th", "thead", "tr"))
{
tb.Error(this);
return false;
}
else
{
tb.ReconstructFormattingElements();
tb.Insert(startTag);
}
break;
case Token.TokenType.EndTag:
Token.EndTag endTag = t.AsEndTag();
name = endTag.Name();
if (name.Equals("body"))
{
if (!tb.InScope("body"))
{
tb.Error(this);
return false;
}
else
{
// todo: error if stack contains something not dd, dt, li, optgroup, option, p, rp, rt, tbody, td, tfoot, th, thead, tr, body, html
tb.Transition(AfterBody);
}
}
else if (name.Equals("html"))
{
bool notIgnored = tb.Process(new Token.EndTag("body"));
if (notIgnored)
{
return tb.Process(endTag);
}
}
else if (StringUtil.In(name,
"address", "article", "aside", "blockquote", "button", "center", "details", "dir", "div",
"dl", "fieldset", "figcaption", "figure", "footer", "header", "hgroup", "listing", "menu",
"nav", "ol", "pre", "section", "summary", "ul"))
{
// todo: refactor these lookups
if (!tb.InScope(name))
{
// nothing to close
tb.Error(this);
return false;
}
else
{
tb.GenerateImpliedEndTags();
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
tb.PopStackToClose(name);
}
}
else if (name.Equals("form"))
{
Element currentForm = tb.FormElement;
tb.FormElement = null;
if (currentForm == null || !tb.InScope(name))
{
tb.Error(this);
return false;
}
else
{
tb.GenerateImpliedEndTags();
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
// remove currentForm from stack. will shift anything under up.
tb.RemoveFromStack(currentForm);
}
}
else if (name.Equals("p"))
{
if (!tb.InButtonScope(name))
{
tb.Error(this);
tb.Process(new Token.StartTag(name)); // if no p to close, creates an empty <p></p>
return tb.Process(endTag);
}
else
{
tb.GenerateImpliedEndTags(name);
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
tb.PopStackToClose(name);
}
}
else if (name.Equals("li"))
{
if (!tb.InListItemScope(name))
{
tb.Error(this);
return false;
}
else
{
tb.GenerateImpliedEndTags(name);
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
tb.PopStackToClose(name);
}
}
else if (StringUtil.In(name, "dd", "dt"))
{
if (!tb.InScope(name))
{
tb.Error(this);
return false;
}
else
{
tb.GenerateImpliedEndTags(name);
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
tb.PopStackToClose(name);
}
}
else if (StringUtil.In(name, "h1", "h2", "h3", "h4", "h5", "h6"))
{
if (!tb.InScope(new string[] { "h1", "h2", "h3", "h4", "h5", "h6" }))
{
tb.Error(this);
return false;
}
else
{
tb.GenerateImpliedEndTags(name);
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
tb.PopStackToClose("h1", "h2", "h3", "h4", "h5", "h6");
}
}
else if (name.Equals("sarcasm"))
{
// *sigh*
return AnyOtherEndTag(t, tb);
}
else if (StringUtil.In(name,
"a", "b", "big", "code", "em", "font", "i", "nobr", "s", "small", "strike", "strong", "tt", "u"))
{
// Adoption Agency Algorithm.
//OUTER:
for (int i = 0; i < 8; i++)
{
Element formatEl = tb.GetActiveFormattingElement(name);
if (formatEl == null)
{
return AnyOtherEndTag(t, tb);
}
else if (!tb.OnStack(formatEl))
{
tb.Error(this);
tb.RemoveFromActiveFormattingElements(formatEl);
return true;
}
else if (!tb.InScope(formatEl.NodeName))
{
tb.Error(this);
return false;
}
else if (tb.CurrentElement != formatEl)
{
tb.Error(this);
}
Element furthestBlock = null;
Element commonAncestor = null;
bool seenFormattingElement = false;
LinkedList<Element> stack = tb.Stack;
for (int si = 0; si < stack.Count; si++)
{
Element el = stack.ElementAt(si);
if (el == formatEl)
{
commonAncestor = stack.ElementAt(si - 1);
seenFormattingElement = true;
}
else if (seenFormattingElement && tb.IsSpecial(el))
{
furthestBlock = el;
break;
}
}
if (furthestBlock == null)
{
tb.PopStackToClose(formatEl.NodeName);
tb.RemoveFromActiveFormattingElements(formatEl);
return true;
}
// todo: Let a bookmark note the position of the formatting element in the list of active formatting elements relative to the elements on either side of it in the list.
// does that mean: int pos of format el in list?
Element node = furthestBlock;
Element lastNode = furthestBlock;
for (int j = 0; j < 3; j++)
{
if (tb.OnStack(node))
node = tb.AboveOnStack(node);
if (!tb.IsInActiveFormattingElements(node))
{ // note no bookmark check
tb.RemoveFromStack(node);
continue;
}
else if (node == formatEl)
{
break;
}
Element replacement = new Element(Tag.ValueOf(node.NodeName), tb.BaseUri);
tb.ReplaceActiveFormattingElement(node, replacement);
tb.ReplaceOnStack(node, replacement);
node = replacement;
if (lastNode == furthestBlock)
{
// todo: move the aforementioned bookmark to be immediately after the new node in the list of active formatting elements.
// not getting how this bookmark both straddles the element above, but is inbetween here...
}
if (lastNode.Parent != null)
{
lastNode.Remove();
}
node.AppendChild(lastNode);
lastNode = node;
}
if (StringUtil.In(commonAncestor.NodeName, "table", "tbody", "tfoot", "thead", "tr"))
{
if (lastNode.Parent != null)
{
lastNode.Remove();
}
tb.InsertInFosterParent(lastNode);
}
else
{
if (lastNode.Parent != null)
{
lastNode.Remove();
}
commonAncestor.AppendChild(lastNode);
}
Element adopter = new Element(Tag.ValueOf(name), tb.BaseUri);
Node[] childNodes = furthestBlock.ChildNodes.ToArray();
foreach (Node childNode in childNodes)
{
adopter.AppendChild(childNode); // append will reparent. thus the clone to avvoid concurrent mod.
}
furthestBlock.AppendChild(adopter);
tb.RemoveFromActiveFormattingElements(formatEl);
// todo: insert the new element into the list of active formatting elements at the position of the aforementioned bookmark.
tb.RemoveFromStack(formatEl);
tb.InsertOnStackAfter(furthestBlock, adopter);
}
}
else if (StringUtil.In(name, "applet", "marquee", "object"))
{
if (!tb.InScope("name"))
{
if (!tb.InScope(name))
{
tb.Error(this);
return false;
}
tb.GenerateImpliedEndTags();
if (!tb.CurrentElement.NodeName.Equals(name))
{
tb.Error(this);
}
tb.PopStackToClose(name);
tb.ClearFormattingElementsToLastMarker();
}
}
else if (name.Equals("br"))
{
tb.Error(this);
tb.Process(new Token.StartTag("br"));
return false;
}
else
{
return AnyOtherEndTag(t, tb);
}
break;
case Token.TokenType.EOF:
// todo: error if stack contains something not dd, dt, li, p, tbody, td, tfoot, th, thead, tr, body, html
// stop parsing
break;
default:
break;
}
return true;
}
public override bool Process(Token t, TreeBuilder tb)
{
if (IsWhitespace(t))
{
tb.Insert(t.AsCharacter());
return true;
}
switch (t.Type)
{
case Token.TokenType.Comment:
tb.Insert(t.AsComment());
break;
case Token.TokenType.Doctype:
tb.Error(this);
return false;
case Token.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"))
{
Element el = tb.InsertEmpty(start);
// jsoup special: update base as it is seen. todo: flip to current browser behaviour of one shot
if (name.Equals("base") && el.HasAttr("href"))
tb.SetBaseUri(el);
}
else if (name.Equals("meta"))
{
Element 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 Token.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;
}
protected void HandleRawText(Token.StartTag startTag, TreeBuilder tb)
{
tb.Insert(startTag);
tb.Tokeniser.Transition(TokeniserState.RawText);
tb.MarkInsertionMode();
tb.Transition(Text);
}
public override bool Process(Token t, TreeBuilder tb)
{
if (t.IsCharacter())
{
tb.NewPendingTableCharacters();
tb.MarkInsertionMode();
tb.Transition(InTableText);
return tb.Process(t);
}
else if (t.IsComment())
{
tb.Insert(t.AsComment());
}
else if (t.IsDoctype())
{
tb.Error(this);
return false;
}
else if (t.IsStartTag())
{
Token.StartTag startTag = t.AsStartTag();
string name = startTag.Name();
if (name.Equals("caption"))
{
tb.ClearStackToTableContext();
tb.InsertMarkerToFormattingElements();
tb.Insert(startTag);
tb.Transition(InCaption);
}
else if (name.Equals("colgroup"))
{
tb.ClearStackToTableContext();
tb.Insert(startTag);
tb.Transition(InColumnGroup);
}
else if (name.Equals("col"))
{
tb.Process(new Token.StartTag("colgroup"));
return tb.Process(t);
}
else if (StringUtil.In(name, "tbody", "tfoot", "thead"))
{
tb.ClearStackToTableContext();
tb.Insert(startTag);
tb.Transition(InTableBody);
}
else if (StringUtil.In(name, "td", "th", "tr"))
{
tb.Process(new Token.StartTag("tbody"));
return tb.Process(t);
}
else if (name.Equals("table"))
{
tb.Error(this);
bool processed = tb.Process(new Token.EndTag("table"));
if (processed) // only ignored if in fragment
{
return tb.Process(t);
}
}
else if (StringUtil.In(name, "style", "script"))
{
return tb.Process(t, InHead);
}
else if (name.Equals("input"))
{
if (!startTag.Attributes["type"].Equals("hidden", StringComparison.InvariantCultureIgnoreCase))
{
return AnythingElse(t, tb);
}
else
{
tb.InsertEmpty(startTag);
}
}
else if (name.Equals("form"))
{
tb.Error(this);
if (tb.FormElement != null)
return false;
else
{
Element form = tb.InsertEmpty(startTag);
tb.FormElement = form;
}
}
else
{
return AnythingElse(t, tb);
}
}
else if (t.IsEndTag())
{
Token.EndTag endTag = t.AsEndTag();
string name = endTag.Name();
if (name.Equals("table"))
{
if (!tb.InTableScope(name))
{
tb.Error(this);
return false;
}
else
{
tb.PopStackToClose("table");
}
tb.ResetInsertionMode();
}
else if (StringUtil.In(name,
"body", "caption", "col", "colgroup", "html", "tbody", "td", "tfoot", "th", "thead", "tr"))
{
tb.Error(this);
return false;
}
else
{
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);
}