Question

It is well understood that forking a process for running Python, as CGI does, is slower than embedding Python, as WSGI does. I would like to implement an XML-RPC interface using the SimpleXMLRPCServer included in the standard Python library and I already have an implementation that works via CGI. I believe there should be a faster way. I’d like to try WSGI but first I need a request handler for WSGI and there does not appear to be one in SimpleXMLRPCServer already. Am I all wet or is there no equivalent of this in the standard library under Python 2.6, 2.7, 3.x?

Here is my initial implementation of a WSGI replacement for CGIXMLRPCRequestHandler:

from xmlrpclib import SimpleXMLRPCDispatcher


class WSGIXMLRPCRequestHandler(SimpleXMLRPCDispatcher):
  """Simple handler for XML-RPC data passed through WSGI."""

  def __init__(self, allow_none = False, encoding = None):
    SimpleXMLRPCDispatcher.__init__(self, allow_none, encoding)


  def __call__(self, environ, start_response):
    """Parse and handle a single XML-RPC request"""
    result  = []
    method  = environ['REQUEST_METHOD']
    headers = [('Content-type', 'text/html')]

    if method != 'POST':
      # Default implementation indicates an error because XML-RPC uses the POST method.
      code             = 400
      message, explain = BaseHTTPServer.BaseHTTPRequestHandler.responses[code]

      status = '%d %s' % (code, message)

      if method == 'HEAD':
        response = ''
      else:
        response = BaseHTTPServer.DEFAULT_ERROR_MESSAGE % {'code' : code, 'message' : message, 'explain' : explain}
    else:
      # Dispatch XML-RPC to implementation
      status   = '200 OK'
      request  = environ['wsgi.input'].read(int(environ['CONTENT_LENGTH']))
      response = self._marshaled_dispatch(request)

    length = len(response)

    if length > 0:
      result = [response]

    headers.append(('Content-length', str(length)))
    start_response(status, headers)

    return result

Answer 1

Take a look at this Hope it’ll help.