Code feedback

Tor Erik Soenvisen toreriks at hotmail.com
Fri Nov 17 09:24:42 EST 2006


Hi, all I would like some feedback on a multithreaded HTTP server I've 
written. The server serves python-scripts by dynamically loading scripts
in the same directory as itself. When a request is made to one of these
scripts the script is executed and its output is returned to the 
requester.

Here is the server code, HTTPServer.py:

# Basic, threaded HTTP server, serving requests via python scripts
# Author: Tor Erik Soenvisen

# Std lib imports
import BaseHTTPServer, os, threading, Queue

class HTTPServer(BaseHTTPServer.HTTPServer):
    """ A threaded HTTP server """

    port = 80
    # Maximum requests on hold
    request_queue_size = 250
    # Maximum requests serviced concurrently
    workers = 20
    
    def __init__(self, port=None, reqSize=None, workers=None):

        if port is not None: self.port = port
        if reqSize is not None: self.request_queue_size = reqSize
        if workers is not None: self.workers = workers
        
        self.sharedPath = os.path.dirname(os.path.abspath(__file__))
        # self.sharedPath must be set before calling self.getHandlers
        self.handler = self.getHandlers()
        self.jobQueue = Queue.Queue(self.request_queue_size)
        addr = ('', self.port)
        BaseHTTPServer.HTTPServer.__init__(self, addr, 
HTTPRequestHandler)

        # Exit threads when application exits
        self.daemon_threads = True        
 
        self.createThreadPool()
        self.serve_forever()

    def getHandlers(self):
        """ Imports all python scripts in the current directory except 
this one.
            These scripts are the response generators corresponding to 
all valid
            path requests. The idea is to comprise something similar to 
a
            lightweight CGI, where each script generate HTTP responses
            to HTTP requests
        """
        import inspect
        # 1. List all files in current directory
        # 2. Skip files not having a .py extension, in addition to this 
file
        # 3. Slice of .py extension
        handler = dict.fromkeys([x[:-3] for x in os.listdir
(self.sharedPath) \
                                 if x.endswith('.py') and \
                                 x != os.path.basename(__file__)])
        for name in handler:
            handler[name] = __import__(name)
            # Make sure the handler contains a run function accepting at 
least
            # one parameter
            if not hasattr(handler[name], 'run') or \
               len(inspect.getargspec(handler[name].run)[0]) != 2:
                print 'Handler %s.py dropped because it lacks ' % name + 
\
                      'a function named run accepting two arguments'
                del handler[name]

        return handler
    
    def createThreadPool(self):
        """ Creates pool of worker threads, servicing requests """
        self.tpool = []
        for i in range(self.workers):
            self.tpool.append(threading.Thread
(target=self.process_request_thread,
                                               args=()))
            if self.daemon_threads:
                self.tpool[-1].setDaemon(1)
            self.tpool[-1].start()

    def serve_forever(self):
        """ Handle requests "forever" """
        while 1:
            self.handle_request()
        
    def process_request(self, request, client_address):
        """ Task source: dispath requests to job queue """
        self.jobQueue.put((request, client_address))

    def process_request_thread(self):
        """ Task sink: process requests from job queue """
        while 1:
            request, client_address = self.jobQueue.get()
            try:
                self.finish_request(request, client_address)
                self.close_request(request)
            except:
                self.handle_error(request, client_address)
                self.close_request(request)
    
class HTTPRequestHandler(BaseHTTPServer.BaseHTTPRequestHandler):
    """ Handles HTTP requests. GET is the only method currently 
supported """

    # Set HTTP protocol version 1.1
    protocol_version = 'HTTP/1.1'
    # Set default response headers
    responseHeaders = {'Content-Type': 'text/html',
                       'Content-Length': '',
                       'Connection': 'close'}
    
    def log_message(self, format, *args):
        """ Log function: see BaseHTTPServer.py for info on arguments 
"""
        pass
    
    def do_GET(self):
        """ Handles HTTP GET requests """

        name = self.path.lstrip('/')
        try:
            handler = self.server.handler[name].run
        except KeyError:
            self.send_error(code=404, message='File %s not found' % 
name)
        else:
            data = handler(self.headers, self.responseHeaders)
            # Set Content-Length header
            self.responseHeaders['Content-Length'] = str(len(data))
            # Send headers and content
            self.send_response(200, message='OK')            
            for name, value in self.responseHeaders.items():
                self.send_header(name, value)
            self.end_headers()
            self.wfile.write(data)
        
if __name__ == '__main__':
    """ For testing purposes """
    HTTPServer(port=8081)


Pointing the browser to http://localhost:8081/index yields Hello, world! 
when the script index.py is contained in the same directory as the 
server.

index.py:

def run(request, response):
    """ Entrypoint function, called by web-server
        Argument request is the request headers as a mimetools.Message
        instance, while response is a dictionary containing the default
        response headers.
    """
    html = '<html><head><title></title></head><body>Hello, world!</body>
</html>'
    return html

Suggestions for making the server faster, more secure, more general, 
and/or more compact will be appreciated.

Regards, Tor Erik

PS: Yes, I know there exist heaps of web-servers out there.



More information about the Python-list mailing list