xmlparser.java

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

      addEntityReference(_buf.toString());

      ch = _reader.read();

      return ch;
    } else if (_strictXml) {
      throw error(L.l("expected name at {0}", badChar(ch)));
    } else {
      addText('&');
      return ch;
    }
  }

  private int parseCharacterReference()
    throws IOException, SAXException
  {
    int ch = _reader.read();

    int radix = 10;
    if (ch == 'x') {
      radix = 16;
      ch = _reader.read();
    }

    int value = 0;
    for (; ch != ';'; ch = _reader.read()) {
      if (ch >= '0' && ch <= '9')
	value = radix * value + ch - '0';
      else if (radix == 16 && ch >= 'a' && ch <= 'f')
	value = radix * value + ch - 'a' + 10;
      else if (radix == 16 && ch >= 'A' && ch <= 'F')
	value = radix * value + ch - 'A' + 10;
      else
	throw error(L.l("malformed entity ref at {0}", badChar(ch)));
    }

    if (value > 0xffff)
      throw error(L.l("malformed entity ref at {0}", "" + value));

    // xml/0072
    if (_strictCharacters && ! isChar(value))
      throw error(L.l("illegal character ref at {0}", badChar(value)));

    return value;
  }

  /**
   * Looks up a named entity reference, filling the text.
   */
  private void addEntityReference(String name)
    throws IOException, SAXException
  {
    boolean expand = ! _entitiesAsText || _hasDoctype || ! _switchToXml;
    // XXX: not quite the right logic.  There should be a soft expandEntities

    if (! expand) {
      addText("&" + name + ";");
      return;
    }
    
    int ch = _entities.getEntity(name);
    if (ch >= 0 && ch <= 0xffff) {
      addText((char) ch);
      return;
    }

    QEntity entity = _dtd == null ? null : _dtd.getEntity(name);

    if (! _expandEntities) {
      addText("&" + name + ";");
      return;
    }

    if (entity == null && (_dtd == null || _dtd.getName() == null ||
                           ! _dtd.isExternal())) {
      if (_strictXml)
	throw error(L.l("`&{0};' is an unknown entity.  XML predefines only `&lt;', `&amp;', `&gt;', `&apos;' and  `&quot;'. All other entities must be defined in an &lt;!ENTITY> definition in the DTD.", name));
      else {
        if (expand && _contentHandler instanceof DOMBuilder) {
          appendText();
          ((DOMBuilder) _contentHandler).entityReference(name);
        }
        else
          addText("&" + name + ";");
      }
    }
    else if (entity != null) {
      if (expand && entity._isSpecial && entity._value != null)
	addText(entity._value);
      else if (entity.getSystemId() != null) {
        if (pushSystemEntity(entity)) {
        }
        /* XXX:??
        else if (strictXml) {
          throw error(L.l("can't open external entity at `&{0};'", name));
        }
        */
        else if (_contentHandler instanceof DOMBuilder) {
          appendText();
          ((DOMBuilder) _contentHandler).entityReference(name);
        }
        else
          addText("&" + name + ";");
      }
      else if (expand && entity._value != null)
	setMacro(entity._value);
      else
        addText("&" + name + ";");
    }
    else {
      if (expand && _contentHandler instanceof DOMBuilder) {
        appendText();
        ((DOMBuilder) _contentHandler).entityReference(name);
      }
      else // XXX: error?
        addText("&" + name + ";");
    }
  }

  private boolean pushSystemEntity(QEntity entity)
    throws IOException, SAXException
  {
    String publicId = entity.getPublicId();
    String systemId = entity.getSystemId();
    String value = null;
    InputSource source = null;
    ReadStream is = null;

    if (_entityResolver != null)
      source = _entityResolver.resolveEntity(publicId, systemId);

    if (source != null && source.getByteStream() != null)
      is = Vfs.openRead(source.getByteStream());
    else if (source != null && source.getCharacterStream() != null)
      is = Vfs.openRead(source.getCharacterStream());
    else if (source != null && source.getSystemId() != null &&
             _searchPath.lookup(source.getSystemId()).isFile()) {
      _owner.addDepend(_searchPath.lookup(source.getSystemId()));
      is = _searchPath.lookup(source.getSystemId()).openRead();
    }
    else if (systemId != null && ! systemId.equals("")) {
      String path = systemId;
      if (path.startsWith("file:"))
        path = path.substring(5);
      if (_searchPath != null && _searchPath.lookup(path).isFile()) {
        _owner.addDepend(_searchPath.lookup(path));
        is = _searchPath.lookup(path).openRead();
      }
    }

    if (is == null)
      return false;

    _filename = systemId;
    _systemId = systemId;

    Path oldSearchPath = _searchPath;
    Path path = is.getPath();
    if (path != null) {
      _owner.addDepend(path);
      
      if (_searchPath != null) {
        _searchPath = path.getParent();
        _reader.setSearchPath(oldSearchPath);
      }
    }

    _is = is;
    _line = 1;
    
    XmlReader oldReader = _reader;
    _reader = null;

    int ch = parseXMLDeclaration(oldReader);
    unread(ch);

    return true;
  }

  /**
   * Parses an attribute value.
   *
   * <pre>
   * value ::= '[^']*'
   *       ::= "[^"]*"
   *       ::= [^ />]*
   * </pre>
   *
   * @param value the CharBuffer which will contain the value.
   * @param ch the next character from the input stream.
   * @param isGeneral true if general entities are allowed.
   *
   * @return the following character from the input stream
   */
  private int parseValue(CharBuffer value, int ch, boolean isGeneral)
    throws IOException, SAXException
  {
    int end = ch;

    value.clear();

    if (end == '\'' || end == '"')
      ch = _reader.read();
    else if (_strictAttributes) {
      value.append((char) end);
      for (ch = _reader.read();
           ch >= 0 && XmlChar.isNameChar(ch);
           ch = _reader.read())
        value.append((char) ch);
      
      throw error(L.l("XML attribute value must be quoted at `{0}'.  XML attribute syntax is either attr=\"value\" or attr='value'.",
                      value));
    }
    else
      end = 0;

    while (ch != -1 && (end != 0 && ch != end ||
			end == 0 && isAttributeChar(ch))) {
      if (end == 0 && ch == '/') {
	ch = _reader.read();
	if (! isWhitespace(ch) && ch != '>') {
	  value.append('/');
	  value.append((char) ch);
	}
	else {
          unread(ch);
	  return '/';
	}
      }
      else if (ch == '&' && ! _entitiesAsText) {
	if ((ch = _reader.read()) == '#')
	  value.append((char) parseCharacterReference());
	else if (! isGeneral) {
	  value.append('&');
	  value.append((char) ch);
	}
        else if (XmlChar.isNameStart(ch)) {
	  ch = _reader.parseName(_buf, ch);
	  String name = _buf.toString();

	  if (ch != ';' && _strictXml)
	    throw error(L.l("expected `{0}' at {1}", ";", badChar(ch)));
	  else if (ch != ';') {
	    value.append('&');
	    value.append(name);
	    continue;
	  } else {
            int lookup = _entities.getEntity(name);

            if (lookup >= 0 && lookup <= 0xffff) {
              ch = _reader.read();
              value.append((char) lookup);
              continue;
            }
            
	    QEntity entity = _dtd == null ? null : _dtd.getEntity(name);
	    if (entity != null && entity._value != null)
              setMacroAttr(entity._value);
	    else if (_strictXml)
	      throw error(L.l("expected local reference at `&{0};'", name));
	    else {
	      value.append('&');
	      value.append(name);
	      value.append(';');
	    }
	  }
	}
      }
      else if (ch == '%' && ! isGeneral) {
        ch = _reader.read();

        if (! XmlChar.isNameStart(ch)) {
          value.append('%');
          continue;
        }
        else {
          ch = _reader.parseName(_buf, ch);

          if (ch != ';')
            throw error(L.l("expected `{0}' at {1}", ";", badChar(ch)));
          else
            addPEReference(value, _buf.toString());
        }
      } 
      else if (ch == '<' && _isJsp) {
        value.append('<');
        
        ch = _reader.read();

        if (ch != '%')
          continue;

        value.append('%');

        ch = _reader.read();
        while (ch >= 0) {
          if (ch == '%') {
            ch = _reader.read();
            if (ch == '>') {
              value.append("%>");
              break;
            }
            else
              value.append('%');
          }
          else {
            value.append((char) ch);
            ch = _reader.read();
          }
        }
      }
      else if (isGeneral) {
        if (ch == '\r') {
          ch = _reader.read();
          if (ch != '\n') {
            value.append('\n');
            continue;
          }
        }
        value.append((char) ch);
      }
      else if (ch == '\r') {
	value.append(' ');
        
        if ((ch = _reader.read()) != '\n')
          continue;
      }
      else if (ch == '\n')
	value.append(' ');
      else
	value.append((char) ch);

      ch = _reader.read();
    }

    if (end != 0)
      ch = _reader.read();

    return ch;
  }

  private boolean isAttributeChar(int ch)
  {
    switch (ch) {
    case ' ': case '\t': case '\n': case '\r':
      return false;
    case '<': case '>': case '\'':case '"': case '=':
      return false;
    default:
      return true;
    }
  }

  private void parsePcdata(QNode node) throws IOException, SAXException
  {
    int ch;
    String tail = "</" + node.getNodeName() + ">";

    _text.clear();
    ch = _reader.read();
    if (ch == '\n')
      ch = _reader.read();

    for (; ch != -1; ch = _reader.read()) {
      addText((char) ch);

      if (_text.endsWith(tail)) {
	_text.setLength(_text.length() - tail.length());
	if (_text.length() > 1 && _text.charAt(_text.length() - 1) == '\n')
	  _text.setLength(_text.length() - 1);
	appendText();
	return;
      }
    }
    
    throw error("bad pcdata");
  }

  private int parseXMLDeclaration(XmlReader oldReader)
    throws IOException, SAXException
  {
    int startOffset = _is.getOffset();
    boolean isEBCDIC = false;
    int ch = _is.read();

    XmlReader reader = null;
    
    // utf-16 starts with \xfe \xff
    if (ch == 0xfe) {
      ch = _is.read();
      if (ch == 0xff) {
	_owner.setAttribute("encoding", "UTF-16");
	_is.setEncoding("utf-16");

        reader = new Utf16Reader(this, _is);
        
        ch = reader.read();
      }
    }
    // utf-16 rev starts with \xff \xfe
    else if (ch == 0xff) {
      ch = _is.read();
      if (ch == 0xfe) {
	_owner.setAttribute("encoding", "UTF-16");
	_is.setEncoding("utf-16");

        reader = new Utf16Reader(this, _is);
        ((Utf16Reader) reader).setReverse(true);
        
        ch = reader.read();
      }
    }
    // utf-16 can also start with \x00 <
    else if (ch == 0x00) {
      ch = _is.read();
      _owner.setAttribute("encoding", "UTF-16");
      _is.setEncoding("utf-16");
      
      reader = new Utf16Reader(this, _is);
    }
    // utf-8 BOM is \xef \xbb \xbf
    else if (ch == 0xef) {
      ch = _is.read();
      if (ch == 0xbb) {
        ch = _is.read();

        if (ch == 0xbf) {
          ch = _is.read();

          _owner.setAttribute("encoding", "UTF-8");
          _is.setEncoding("utf-8");
      
          reader = new Utf8Reader(this, _is);
        }
      }
    }
    else if (ch == 0x4c) {
      // ebcdic
      // xml/00l1
      _is.unread();
      // _is.setEncoding("cp037");
      _is.setEncoding("cp500");

      isEBCDIC = true;

      reader = new XmlReader(this, _is);

      ch = reader.read();
    }
    else {
      int ch2 = _is.read();

      if (ch2 == 0x00) {
	_owner.setAttribute("encoding", "UTF-16LE");
	_is.setEncoding("utf-16le");

        reader = new Utf16Reader(this, _is);
        ((Utf16Reader) reader).setReverse(true);
      }
      else if (ch2 > 0)
	_is.unread();
    }

    if (reader != null && reader != oldReader) {
    }
    else if (_policy instanceof HtmlPolicy ||
             _is.getSource() instanceof ReaderWriterStream) {
      reader = new XmlReader(this, _is);
    }
    else {
      reader = new Utf8Reader(this, _is);
    }

    if (ch == '\n')
      reader.setLine(2);

    reader.setSystemId(_systemId);
    if (_systemId == null)
      reader.setSystemId(_filename);
    reader.setFilename(_filename);
    reader.setPublicId(_publicId);

    reader.setNext(oldReader);

    _reader = reader;

    /* XXX: this might be too strict. */
    /*
    if (! strictXml) {
      for (; XmlChar.isWhitespace(ch); ch = reader.read()) {
      }
    }
    */

    if (ch != '<')
      return ch;

    if (parseXMLDecl(_reader) && isEBCDIC) {
      // EBCDIC requires a re-read
      _is.setOffset(startOffset);

      ch = _reader.read();
      if (ch != '<')
	throw new IllegalStateException();
      
      parseXMLDecl(_reader);
    }

    return _reader.read();    
  }

  private boolean parseXMLDecl(XmlReader reader)
    throws IOException, SAXException
  {
    int ch = reader.read();
    if (ch != '?') {
      unread((char) ch);
      unread('<');
      return false;
    }

    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")) {
      ch = parsePITail(piName, ch);
      unread(ch);