xmlparser.java

来自「RESIN 3.2 最新源码」· Java 代码 · 共 2,651 行 · 第 1/5 页

      return false;
    }
          
    if (_switchToXml && _activeNode == DOC_NAME && ! _inDtd) {
      _policy = new XmlPolicy();
    }

    ch = parseAttributes(ch, false);
      
    if (ch != '?')
      throw error(L.l("expected `?' at {0}.  Processing instructions end with `?>' like <?foo ... ?>", badChar(ch)));
    if ((ch = _reader.read()) != '>')
      throw error(L.l("expected `>' at {0}.  Processing instructions end with `?>' like <?foo ... ?>", ">", badChar(ch)));

    for (int i = 0; i < _attributes.getLength(); i++) {
      QName name = _attributes.getName(i);
      String value = _attributes.getValue(i);

      if (_owner != null)
        _owner.setAttribute(name.getName(), value);

      if (name.getName().equals("encoding")) { // xml/00hb // && ! _inDtd) {
        String encoding = value;

        if (! _isStaticEncoding &&
            ! encoding.equalsIgnoreCase("UTF-8") &&
            ! encoding.equalsIgnoreCase("UTF-16") &&
            ! (_is.getSource() instanceof ReaderWriterStream)) {
	  _is.setEncoding(encoding);

	  XmlReader oldReader = _reader;
	  
	  _reader = new XmlReader(this, _is);
	  // _reader.setNext(oldReader);
	  
	  _reader.setLine(oldReader.getLine());

          _reader.setSystemId(_filename);
          _reader.setPublicId(null);
        }
      }
    }

    return true;
  }

  private int parsePI()
    throws IOException, SAXException
  {
    int ch;

    appendText();
    ch = _reader.read();
    if (! XmlChar.isNameStart(ch))
      throw error(L.l("expected name after '<?' at {0}.  Processing instructions expect a name like <?foo ... ?>", badChar(ch)));
    ch = _reader.parseName(_text, ch);

    String piName = _text.toString();
    if (! piName.equals("xml"))
      return parsePITail(piName, ch);
    else if (_switchToXml && _activeNode == DOC_NAME && ! _inDtd) {
      _policy = new XmlPolicy();
      return parsePITail(piName, ch);
    }
    else {
      throw error(L.l("<?xml ... ?> occurs after content.  The <?xml ... ?> prolog must be at the document start."));

    }
  }

  private int parsePITail(String piName, int ch)
    throws IOException, SAXException
  {
    ch = skipWhitespace(ch);

    _text.clear();
    while (ch != -1) {
      if (ch == '?') {
        if ((ch = _reader.read()) == '>')
          break;
        else
          _text.append('?');
      } else {
        _text.append((char) ch);
        ch = _reader.read();
      }
    }

    if (_inDtd) {
      QProcessingInstruction pi;
      pi = new QProcessingInstruction(piName, _text.toString());
      pi._owner = _dtd._owner;
      _dtd.appendChild(pi);
    }
    else
      _contentHandler.processingInstruction(piName, _text.toString());

    return _reader.read();
  }

  /**
   * Parses a comment.  The "&lt;!--" has already been read.
   */
  private void parseComment()
    throws IOException, SAXException
  {
    if (! _skipComments)
      appendText();

    int ch = _reader.read();

    if (ch != '-')
      throw error(L.l("expected comment at {0}", badChar(ch)));

    ch = _reader.read();

    if (! _skipComments)
      _buf.clear();
	      
  comment:
    while (ch != -1) {
      if (ch == '-') {
	ch = _reader.read();

	while (ch == '-') {
	  if ((ch = _reader.read()) == '>')
	    break comment;
	  else if (_strictComments)
	    throw error(L.l("XML forbids `--' in comments"));
	  else if (ch == '-') {
            if (! _skipComments)
              _buf.append('-');
          }
	  else {
            if (! _skipComments)
              _buf.append("--");
	    break;
          }
	}

        _buf.append('-');
      } else if (! XmlChar.isChar(ch)) {
        throw error(L.l("bad character {0}", hex(ch)));
      } else {
	_buf.append((char) ch);
	ch = _reader.read();
      }
    }

    if (_inDtd) {
      QComment comment = new QComment(_buf.toString());
      comment._owner = _dtd._owner;
      _dtd.appendChild(comment);
    }
    else if (_skipComments) {
    }
    else if (_contentHandler instanceof XMLWriter && ! _skipComments) {
      ((XMLWriter) _contentHandler).comment(_buf.toString());
      _isIgnorableWhitespace = true;
    }
    else if (_lexicalHandler != null) {
      _lexicalHandler.comment(_buf.getBuffer(), 0, _buf.getLength());
      _isIgnorableWhitespace = true;
    }
  }

  /**
   * Parses the contents of a cdata section.
   *
   * <pre>
   * cdata ::= &lt;![CDATA[ ... ]]>
   * </pre>
   */
  private void parseCdata()
    throws IOException, SAXException
  {
    int ch;

    if (_forgiving) {
      if ((ch = _reader.read()) != 'C') {
	appendText("<![" + (char) ch);
	return;
      }
      else if ((ch = _reader.read()) != 'D') {
	appendText("<![C" + (char) ch);
	return;
      }
      else if ((ch = _reader.read()) != 'A') {
	appendText("<![CD" + (char) ch);
	return;
      }
      else if ((ch = _reader.read()) != 'T') {
	appendText("<![CDA" + (char) ch);
	return;
      }
      else if ((ch = _reader.read()) != 'A') {
	appendText("<![CDAT" + (char) ch);
	return;
      }
      else if ((ch = _reader.read()) != '[') {
	appendText("<![CDATA" + (char) ch);
	return;
      }
    }
    else if ((ch = _reader.read()) != 'C' ||
             (ch = _reader.read()) != 'D' ||
             (ch = _reader.read()) != 'A' ||
             (ch = _reader.read()) != 'T' ||
             (ch = _reader.read()) != 'A' ||
             (ch = _reader.read()) != '[') {
      throw error(L.l("expected `<![CDATA[' at {0}", badChar(ch)));
    }

    ch = _reader.read();

    if (_lexicalHandler != null) {
      _lexicalHandler.startCDATA();
      appendText();
    }
    else if (! _isCoalescing)
      appendText();

  cdata:
    while (ch != -1) {
      if (ch == ']') {
	ch = _reader.read();

	while (ch == ']') {
	  if ((ch = _reader.read()) == '>')
	    break cdata;
	  else if (ch == ']')
	    addText(']');
	  else {
	    addText(']');
	    break;
	  }
	}

	addText(']');
      } else if (_strictCharacters && ! isChar(ch)) {
	throw error(L.l("expected character in cdata at {0}", badChar(ch)));
      } else {
	addText((char) ch);
	ch = _reader.read();
      }
    }
    
    if (_lexicalHandler != null) {
      appendText();
      _lexicalHandler.endCDATA();
    }
    else if (! _isCoalescing)
      appendText();
  }

  /**
   * Ignores content to the ']]>'
   */
  private void parseIgnore()
    throws IOException, SAXException
  {
    int ch = read();

    while (ch >= 0) {
      if (ch != ']') {
	ch = read();
      }
      else if ((ch = read()) != ']') {
      }
      else if ((ch = read()) == '>')
	return;
    }
  }

  private int parseContentSpec(QElementDef def, int ch)
    throws IOException, SAXException
  {
    ch = expandPE(ch);
    
    if (XmlChar.isNameStart(ch)) {
      ch = _reader.parseName(_text, ch);
      String name = _text.toString();

      if (name.equals("EMPTY")) {
	def._content = "EMPTY";
	return ch;
      }
      else if (name.equals("ANY")) {
	def._content = "ANY";
	return ch;
      }
      else
	throw error(L.l("expected EMPTY or ANY at `{0}'", name));
    }
    else if (ch != '(') {
      throw error(L.l("expected grammar definition starting with '(' at {0}.  <!ELEMENT> definitions have the syntax <!ELEMENT name - - (grammar)>", badChar(ch)));
    }
    else {
      QContentParticle cp = new QContentParticle();
      def._content = cp;

      return parseContentParticle(cp, true);
    }
  }

  /**
   * Parses a content-particle, i.e. a grammer particle in the DTD
   * regexp.
   */
  private int parseContentParticle(QContentParticle cp, boolean isTop)
    throws IOException, SAXException
  {
    boolean hasCdata = false;
    cp._separator = 0;
    cp._repeat = 0;
    int ch;

    ch = expandPE(_reader.read());
    
    for (; ch != -1; ch = expandPE(ch)) {
      if (ch == '(') {
	QContentParticle child = new QContentParticle();
	cp.addChild(child);
	
	ch = parseContentParticle(child, false);
      }
      else if (XmlChar.isNameStart(ch)) {
	ch = _reader.parseName(_text, ch);
	cp.addChild(_text.toString());
      }
      else if (ch == '#') {
	ch = _reader.parseName(_text, _reader.read());
	String name = _text.toString();

	if (_strictXml && cp._children.size() != 0)
	  throw error(L.l("`#{0}' must occur first", name));
	if (_strictXml && ! isTop)
	  throw error(L.l("`#{0}' may only occur at top level", name));

	if (name.equals("PCDATA"))
	  cp.addChild("#PCDATA");
	else
	  throw error(L.l("illegal content particle at `#{0}'", name));

	hasCdata = true;
      }
      else
	throw error(L.l("expected content particle at {0}", badChar(ch)));

      ch = expandPE(ch);

      if (ch == '?' || ch == '*' || ch == '+') {
        Object child = cp.getChild(cp.getChildSize() - 1);
        if (child instanceof QContentParticle) {
          QContentParticle cpChild = (QContentParticle) child;
          cpChild._repeat = ch;
        }
        else {
          QContentParticle cpChild = new QContentParticle();
          cpChild.addChild(child);
          cpChild._repeat = ch;
          cp.setChild(cp.getChildSize() - 1, cpChild);
        }

        ch = expandPE(_reader.read());
      }

      if (ch == ')')
	break;
      else if (cp._separator == 0) {
	if (ch == '|') 
	  cp._separator = ch;
	else if (hasCdata && _strictXml)
	  throw error(L.l("#PCDATA must be separated by `|' at {0}",
                          badChar(ch)));
	else if (ch == ',')
	  cp._separator = ch;
	else if (! _strictXml && ch =='&')
	  cp._separator = ch;
	else
	  throw error(L.l("expected separator at {0}", badChar(ch)));

	ch = _reader.read();
      } else if (ch != cp._separator)
	throw error(L.l("expected `{0}' at {1}",
                        "" + (char) cp._separator, badChar(ch)));
      else
	ch = _reader.read();
    }

    ch = expandPE(_reader.read());

    if (_strictXml && hasCdata && (ch == '+' || ch == '?'))
      throw error(L.l("pcdata clause can not have {0}", badChar(ch)));
    else if (ch == '*' || ch == '+' || ch == '?') {
      cp._repeat = ch;
      return _reader.read();
    }
    else
      return ch;
  }

  private int expandPE(int ch)
    throws IOException, SAXException
  {
    ch = skipWhitespace(ch);
    
    while (ch == '%') {
      parsePEReference();
      ch = skipWhitespace(_reader.read());
    }

    return ch;
  }

  /**
   * Parses a PE reference %foo; and inserts the macro text to the input
   * stream.
   */
  private void parsePEReference()
    throws IOException, SAXException
  {
    int ch = _reader.parseName(_buf, _reader.read());

    if (ch != ';')
      throw error(L.l("`%{0};' expects `;' at {1}.  Parameter entities have a `%name;' syntax.", _buf, badChar(ch)));

    addPEReference(_text, _buf.toString());
  }

  /**
   * Expands the macro value of a PE reference.
   */
  private void addPEReference(CharBuffer value, String name)
    throws IOException, SAXException
  {
    QEntity entity = _dtd.getParameterEntity(name);

    if (entity == null && ! _dtd.isExternal())
      throw error(L.l("`%{0};' is an unknown parameter entity.  Parameter entities must be defined in an <!ENTITY> declaration before use.", name));
    else if (entity != null && entity._value != null) {
      setMacro(entity._value);
    }
    else if (entity != null && entity.getSystemId() != null) {
      pushInclude(entity.getPublicId(), entity.getSystemId());
    }
    else {
      value.append("%");
      value.append(name);
      value.append(";");
    }
  }

  /**
   * <!ELEMENT name contentspec>
   */
  private void parseElementDecl(QDocumentType doctype)
    throws IOException, SAXException
  {
    int ch = skipWhitespace(_reader.read());

    ch = _reader.parseName(_text, ch);
    String name = _text.toString();

    ch = skipWhitespace(ch);

    QElementDef def = _dtd.addElement(name);
    def.setLocation(getSystemId(), getFilename(), getLine(), getColumn());

    boolean needsStartTag = true;
    boolean needsEndTag = true;

    if (_optionalTags && (ch == 'O' || ch == '-')) {
      needsStartTag = ch == '-';

      ch = skipWhitespace(ch);

      if (ch == '0')
	needsEndTag = false;
      else if (ch == '-')
	needsEndTag = true;
      else
	throw error(L.l("unknown short tag"));
    }

    ch = parseContentSpec(def, ch);

    ch = skipWhitespace(ch);

    if (ch != '>')
      throw error(L.l("`<!ELEMENT' must close with `>' at {0}", badChar(ch)));
  }

  private static String toAttrDefault(CharBuffer text)
  {
    for (int i = 0; i < text.length(); i++) {
      int ch = text.charAt(i);

      if (ch == '"') {
	text.delete(i, i + 1);
	text.insert(i, "&#34;");
	i--;
      } else if (ch == '\'') {
	text.delete(i, i + 1);
	text.insert(i, "&#39;");
	i--;
      }
    }

    return text.toString();
  }

  /**
   * <!ATTLIST name (attr type def)*>
   */
  private void parseAttlistDecl(QDocumentType doctype)
    throws IOException, SAXException
  {
    int ch = skipWhitespace(_reader.read());

    ch = _reader.parseName(_text, ch);
    String name = _text.toString();

    ch = skipWhitespace(ch);

    QElementDef def = _dtd.addElement(name);

    while (XmlChar.isNameStart((ch = expandPE(ch)))) {
      ch = _reader.parseName(_text, ch);
      String attrName = _text.toString();