Esempio n. 1
0
  /**
   * Converts an HTML document to XML.
   *
   * @param io io reference
   * @param opts html options
   * @return parser
   * @throws IOException I/O exception
   */
  private static IO toXML(final IO io, final HtmlOptions opts) throws IOException {
    // reader could not be initialized; fall back to XML
    if (READER == null) return io;

    try {
      // tries to extract the encoding from the input
      final TextInput ti = new TextInput(io);
      String enc = ti.encoding();
      final byte[] content = ti.content();

      // looks for a charset definition
      final byte[] encoding = token("charset=");
      int cs = indexOf(content, encoding);
      if (cs > 0) {
        // extracts the encoding string
        cs += encoding.length;
        int ce = cs;
        final int cl = content.length;
        while (++ce < cl && content[ce] > 0x28) ;
        enc = string(substring(content, cs, ce));
      }

      // define input
      final InputSource is = new InputSource(new ArrayInput(content));
      is.setEncoding(supported(enc) ? normEncoding(enc) : UTF8);
      // define output
      final StringWriter sw = new StringWriter();
      final XMLReader reader = (XMLReader) Reflect.get(READER);
      final Object writer = Reflect.get(WRITER, sw);

      // set TagSoup options
      if (opts.get(HtmlOptions.HTML)) {
        reader.setFeature("http://xml.org/sax/features/namespaces", false);
        opt("method", "html");
        opt("omit-xml-declaration", "yes");
      }
      if (opts.get(HtmlOptions.NONS))
        reader.setFeature("http://xml.org/sax/features/namespaces", false);
      if (opts.get(HtmlOptions.OMITXML)) opt("omit-xml-declaration", "yes");
      if (opts.get(HtmlOptions.NOBOGONS)) reader.setFeature(FEATURES + "ignore-bogons", true);
      if (opts.get(HtmlOptions.NODEFAULTS))
        reader.setFeature(FEATURES + "default-attributes", false);
      if (opts.get(HtmlOptions.NOCOLONS)) reader.setFeature(FEATURES + "translate-colons", true);
      if (opts.get(HtmlOptions.NORESTART)) reader.setFeature(FEATURES + "restart-elements", false);
      if (opts.get(HtmlOptions.IGNORABLE))
        reader.setFeature(FEATURES + "ignorable-whitespace", true);
      if (opts.get(HtmlOptions.EMPTYBOGONS)) reader.setFeature(FEATURES + "bogons-empty", true);
      if (opts.get(HtmlOptions.ANY)) reader.setFeature(FEATURES + "bogons-empty", false);
      if (opts.get(HtmlOptions.NOROOTBOGONS)) reader.setFeature(FEATURES + "root-bogons", false);
      if (opts.get(HtmlOptions.NOCDATA)) reader.setFeature(FEATURES + "cdata-elements", false);
      if (opts.get(HtmlOptions.LEXICAL))
        reader.setProperty("http://xml.org/sax/properties/lexical-handler", writer);
      if (opts.contains(HtmlOptions.METHOD)) opt("method", opts.get(HtmlOptions.METHOD));
      if (opts.contains(HtmlOptions.DOCTYPESYS))
        opt("doctype-system", opts.get(HtmlOptions.DOCTYPESYS));
      if (opts.contains(HtmlOptions.DOCTYPEPUB))
        opt("doctype-public", opts.get(HtmlOptions.DOCTYPEPUB));
      if (opts.contains(HtmlOptions.ENCODING)) is.setEncoding(opts.get(HtmlOptions.ENCODING));
      // end TagSoup options

      reader.setContentHandler((ContentHandler) writer);
      reader.parse(is);
      return new IOContent(token(sw.toString()), io.name());

    } catch (final SAXException ex) {
      Util.errln(ex);
      return io;
    }
  }
Esempio n. 2
0
  @Override
  protected boolean run() {
    final boolean create = context.user.has(Perm.CREATE);
    String path = MetaData.normPath(args[0]);
    if (path == null || path.endsWith(".")) return error(NAME_INVALID_X, args[0]);

    if (in == null) {
      final IO io = IO.get(args[1]);
      if (!io.exists() || io.isDir()) return error(RES_NOT_FOUND_X, create ? io : args[1]);
      in = io.inputSource();
      // set/add name of document
      if ((path.isEmpty() || path.endsWith("/")) && !(io instanceof IOContent)) path += io.name();
    }

    // ensure that the final name is not empty
    if (path.isEmpty()) return error(NAME_INVALID_X, path);

    // ensure that the name is not empty and contains no trailing dots
    final Data data = context.data();
    if (data.inMemory()) return error(NO_MAINMEM);

    final IOFile file = data.meta.binary(path);
    if (path.isEmpty() || path.endsWith(".") || file == null || file.isDir())
      return error(NAME_INVALID_X, create ? path : args[0]);

    // start update
    if (!data.startUpdate()) return error(DB_PINNED_X, data.meta.name);

    try {
      store(in, file);
      return info(QUERY_EXECUTED_X_X, "", perf);
    } catch (final IOException ex) {
      return error(FILE_NOT_STORED_X, Util.message(ex));
    } finally {
      data.finishUpdate();
    }
  }