Jsoup源码以及chm文件

package org.jsoup.parser; import org.jsoup.helper.StringUtil; import org.jsoup.nodes.*; import java.util.ArrayList; /** * The Tree Builder's current state. Each state embodies the processing for the state, and transitions to other states. */ enum HtmlTreeBuilderState { Initial { 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.getPubSysKey(), d.getPublicIdentifier(), d.getSystemIdentifier(), tb.getBaseUri()); 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; } }, BeforeHtml { 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; } private boolean anythingElse(Token t, HtmlTreeBuilder tb) { tb.insertStartTag("html"); tb.transition(BeforeHead); return tb.process(t); } }, BeforeHead { 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; } }, InHead { boolean process(Token t, HtmlTreeBuilder tb) { if (isWhitespace(t)) { tb.insert(t.asCharacter()); return true; } switch (t.type) { case Comment: tb.insert(t.asComment()); break; case Doctype: tb.error(this); return false; case StartTag: Token.StartTag start = t.asStartTag(); String name = start.normalName(); 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 the frist time it is seen if (name.equals("base") && el.hasAttr("href")) tb.maybeSetBaseUri(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.tokeniser.transition(TokeniserState.ScriptData); tb.markInsertionMode(); tb.transition(Text); tb.insert(start); } else if (name.equals("head")) { tb.error(this); return false; } else { return anythingElse(t, tb); } break; case EndTag: Token.EndTag end = t.asEndTag(); name = end.normalName(); 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; } private boolean anythingElse(Token t, TreeBuilder tb) { tb.processEndTag("head"); return tb.process(t); } }, InHeadNoscript { boolean process(Token t, HtmlTreeBuilder tb) { if (t.isDoctype()) { tb.error(this); } else if (t.isStartTag() && t.asStartTag().normalName().equals("html")) { return tb.process(t, InBody); } else if (t.isEndTag() && t.asEndTag().normalName().equals("noscript")) { tb.pop(); tb.transition(InHead); } else if (isWhitespace(t) || t.isComment() || (t.isStartTag() && StringUtil.in(t.asStartTag().normalName(), "basefont", "bgsound", "link", "meta", "noframes", "style"))) { return tb.process(t, InHead); } else if (t.isEndTag() && t.asEndTag().normalName().equals("br")) { return anythingElse(t, tb); } else if ((t.isStartTag() && StringUtil.in(t.asStartTag().normalName(), "head", "noscript")) || t.isEndTag()) { tb.error(this); return false; } else { return anythingElse(t, tb); } return true; } private boolean anythingElse(Token t, HtmlTreeBuilder tb) { tb.error(this); tb.insert(new Token.Character().data(t.toStrin