htmlparser.java

HTML解释器JAVA源码

/* Generated By:JavaCC: Do not edit this line. HtmlParser.java */
package com.quiotix.html.parser;

public class HtmlParser implements HtmlParserConstants {

  final static String NL = System.getProperty("line.separator");

  private static String getTokenText(Token first, Token cur) {
    Token t;
    StringBuffer sb = new StringBuffer();

    for (t=first; t != cur.next; t = t.next) {
      if (t.specialToken != null) {
        Token tt=t.specialToken;
        while (tt.specialToken != null)
          tt = tt.specialToken;
        for (; tt != null; tt = tt.next)
          sb.append(tt.image);
      };
      sb.append(t.image);
    };
    return sb.toString();
  }

  public static void main(String[] args) throws ParseException {
    HtmlParser parser = new HtmlParser(System.in);
    HtmlDocument doc = parser.HtmlDocument();
    doc.accept(new HtmlDumper(System.out));
    System.exit(0);
  }

  final public HtmlDocument HtmlDocument() throws ParseException {
  HtmlDocument.ElementSequence s;
    s = ElementSequence();
    jj_consume_token(0);
    {if (true) return new HtmlDocument(s);}
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.ElementSequence ElementSequence() throws ParseException {
  HtmlDocument.ElementSequence s = new HtmlDocument.ElementSequence();
  HtmlDocument.HtmlElement h;
    label_1:
    while (true) {
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case EOL:
      case TAG_START:
      case ENDTAG_START:
      case COMMENT_START:
      case DECL_START:
      case PCDATA:
      case PCDATA_QS:
      case PCDATA_Q:
        ;
        break;
      default:
        jj_la1[0] = jj_gen;
        break label_1;
      }
      h = Element();
                  s.addElement(h);
    }
    {if (true) return s;}
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.HtmlElement Element() throws ParseException {
  HtmlDocument.HtmlElement e;
  Token text;
    if (jj_2_1(2)) {
      e = Tag();
                            {if (true) return e;}
    } else {
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case ENDTAG_START:
        e = EndTag();
                            {if (true) return e;}
        break;
      case COMMENT_START:
        e = CommentTag();
                            {if (true) return e;}
        break;
      case DECL_START:
        e = DeclTag();
                            {if (true) return e;}
        break;
      default:
        jj_la1[1] = jj_gen;
        if (jj_2_2(2)) {
          e = ScriptBlock();
                             {if (true) return e;}
        } else if (jj_2_3(2)) {
          e = StyleBlock();
                             {if (true) return e;}
        } else if (jj_2_4(2)) {
          jj_consume_token(TAG_START);
          text = jj_consume_token(LST_ERROR);
                            {if (true) return new HtmlDocument.Text("<" + text.image);}
        } else {
          switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
          case PCDATA:
            text = jj_consume_token(PCDATA);
                            {if (true) return new HtmlDocument.Text(text.image);}
            break;
          case PCDATA_QS:
            text = jj_consume_token(PCDATA_QS);
                            {if (true) return new HtmlDocument.Text(text.image);}
            break;
          case PCDATA_Q:
            text = jj_consume_token(PCDATA_Q);
                            {if (true) return new HtmlDocument.Text(text.image);}
            break;
          case EOL:
            jj_consume_token(EOL);
                            {if (true) return new HtmlDocument.Newline();}
            break;
          default:
            jj_la1[2] = jj_gen;
            jj_consume_token(-1);
            throw new ParseException();
          }
        }
      }
    }
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.Attribute Attribute() throws ParseException {
  HtmlDocument.Attribute a;
  Token t1, t2=null;
    t1 = jj_consume_token(ATTR_NAME);
    switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
    case ATTR_EQ:
      jj_consume_token(ATTR_EQ);
      t2 = jj_consume_token(ATTR_VAL);
      break;
    default:
      jj_la1[3] = jj_gen;
      ;
    }
    if (t2 == null)
      {if (true) return new HtmlDocument.Attribute(t1.image);}
    else
      {if (true) return new HtmlDocument.Attribute(t1.image, t2.image);}
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.AttributeList AttributeList() throws ParseException {
  HtmlDocument.AttributeList alist = new HtmlDocument.AttributeList();
  HtmlDocument.Attribute a;
    label_2:
    while (true) {
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case ATTR_NAME:
        ;
        break;
      default:
        jj_la1[4] = jj_gen;
        break label_2;
      }
      a = Attribute();
                   alist.addAttribute(a);
    }
    {if (true) return alist;}
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.HtmlElement Tag() throws ParseException {
  Token t, et;
  HtmlDocument.AttributeList alist;
  Token firstToken = getToken(1);
    try {
      jj_consume_token(TAG_START);
      t = jj_consume_token(TAG_NAME);
      alist = AttributeList();
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case TAG_END:
        et = jj_consume_token(TAG_END);
        break;
      case TAG_SLASHEND:
        et = jj_consume_token(TAG_SLASHEND);
        break;
      default:
        jj_la1[5] = jj_gen;
        jj_consume_token(-1);
        throw new ParseException();
      }
      HtmlDocument.Tag tag = new HtmlDocument.Tag(t.image, alist);
      if (et.kind == TAG_SLASHEND) tag.setEmpty(true);
      {if (true) return tag;}
    } catch (ParseException ex) {
    token_source.SwitchTo(DEFAULT);
    String s = getTokenText(firstToken, getNextToken());
    {if (true) return new HtmlDocument.Text(s);}
    }
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.ElementSequence BlockContents() throws ParseException {
  Token t;
  StringBuffer s = new StringBuffer();
  HtmlDocument.ElementSequence e = new HtmlDocument.ElementSequence();
    label_3:
    while (true) {
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case BLOCK_EOL:
      case BLOCK_LBR:
      case BLOCK_WORD:
        ;
        break;
      default:
        jj_la1[6] = jj_gen;
        break label_3;
      }
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case BLOCK_EOL:
        jj_consume_token(BLOCK_EOL);
      if (s.length() > 0) {
        e.addElement(new HtmlDocument.Text(s.toString()));
        s.setLength(0);
      };
      e.addElement(new HtmlDocument.Newline());
        break;
      case BLOCK_WORD:
        t = jj_consume_token(BLOCK_WORD);
                       s.append(t.image);
        break;
      case BLOCK_LBR:
        t = jj_consume_token(BLOCK_LBR);
                       s.append(t.image);
        break;
      default:
        jj_la1[7] = jj_gen;
        jj_consume_token(-1);
        throw new ParseException();
      }
    }
    if (s.length() > 0)
      e.addElement(new HtmlDocument.Text(s.toString()));
    e.addElement(new HtmlDocument.Newline());
    {if (true) return e;}
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.HtmlElement ScriptBlock() throws ParseException {
  HtmlDocument.AttributeList alist;
  HtmlDocument.ElementSequence e;
  Token firstToken = getToken(1);
    try {
      jj_consume_token(TAG_START);
      jj_consume_token(TAG_SCRIPT);
      alist = AttributeList();
      jj_consume_token(TAG_END);
      token_source.SwitchTo(LexScript);
      e = BlockContents();
      jj_consume_token(SCRIPT_END);
      {if (true) return new HtmlDocument.TagBlock("SCRIPT", alist, e);}
    } catch (ParseException ex) {
    token_source.SwitchTo(DEFAULT);
    String s = getTokenText(firstToken, getNextToken());
    {if (true) return new HtmlDocument.Text(s);}
    }
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.HtmlElement StyleBlock() throws ParseException {
  HtmlDocument.AttributeList alist;
  HtmlDocument.ElementSequence e;
  Token firstToken = getToken(1);
    try {
      jj_consume_token(TAG_START);
      jj_consume_token(TAG_STYLE);
      alist = AttributeList();
      jj_consume_token(TAG_END);
      token_source.SwitchTo(LexStyle);
      e = BlockContents();
      jj_consume_token(STYLE_END);
      {if (true) return new HtmlDocument.TagBlock("STYLE", alist, e);}
    } catch (ParseException ex) {
    token_source.SwitchTo(DEFAULT);
    String s = getTokenText(firstToken, getNextToken());
    {if (true) return new HtmlDocument.Text(s);}
    }
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.HtmlElement EndTag() throws ParseException {
  Token t;
  Token firstToken = getToken(1);
    try {
      jj_consume_token(ENDTAG_START);
      t = jj_consume_token(TAG_NAME);
      jj_consume_token(TAG_END);
      {if (true) return new HtmlDocument.EndTag(t.image);}
    } catch (ParseException ex) {
    token_source.SwitchTo(DEFAULT);
    String s = getTokenText(firstToken, getNextToken());
    {if (true) return new HtmlDocument.Text(s);}
    }
    throw new Error("Missing return statement in function");
  }

  final public HtmlDocument.Comment CommentTag() throws ParseException {
  Token t;
  StringBuffer s = new StringBuffer("--");
    jj_consume_token(COMMENT_START);
    label_4:
    while (true) {
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case DASH:
      case COMMENT_EOL:
      case COMMENT_WORD:
        ;
        break;
      default:
        jj_la1[8] = jj_gen;
        break label_4;
      }
      switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
      case DASH:
        t = jj_consume_token(DASH);
               s.append(t.image);
        break;
      case COMMENT_EOL:
        jj_consume_token(COMMENT_EOL);
                       s.append(NL);
        break;
      case COMMENT_WORD:
        t = jj_consume_token(COMMENT_WORD);
                         s.append(t.image);
        break;
      default:
        jj_la1[9] = jj_gen;
        jj_consume_token(-1);
        throw new ParseException();
      }
    }
    switch ((jj_ntk==-1)?jj_ntk():jj_ntk) {
    case 0:
      jj_consume_token(0);
      break;
    case COMMENT_END:
      jj_consume_token(COMMENT_END);
      break;
    default:
      jj_la1[10] = jj_gen;
      jj_consume_token(-1);
      throw new ParseException();
    }