Element insertStartTag(String startTagName) { Element el = new Element(Tag.valueOf(startTagName, settings), baseUri); insert(el); return el; }
Element insert(Token.StartTag startTag) { // handle empty unknown tags // when the spec expects an empty tag, will directly hit insertEmpty, so won't generate this fake end tag. if (startTag.isSelfClosing()) { Element el = insertEmpty(startTag); stack.add(el); tokeniser.transition(TokeniserState.Data); // handles <script />, otherwise needs breakout steps from script data tokeniser.emit(emptyEnd.reset().name(el.tagName())); // ensure we get out of whatever state we are in. emitted for yielded processing return el; } Element el = new Element(Tag.valueOf(startTag.name(), settings), baseUri, settings.normalizeAttributes(startTag.attributes)); insert(el); return el; }
boolean process(Token t, HtmlTreeBuilder tb) { if (isWhitespace(t)) { tb.insert(t.asCharacter()); return true; tb.insert(t.asComment()); break; case Doctype: } else if (name.equals("noscript")) { tb.insert(start); tb.transition(InHeadNoscript); } else if (name.equals("script")) { tb.markInsertionMode(); tb.transition(Text); tb.insert(start); } else if (name.equals("head")) { tb.error(this);
private boolean anythingElse(Token t, HtmlTreeBuilder tb) { tb.error(this); tb.insert(new Token.Character().data(t.toString())); return true; } },
private static void handleRcData(Token.StartTag startTag, HtmlTreeBuilder tb) { tb.tokeniser.transition(TokeniserState.Rcdata); tb.markInsertionMode(); tb.transition(Text); tb.insert(startTag); }
private static void handleRawtext(Token.StartTag startTag, HtmlTreeBuilder tb) { tb.tokeniser.transition(TokeniserState.Rawtext); tb.markInsertionMode(); tb.transition(Text); tb.insert(startTag); }
boolean 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() && t.asStartTag().normalName().equals("html")) { return tb.process(t, InBody); } else if (t.isEndTag() && t.asEndTag().normalName().equals("html")) { tb.transition(AfterAfterFrameset); } else if (t.isStartTag() && t.asStartTag().normalName().equals("noframes")) { return tb.process(t, InHead); } else if (t.isEOF()) { // cool your heels, we're complete } else { tb.error(this); return false; } return true; } },
boolean process(Token t, HtmlTreeBuilder tb) { if (t.isComment()) { tb.insert(t.asComment()); } else if (t.isDoctype() || isWhitespace(t) || (t.isStartTag() && t.asStartTag().normalName().equals("html"))) { return tb.process(t, InBody); } else if (t.isEOF()) { // nice work chuck } else if (t.isStartTag() && t.asStartTag().normalName().equals("noframes")) { return tb.process(t, InHead); } else { tb.error(this); return false; } return true; } },
boolean process(Token t, HtmlTreeBuilder tb) { if (t.isComment()) { tb.insert(t.asComment()); } else if (t.isDoctype() || isWhitespace(t) || (t.isStartTag() && t.asStartTag().normalName().equals("html"))) { return tb.process(t, InBody); } else if (t.isEOF()) { // nice work chuck } else { tb.error(this); tb.transition(InBody); return tb.process(t); } return true; } },
boolean 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 tb.process(t, InBody); } else if (name.equals("body")) { tb.insert(startTag); tb.framesetOk(false); tb.transition(InBody); } else if (name.equals("frameset")) { tb.insert(startTag); tb.transition(InFrameset); } else if (StringUtil.in(name, "base", "basefont", "bgsound", "link", "meta", "noframes", "script", "style", "title")) {
boolean process(Token t, HtmlTreeBuilder tb) { if (t.isDoctype()) { tb.error(this); return false; } else if (t.isComment()) { tb.insert(t.asComment()); } else if (isWhitespace(t)) { return true; // ignore whitespace } else if (t.isStartTag() && t.asStartTag().normalName().equals("html")) { tb.insert(t.asStartTag()); tb.transition(BeforeHead); } else if (t.isEndTag() && (StringUtil.in(t.asEndTag().normalName(), "head", "body", "html", "br"))) { return anythingElse(t, tb); } else if (t.isEndTag()) { tb.error(this); return false; } else { return anythingElse(t, tb); } return true; }
boolean process(Token t, HtmlTreeBuilder tb) { if (isWhitespace(t)) { return true; } else if (t.isComment()) { tb.insert(t.asComment()); } else if (t.isDoctype()) { tb.error(this); return false; } else if (t.isStartTag() && t.asStartTag().normalName().equals("html")) { return InBody.process(t, tb); // does not transition } else if (t.isStartTag() && t.asStartTag().normalName().equals("head")) { Element head = tb.insert(t.asStartTag()); tb.setHeadElement(head); tb.transition(InHead); } else if (t.isEndTag() && (StringUtil.in(t.asEndTag().normalName(), "head", "body", "html", "br"))) { tb.processStartTag("head"); return tb.process(t); } else if (t.isEndTag()) { tb.error(this); return false; } else { tb.processStartTag("head"); return tb.process(t); } return true; } },
boolean process(Token t, HtmlTreeBuilder tb) { if (isWhitespace(t)) { return tb.process(t, InBody); } else if (t.isComment()) { tb.insert(t.asComment()); // into html node } else if (t.isDoctype()) { tb.error(this); return false; } else if (t.isStartTag() && t.asStartTag().normalName().equals("html")) { return tb.process(t, InBody); } else if (t.isEndTag() && t.asEndTag().normalName().equals("html")) { if (tb.isFragmentParsing()) { tb.error(this); return false; } else { tb.transition(AfterAfterBody); } } else if (t.isEOF()) { // chillax! we're done } else { tb.error(this); tb.transition(InBody); return tb.process(t); } return true; } },
boolean process(Token t, HtmlTreeBuilder tb) { if (t.isCharacter()) { tb.insert(t.asCharacter()); } else if (t.isEOF()) { tb.error(this); // if current node is script: already started tb.pop(); tb.transition(tb.originalState()); return tb.process(t); } else if (t.isEndTag()) { // if: An end tag whose tag name is "script" -- scripting nesting level, if evaluating scripts tb.pop(); tb.transition(tb.originalState()); } return true; } },
tb.insert(new Token.Character().data(character));
boolean process(Token t, HtmlTreeBuilder tb) { if (isWhitespace(t)) { return true; // ignore whitespace } else if (t.isComment()) { tb.insert(t.asComment()); } else if (t.isDoctype()) { // todo: parse error check on expected doctypes // todo: quirk state check on doctype ids Token.Doctype d = t.asDoctype(); DocumentType doctype = new DocumentType( tb.settings.normalizeTag(d.getName()), d.getPublicIdentifier(), d.getSystemIdentifier()); doctype.setPubSysKey(d.getPubSysKey()); tb.getDocument().appendChild(doctype); if (d.isForceQuirks()) tb.getDocument().quirksMode(Document.QuirksMode.quirks); tb.transition(BeforeHtml); } else { // todo: check not iframe srcdoc tb.transition(BeforeHtml); return tb.process(t); // re-process token } return true; } },