Access URIs that violate RFC 2396

[ChristianGruen]

It is currently not possible to access the URI https://www.läuft-bei-dir.de/ via fn:doc and other functions, as Java’s HttpRequest.newBuilder method will reject the input.

As “invalid” URIs can be expected to occur more frequently in the future, we should think about workarounds. One way could be to manually convert the URI to its ASCII representation (should definitely be checked in more detail):

URI toASCII(String input) throws URISyntaxException {
  URI uri = new URI(input);
  String string = uri.getScheme() + "://";
  if(uri.getRawUserInfo() != null) string += uri.getRawUserInfo() + "@";
  String host = uri.getHost();
  if(host == null) {
    String auth = uri.getRawAuthority();
    String hp = auth.substring(auth.indexOf('@') + 1);
    int c = hp.lastIndexOf(':');
    host = c < 0 ? hp : hp.substring(0, c);
  }
  string += IDN.toASCII(host, IDN.ALLOW_UNASSIGNED);
  if(uri.getPort() != -1) string += ":" + uri.getPort();
  if(uri.getRawPath() != null) string += uri.getRawPath();
  if(uri.getRawQuery() != null) string += "?" + uri.getRawQuery();
  if(uri.getRawFragment() != null) string += "#" + uri.getRawFragment();
  return URI.create(new URI(string).toASCIIString());
}