root/branches/cherrypy-3.0.x/cherrypy/wsgiserver/__init__.py

"""A high-speed, production ready, thread pooled, generic WSGI server.

Simplest example on how to use this module directly
(without using CherryPy's application machinery):

    from cherrypy import wsgiserver
    
    def my_crazy_app(environ, start_response):
        status = '200 OK'
        response_headers = [('Content-type','text/plain')]
        start_response(status, response_headers)
        return ['Hello world!\n']
    
    # Here we set our application to the script_name '/' 
    wsgi_apps = [('/', my_crazy_app)]
    
    server = wsgiserver.CherryPyWSGIServer(('localhost', 8070), wsgi_apps,
                                           server_name='localhost')
    
    # Want SSL support? Just set these attributes
    # server.ssl_certificate = <filename>
    # server.ssl_private_key = <filename>
    
    if __name__ == '__main__':
        try:
            server.start()
        except KeyboardInterrupt:
            server.stop()

This won't call the CherryPy engine (application side) at all, only the
WSGI server, which is independant from the rest of CherryPy. Don't
let the name "CherryPyWSGIServer" throw you; the name merely reflects
its origin, not it's coupling.

The CherryPy WSGI server can serve as many WSGI applications
as you want in one instance:

    wsgi_apps = [('/', my_crazy_app), ('/blog', my_blog_app)]

"""


import base64
import Queue
import os
import re
quoted_slash = re.compile("(?i)%2F")
import rfc822
import socket
try:
    import cStringIO as StringIO
except ImportError:
    import StringIO
import sys
import threading
import time
import traceback
from urllib import unquote
from urlparse import urlparse

try:
    from OpenSSL import SSL
    from OpenSSL import crypto
except ImportError:
    SSL = None

import errno
socket_errors_to_ignore = []
# Not all of these names will be defined for every platform.
for _ in ("EPIPE", "ETIMEDOUT", "ECONNREFUSED", "ECONNRESET",
          "EHOSTDOWN", "EHOSTUNREACH",
          "WSAECONNABORTED", "WSAECONNREFUSED", "WSAECONNRESET",
          "WSAENETRESET", "WSAETIMEDOUT"):
    if _ in dir(errno):
        socket_errors_to_ignore.append(getattr(errno, _))
# de-dupe the list
socket_errors_to_ignore = dict.fromkeys(socket_errors_to_ignore).keys()
socket_errors_to_ignore.append("timed out")

comma_separated_headers = ['ACCEPT', 'ACCEPT-CHARSET', 'ACCEPT-ENCODING',
    'ACCEPT-LANGUAGE', 'ACCEPT-RANGES', 'ALLOW', 'CACHE-CONTROL',
    'CONNECTION', 'CONTENT-ENCODING', 'CONTENT-LANGUAGE', 'EXPECT',
    'IF-MATCH', 'IF-NONE-MATCH', 'PRAGMA', 'PROXY-AUTHENTICATE', 'TE',
    'TRAILER', 'TRANSFER-ENCODING', 'UPGRADE', 'VARY', 'VIA', 'WARNING',
    'WWW-AUTHENTICATE']

class HTTPRequest(object):
    """An HTTP Request (and response).
    
    A single HTTP connection may consist of multiple request/response pairs.
    
    connection: the HTTP Connection object which spawned this request.
    rfile: the 'read' fileobject from the connection's socket
    ready: when True, the request has been parsed and is ready to begin
        generating the response. When False, signals the calling Connection
        that the response should not be generated and the connection should
        close.
    close_connection: signals the calling Connection that the request
        should close. This does not imply an error! The client and/or
        server may each request that the connection be closed.
    chunked_write: if True, output will be encoded with the "chunked"
        transfer-coding. This value is set automatically inside
        send_headers.
    """
    
    def __init__(self, connection):
        self.connection = connection
        self.rfile = self.connection.rfile
        self.sendall = self.connection.sendall
        self.environ = connection.environ.copy()
        
        self.ready = False
        self.started_response = False
        self.status = ""
        self.outheaders = []
        self.sent_headers = False
        self.close_connection = False
        self.chunked_write = False
    
    def parse_request(self):
        """Parse the next HTTP request start-line and message-headers."""
        # HTTP/1.1 connections are persistent by default. If a client
        # requests a page, then idles (leaves the connection open),
        # then rfile.readline() will raise socket.error("timed out").
        # Note that it does this based on the value given to settimeout(),
        # and doesn't need the client to request or acknowledge the close
        # (although your TCP stack might suffer for it: cf Apache's history
        # with FIN_WAIT_2).
        request_line = self.rfile.readline()
        if not request_line:
            # Force self.ready = False so the connection will close.
            self.ready = False
            return
        
        if request_line == "\r\n":
            # RFC 2616 sec 4.1: "...if the server is reading the protocol
            # stream at the beginning of a message and receives a CRLF
            # first, it should ignore the CRLF."
            # But only ignore one leading line! else we enable a DoS.
            request_line = self.rfile.readline()
            if not request_line:
                self.ready = False
                return
        
        server = self.connection.server
        environ = self.environ
        environ["SERVER_SOFTWARE"] = "%s WSGI Server" % server.version
        
        method, path, req_protocol = request_line.strip().split(" ", 2)
        environ["REQUEST_METHOD"] = method
        
        # path may be an abs_path (including "http://host.domain.tld");
        scheme, location, path, params, qs, frag = urlparse(path)
        
        if frag:
            self.simple_response("400 Bad Request",
                                 "Illegal #fragment in Request-URI.")
            return
        
        if scheme:
            environ["wsgi.url_scheme"] = scheme
        if params:
            path = path + ";" + params
        
        # Unquote the path+params (e.g. "/this%20path" -> "this path").
        # http://www.w3.org/Protocols/rfc2616/rfc2616-sec5.html#sec5.1.2
        #
        # But note that "...a URI must be separated into its components
        # before the escaped characters within those components can be
        # safely decoded." http://www.ietf.org/rfc/rfc2396.txt, sec 2.4.2
        atoms = [unquote(x) for x in quoted_slash.split(path)]
        path = "%2F".join(atoms)
        
        if path == "*":
            # This means, of course, that the last wsgi_app (shortest path)
            # will always handle a URI of "*".
            environ["SCRIPT_NAME"] = ""
            environ["PATH_INFO"] = "*"
            self.wsgi_app = server.mount_points[-1][1]
        else:
            for mount_point, wsgi_app in server.mount_points:
                # The mount_points list should be sorted by length, descending.
                if path.startswith(mount_point + "/") or path == mount_point:
                    environ["SCRIPT_NAME"] = mount_point
                    environ["PATH_INFO"] = path[len(mount_point):]
                    self.wsgi_app = wsgi_app
                    break
            else:
                self.simple_response("404 Not Found")
                return
        
        # Note that, like wsgiref and most other WSGI servers,
        # we unquote the path but not the query string.
        environ["QUERY_STRING"] = qs
        
        # Compare request and server HTTP protocol versions, in case our
        # server does not support the requested protocol. Limit our output
        # to min(req, server). We want the following output:
        #     request    server     actual written   supported response
        #     protocol   protocol  response protocol    feature set
        # a     1.0        1.0           1.0                1.0
        # b     1.0        1.1           1.1                1.0
        # c     1.1        1.0           1.0                1.0
        # d     1.1        1.1           1.1                1.1
        # Notice that, in (b), the response will be "HTTP/1.1" even though
        # the client only understands 1.0. RFC 2616 10.5.6 says we should
        # only return 505 if the _major_ version is different.
        rp = int(req_protocol[5]), int(req_protocol[7])
        sp = int(server.protocol[5]), int(server.protocol[7])
        if sp[0] != rp[0]:
            self.simple_response("505 HTTP Version Not Supported")
            return
        # Bah. "SERVER_PROTOCOL" is actually the REQUEST protocol.
        environ["SERVER_PROTOCOL"] = req_protocol
        # set a non-standard environ entry so the WSGI app can know what
        # the *real* server protocol is (and what features to support).
        # See http://www.faqs.org/rfcs/rfc2145.html.
        environ["ACTUAL_SERVER_PROTOCOL"] = server.protocol
        self.response_protocol = "HTTP/%s.%s" % min(rp, sp)
        
        # If the Request-URI was an absoluteURI, use its location atom.
        if location:
            environ["SERVER_NAME"] = location
        
        # then all the http headers
        try:
            self.read_headers()
        except ValueError, ex:
            self.simple_response("400 Bad Request", repr(ex.args))
            return
        
        creds = environ.get("HTTP_AUTHORIZATION", "").split(" ", 1)
        environ["AUTH_TYPE"] = creds[0]
        if creds[0].lower() == 'basic':
            user, pw = base64.decodestring(creds[1]).split(":", 1)
            environ["REMOTE_USER"] = user
        
        # Persistent connection support
        if self.response_protocol == "HTTP/1.1":
            if environ.get("HTTP_CONNECTION", "") == "close":
                self.close_connection = True
        else:
            # HTTP/1.0
            if environ.get("HTTP_CONNECTION", "") != "Keep-Alive":
                self.close_connection = True
        
        # Transfer-Encoding support
        te = None
        if self.response_protocol == "HTTP/1.1":
            te = environ.get("HTTP_TRANSFER_ENCODING")
            if te:
                te = [x.strip().lower() for x in te.split(",") if x.strip()]
        
        read_chunked = False
        
        if te:
            for enc in te:
                if enc == "chunked":
                    read_chunked = True
                else:
                    # Note that, even if we see "chunked", we must reject
                    # if there is an extension we don't recognize.
                    self.simple_response("501 Unimplemented")
                    self.close_connection = True
                    return
        
        if read_chunked:
            if not self.decode_chunked():
                return
        
        # From PEP 333:
        # "Servers and gateways that implement HTTP 1.1 must provide
        # transparent support for HTTP 1.1's "expect/continue" mechanism.
        # This may be done in any of several ways:
        #   1. Respond to requests containing an Expect: 100-continue request
        #      with an immediate "100 Continue" response, and proceed normally.
        #   2. Proceed with the request normally, but provide the application
        #      with a wsgi.input stream that will send the "100 Continue"
        #      response if/when the application first attempts to read from
        #      the input stream. The read request must then remain blocked
        #      until the client responds.
        #   3. Wait until the client decides that the server does not support
        #      expect/continue, and sends the request body on its own.
        #      (This is suboptimal, and is not recommended.)
        #
        # We used to do 3, but are now doing 1. Maybe we'll do 2 someday,
        # but it seems like it would be a big slowdown for such a rare case.
        if environ.get("HTTP_EXPECT", "") == "100-continue":
            self.simple_response(100)
        
        self.ready = True
    
    def read_headers(self):
        """Read header lines from the incoming stream."""
        environ = self.environ
        
        while True:
            line = self.rfile.readline()
            if not line:
                # No more data--illegal end of headers
                raise ValueError("Illegal end of headers.")
            
            if line == '\r\n':
                # Normal end of headers
                break
            
            if line[0] in ' \t':
                # It's a continuation line.
                v = line.strip()
            else:
                k, v = line.split(":", 1)
                k, v = k.strip().upper(), v.strip()
                envname = "HTTP_" + k.replace("-", "_")
            
            if k in comma_separated_headers:
                existing = environ.get(envname)
                if existing:
                    v = ", ".join((existing, v))
            environ[envname] = v
        
        ct = environ.pop("HTTP_CONTENT_TYPE", None)
        if ct:
            environ["CONTENT_TYPE"] = ct
        cl = environ.pop("HTTP_CONTENT_LENGTH", None)
        if cl:
            environ["CONTENT_LENGTH"] = cl
    
    def decode_chunked(self):
        """Decode the 'chunked' transfer coding."""
        cl = 0
        data = StringIO.StringIO()
        while True:
            line = self.rfile.readline().strip().split(";", 1)
            chunk_size = int(line.pop(0), 16)
            if chunk_size <= 0:
                break
##            if line: chunk_extension = line[0]
            cl += chunk_size
            data.write(self.rfile.read(chunk_size))
            crlf = self.rfile.read(2)
            if crlf != "\r\n":
                self.simple_response("400 Bad Request",
                                     "Bad chunked transfer coding "
                                     "(expected '\\r\\n', got %r)" % crlf)
                return
        
        # Grab any trailer headers
        self.read_headers()
        
        data.seek(0)
        self.environ["wsgi.input"] = data
        self.environ["CONTENT_LENGTH"] = str(cl) or ""
        return True
    
    def respond(self):
        """Call the appropriate WSGI app and write its iterable output."""
        response = self.wsgi_app(self.environ, self.start_response)
        try:
            for chunk in response:
                # "The start_response callable must not actually transmit
                # the response headers. Instead, it must store them for the
                # server or gateway to transmit only after the first
                # iteration of the application return value that yields
                # a NON-EMPTY string, or upon the application's first
                # invocation of the write() callable." (PEP 333)
                if chunk:
                    self.write(chunk)
        finally:
            if hasattr(response, "close"):
                response.close()
        if (self.ready and not self.sent_headers
                and not self.connection.server.interrupt):
            self.sent_headers = True
            self.send_headers()
        if self.chunked_write:
            self.sendall("0\r\n\r\n")
    
    def simple_response(self, status, msg=""):
        """Write a simple response back to the client."""
        status = str(status)
        buf = ["%s %s\r\n" % (self.connection.server.protocol, status),
               "Content-Length: %s\r\n" % len(msg)]
        
        if status[:3] == "413" and self.response_protocol == 'HTTP/1.1':
            # Request Entity Too Large
            self.close_connection = True
            buf.append("Connection: close\r\n")
        
        buf.append("\r\n")
        if msg:
            buf.append(msg)
        self.sendall("".join(buf))
    
    def start_response(self, status, headers, exc_info = None):
        """WSGI callable to begin the HTTP response."""
        if self.started_response:
            if not exc_info:
                raise AssertionError("WSGI start_response called a second "
                                     "time with no exc_info.")
            else:
                try:
                    raise exc_info[0], exc_info[1], exc_info[2]
                finally:
                    exc_info = None
        self.started_response = True
        self.status = status
        self.outheaders.extend(headers)
        return self.write
    
    def write(self, chunk):
        """WSGI callable to write unbuffered data to the client.
        
        This method is also used internally by start_response (to write
        data from the iterable returned by the WSGI application).
        """
        if not self.started_response:
            raise AssertionError("WSGI write called before start_response.")
        
        if not self.sent_headers:
            self.sent_headers = True
            self.send_headers()
        
        if self.chunked_write and chunk:
            buf = [hex(len(chunk))[2:], "\r\n", chunk, "\r\n"]
            self.sendall("".join(buf))
        else:
            self.sendall(chunk)
    
    def send_headers(self):
        """Assert, process, and send the HTTP response message-headers."""
        hkeys = [key.lower() for key, value in self.outheaders]
        status = int(self.status[:3])
        
        if status == 413:
            # Request Entity Too Large. Close conn to avoid garbage.
            self.close_connection = True
        elif "content-length" not in hkeys:
            # "All 1xx (informational), 204 (no content),
            # and 304 (not modified) responses MUST NOT
            # include a message-body." So no point chunking.
            if status < 200 or status in (204, 205, 304):
                pass
            else:
                if self.response_protocol == 'HTTP/1.1':
                    # Use the chunked transfer-coding
                    self.chunked_write = True
                    self.outheaders.append(("Transfer-Encoding", "chunked"))
                else:
                    # Closing the conn is the only way to determine len.
                    self.close_connection = True
        
        if "connection" not in hkeys:
            if self.response_protocol == 'HTTP/1.1':
                if self.close_connection:
                    self.outheaders.append(("Connection", "close"))
            else:
                if not self.close_connection:
                    self.outheaders.append(("Connection", "Keep-Alive"))
        
        if "date" not in hkeys:
            self.outheaders.append(("Date", rfc822.formatdate()))
        
        server = self.connection.server
        
        if "server" not in hkeys:
            self.outheaders.append(("Server", server.version))
        
        buf = [server.protocol, " ", self.status, "\r\n"]
        try:
            buf += [k + ": " + v + "\r\n" for k, v in self.outheaders]
        except TypeError:
            if not isinstance(k, str):
                raise TypeError("WSGI response header key %r is not a string.")
            if not isinstance(v, str):
                raise TypeError("WSGI response header value %r is not a string.")
            else:
                raise
        buf.append("\r\n")
        self.sendall("".join(buf))


class NoSSLError(Exception):
    """Exception raised when a client speaks HTTP to an HTTPS socket."""
    pass


def _ssl_wrap_method(method, is_reader=False):
    """Wrap the given method with SSL error-trapping.
    
    is_reader: if False (the default), EOF errors will be raised.
        If True, EOF errors will return "" (to emulate normal sockets).
    """
    def ssl_method_wrapper(self, *args, **kwargs):
##        print (id(self), method, args, kwargs)
        start = time.time()
        while True:
            try:
                return method(self, *args, **kwargs)
            except (SSL.WantReadError, SSL.WantWriteError):
                # Sleep and try again. This is dangerous, because it means
                # the rest of the stack has no way of differentiating
                # between a "new handshake" error and "client dropped".
                # Note this isn't an endless loop: there's a timeout below.
                time.sleep(self.ssl_retry)