def __call__(self, addr):
'''Since an instance of HttpServer is passed to the Service
class (with appropriate request_handler established during
initialization), this __call__ method is what's actually
invoked by diesel.
'''
data = None
while True:
try:
h = HttpParser()
body = []
while True:
if data:
used = h.execute(data, len(data))
if h.is_headers_complete():
body.append(h.recv_body())
if h.is_message_complete():
data = data[used:]
break
data = receive()
env = h.get_wsgi_environ()
if 'HTTP_CONTENT_LENGTH' in env:
env['CONTENT_LENGTH'] = env.pop("HTTP_CONTENT_LENGTH")
if 'HTTP_CONTENT_TYPE' in env:
env['CONTENT_TYPE'] = env.pop("HTTP_CONTENT_TYPE")
env.update({
'wsgi.version': (1, 0),
'wsgi.url_scheme': 'http', # XXX incomplete
'wsgi.input': cStringIO.StringIO(''.join(body)),
'wsgi.errors': FileLikeErrorLogger(hlog),
'wsgi.multithread': False,
'wsgi.multiprocess': False,
'wsgi.run_once': False,
'REMOTE_ADDR': addr[0],
'SERVER_NAME': HOSTNAME,
'SERVER_PORT': str(self.port),
})
req = Request(env)
resp = self.request_handler(req)
if 'Server' not in resp.headers:
resp.headers.add('Server', SERVER_TAG)
if 'Date' not in resp.headers:
resp.headers.add(
'Date',
utcnow().strftime("%a, %d %b %Y %H:%M:%S UTC"))
assert resp, "HTTP request handler _must_ return a response"
self.send_response(resp, version=h.get_version())
if (not h.should_keep_alive()) or \
resp.headers.get('Connection', '').lower() == "close" or \
resp.headers.get('Content-Length') == None:
return
except ConnectionClosed:
break
def get_headers_data(data):
""" """
parser = HttpParser()
parser.execute(data, len(data))
url = parser.get_url()
method = parser.get_method()
if method == 'CONNECT':
host, _, port = url.partition(":")
else:
url = urlparse(url)
host, _, port = url.netloc.partition(":")
port = port and port.isdigit() and int(port) or 80
return (host, port), method, parser.get_version()
def __call__(self, addr):
'''Since an instance of HttpServer is passed to the Service
class (with appropriate request_handler established during
initialization), this __call__ method is what's actually
invoked by diesel.
'''
data = None
while True:
try:
h = HttpParser()
body = []
while True:
if data:
used = h.execute(data, len(data))
if h.is_headers_complete():
body.append(h.recv_body())
if h.is_message_complete():
data = data[used:]
break
data = receive()
env = h.get_wsgi_environ()
env.update({
'wsgi.version' : (1,0),
'wsgi.url_scheme' : 'http', # XXX incomplete
'wsgi.input' : cStringIO.StringIO(''.join(body)),
'wsgi.errors' : FileLikeErrorLogger(hlog),
'wsgi.multithread' : False,
'wsgi.multiprocess' : False,
'wsgi.run_once' : False,
})
req = Request(env)
resp = self.request_handler(req)
if 'Server' not in resp.headers:
resp.headers.add('Server', SERVER_TAG)
if 'Date' not in resp.headers:
resp.headers.add('Date', utcnow().strftime("%a, %d %b %Y %H:%M:%S UTC"))
assert resp, "HTTP request handler _must_ return a response"
self.send_response(resp, version=h.get_version())
if (not h.should_keep_alive()) or \
resp.headers.get('Connection', '').lower() == "close" or \
resp.headers.get('Content-Length') == None:
return
except ConnectionClosed:
break
class HttpStream(object):
""" An HTTP parser providing higher-level access to a readable,
sequential io.RawIOBase object. You can use implementions of
http_parser.reader (IterReader, StringReader, SocketReader) or
create your own.
"""
def __init__(self, stream, kind=HTTP_BOTH, decompress=False):
""" constructor of HttpStream.
:attr stream: an io.RawIOBase object
:attr kind: Int, could be 0 to parseonly requests,
1 to parse only responses or 2 if we want to let
the parser detect the type.
"""
self.parser = HttpParser(kind=kind, decompress=decompress)
self.stream = stream
def _check_headers_complete(self):
if self.parser.is_headers_complete():
return
while True:
try:
next(self)
except StopIteration:
if self.parser.is_headers_complete():
return
raise NoMoreData("Can't parse headers")
if self.parser.is_headers_complete():
return
def _wait_status_line(self, cond):
if self.parser.is_headers_complete():
return True
data = []
if not cond():
while True:
try:
d = next(self)
data.append(d)
except StopIteration:
if self.parser.is_headers_complete():
return True
raise BadStatusLine(b"".join(data))
if cond():
return True
return True
def _wait_on_url(self):
return self._wait_status_line(self.parser.get_url)
def _wait_on_status(self):
return self._wait_status_line(self.parser.get_status_code)
def url(self):
""" get full url of the request """
self._wait_on_url()
return self.parser.get_url()
def path(self):
""" get path of the request (url without query string and
fragment """
self._wait_on_url()
return self.parser.get_path()
def query_string(self):
""" get query string of the url """
self._wait_on_url()
return self.parser.get_query_string()
def fragment(self):
""" get fragment of the url """
self._wait_on_url()
return self.parser.get_fragment()
def version(self):
self._wait_on_status()
return self.parser.get_version()
def status_code(self):
""" get status code of a response as integer """
self._wait_on_status()
return self.parser.get_status_code()
def status(self):
""" return complete status with reason """
status_code = self.status_code()
reason = status_reasons.get(int(status_code), 'unknown')
return "%s %s" % (status_code, reason)
def method(self):
""" get HTTP method as string"""
self._wait_on_status()
return self.parser.get_method()
def headers(self):
""" get request/response headers, headers are returned in a
OrderedDict that allows you to get value using insensitive
keys."""
self._check_headers_complete()
return self.parser.get_headers()
def should_keep_alive(self):
""" return True if the connection should be kept alive
"""
self._check_headers_complete()
return self.parser.should_keep_alive()
def is_chunked(self):
""" return True if Transfer-Encoding header value is chunked"""
self._check_headers_complete()
return self.parser.is_chunked()
def wsgi_environ(self, initial=None):
""" get WSGI environ based on the current request.
:attr initial: dict, initial values to fill in environ.
"""
self._check_headers_complete()
return self.parser.get_wsgi_environ()
def body_file(self, buffering=None, binary=True, encoding=None,
errors=None, newline=None):
""" return the body as a buffered stream object. If binary is
true an io.BufferedReader will be returned, else an
io.TextIOWrapper.
"""
self._check_headers_complete()
if buffering is None:
buffering = -1
if buffering < 0:
buffering = DEFAULT_BUFFER_SIZE
raw = HttpBodyReader(self)
buf = BufferedReader(raw, buffering)
if binary:
return buf
text = TextIOWrapper(buf, encoding, errors, newline)
return text
def body_string(self, binary=True, encoding=None, errors=None,
newline=None):
""" return body as string """
return self.body_file(binary=binary, encoding=encoding,
newline=newline).read()
def __iter__(self):
return self
def __next__(self):
if self.parser.is_message_complete():
raise StopIteration
# fetch data
b = bytearray(DEFAULT_BUFFER_SIZE)
recved = self.stream.readinto(b)
if recved is None:
raise NoMoreData("no more data")
del b[recved:]
to_parse = bytes(b)
# parse data
nparsed = self.parser.execute(to_parse, recved)
if nparsed != recved and not self.parser.is_message_complete():
raise ParserError("nparsed != recved (%s != %s) [%s]" % (nparsed,
recved, bytes_to_str(to_parse)))
if recved == 0:
raise StopIteration
return to_parse
next = __next__
def __call__(self, addr):
'''Since an instance of HttpServer is passed to the Service
class (with appropriate request_handler established during
initialization), this __call__ method is what's actually
invoked by diesel.
'''
data = None
while True:
try:
h = HttpParser()
body = []
while True:
if data:
used = h.execute(data, len(data))
if h.is_headers_complete():
body.append(h.recv_body())
if h.is_message_complete():
data = data[used:]
break
data = receive()
env = h.get_wsgi_environ()
if 'HTTP_CONTENT_LENGTH' in env:
env['CONTENT_LENGTH'] = env.pop("HTTP_CONTENT_LENGTH")
if 'HTTP_CONTENT_TYPE' in env:
env['CONTENT_TYPE'] = env.pop("HTTP_CONTENT_TYPE")
env.update({
'wsgi.version' : (1,0),
'wsgi.url_scheme' : 'http', # XXX incomplete
'wsgi.input' : cStringIO.StringIO(''.join(body)),
'wsgi.errors' : FileLikeErrorLogger(hlog),
'wsgi.multithread' : False,
'wsgi.multiprocess' : False,
'wsgi.run_once' : False,
'REMOTE_ADDR' : addr[0],
'SERVER_NAME' : HOSTNAME,
'SERVER_PORT': str(self.port),
})
req = Request(env)
if req.headers.get('Connection', '').lower() == 'upgrade':
req.data = data
resp = self.request_handler(req)
if 'Server' not in resp.headers:
resp.headers.add('Server', SERVER_TAG)
if 'Date' not in resp.headers:
resp.headers.add('Date', utcnow().strftime("%a, %d %b %Y %H:%M:%S UTC"))
assert resp, "HTTP request handler _must_ return a response"
self.send_response(resp, version=h.get_version())
if (not h.should_keep_alive()) or \
resp.headers.get('Connection', '').lower() == "close" or \
resp.headers.get('Content-Length') == None:
return
# Switching Protocols
if resp.status_code == 101 and hasattr(resp, 'new_protocol'):
resp.new_protocol(req)
break
except ConnectionClosed:
break
class HTTPProtocol(FlowControlMixin, asyncio.Protocol):
def __init__(self, stream_reader, callback, loop=None):
super().__init__(loop=loop)
self._stream_reader = stream_reader
self._stream_writer = None
self._callback = callback
self._task = None
self._server = None
def connection_made(self, transport):
self._parser = HttpParser()
self._stream_reader.set_transport(transport)
self._stream_writer = asyncio.StreamWriter(
transport,
self,
self._stream_reader,
self._loop,
)
# Grab the name of our socket if we have it
self._server = transport.get_extra_info("sockname")
def connection_lost(self, exc):
if exc is None:
self._stream_reader.feed_eof()
else:
self._stream_reader.set_exception(exc)
super().connection_lost(exc)
def data_received(self, data):
# Parse our incoming data with our HTTP parser
self._parser.execute(data, len(data))
# If we have not already handled the headers and we've gotten all of
# them, then invoke the callback with the headers in them.
if self._task is None and self._parser.is_headers_complete():
coro = self.dispatch(
{
"server": self._server,
"protocol": b"HTTP/" + b".".join(
str(x).encode("ascii")
for x in self._parser.get_version()
),
"method": self._parser.get_method().encode("latin1"),
"path": self._parser.get_path().encode("latin1"),
"query": self._parser.get_query_string().encode("latin1"),
"headers": self._parser.get_headers(),
},
self._stream_reader,
self._stream_writer,
)
self._task = asyncio.Task(coro, loop=self._loop)
# Determine if we have any data in the body buffer and if so feed it
# to our StreamReader
if self._parser.is_partial_body():
self._stream_reader.feed_data(self._parser.recv_body())
# Determine if we've completed the end of the HTTP request, if we have
# then we should close our stream reader because there is nothing more
# to read.
if self._parser.is_message_complete():
self._stream_reader.feed_eof()
def eof_received(self):
# We've gotten an EOF from the client, so we'll propagate this to our
# StreamReader
self._stream_reader.feed_eof()
@asyncio.coroutine
def dispatch(self, request, request_body, response):
# Get the status, headers, and body from the callback. The body must
# be iterable, and each item can either be a bytes object, or an
# asyncio coroutine, in which case we'll ``yield from`` on it to wait
# for it's value.
status, resp_headers, body = yield from self._callback(
request,
request_body,
)
# Write out the status line to the client for this request
# TODO: We probably don't want to hard code HTTP/1.1 here
response.write(b"HTTP/1.1 " + status + b"\r\n")
# Write out the headers, taking special care to ensure that any
# mandatory headers are added.
# TODO: We need to handle some required headers
for key, values in resp_headers.items():
# In order to handle headers which need to have multiple values
# like Set-Cookie, we allow the value of the header to be an
# iterable instead of a bytes object, in which case we'll write
# multiple header lines for this header.
if isinstance(values, (bytes, bytearray)):
values = [values]
for value in values:
response.write(key + b": " + value + b"\r\n")
# Before we get to the body, we need to write a blank line to separate
# the headers and the response body
response.write(b"\r\n")
for chunk in body:
# If the chunk is a coroutine, then we want to wait for the result
# before we write it.
if asyncio.iscoroutine(chunk):
chunk = yield from chunk
# Write our chunk out to the connect client
response.write(chunk)
# We've written everything in our iterator, so we want to close the
# connection.
response.close()
class HttpStream(object):
""" An HTTP parser providing higher-level access to a readable,
sequential io.RawIOBase object. You can use implementions of
http_parser.reader (IterReader, StringReader, SocketReader) or
create your own.
"""
def __init__(self, stream, kind=HTTP_BOTH, decompress=False):
""" constructor of HttpStream.
:attr stream: an io.RawIOBase object
:attr kind: Int, could be 0 to parseonly requests,
1 to parse only responses or 2 if we want to let
the parser detect the type.
"""
self.parser = HttpParser(kind=kind, decompress=decompress)
self.stream = stream
def _check_headers_complete(self):
if self.parser.is_headers_complete():
return
while True:
try:
data = self.next()
except StopIteration:
if self.parser.is_headers_complete():
return
raise NoMoreData()
if self.parser.is_headers_complete():
return
def url(self):
""" get full url of the request """
self._check_headers_complete()
return self.parser.get_url()
def path(self):
""" get path of the request (url without query string and
fragment """
self._check_headers_complete()
return self.parser.get_path()
def query_string(self):
""" get query string of the url """
self._check_headers_complete()
return self.parser.get_query_string()
def fragment(self):
""" get fragment of the url """
self._check_headers_complete()
return self.parser.get_fragment()
def version(self):
self._check_headers_complete()
return self.parser.get_version()
def status_code(self):
""" get status code of a response as integer """
self._check_headers_complete()
return self.parser.get_status_code()
def status(self):
""" return complete status with reason """
status_code = self.status_code()
reason = status_reasons.get(int(status_code), 'unknown')
return "%s %s" % (status_code, reason)
def method(self):
""" get HTTP method as string"""
self._check_headers_complete()
return self.parser.get_method()
def headers(self):
""" get request/response headers, headers are returned in a
OrderedDict that allows you to get value using insensitive
keys."""
self._check_headers_complete()
return self.parser.get_headers()
def should_keep_alive(self):
""" return True if the connection should be kept alive
"""
self._check_headers_complete()
return self.parser.should_keep_alive()
def is_chunked(self):
""" return True if Transfer-Encoding header value is chunked"""
self._check_headers_complete()
return self.parser.is_chunked()
def wsgi_environ(self, initial=None):
""" get WSGI environ based on the current request.
:attr initial: dict, initial values to fill in environ.
"""
self._check_headers_complete()
return self.parser.get_wsgi_environ()
def body_file(self,
buffering=None,
binary=True,
encoding=None,
errors=None,
newline=None):
""" return the body as a buffered stream object. If binary is
true an io.BufferedReader will be returned, else an
io.TextIOWrapper.
"""
self._check_headers_complete()
if buffering is None:
buffering = -1
if buffering < 0:
buffering = DEFAULT_BUFFER_SIZE
raw = HttpBodyReader(self)
buffer = BufferedReader(raw, buffering)
if binary:
return buffer
text = TextIOWrapper(buffer, encoding, errors, newline)
return text
def body_string(self,
binary=True,
encoding=None,
errors=None,
newline=None):
""" return body as string """
return self.body_file(binary=binary,
encoding=encoding,
newline=newline).read()
def __iter__(self):
return self
def next(self):
if self.parser.is_message_complete():
raise StopIteration
# fetch data
b = bytearray(DEFAULT_BUFFER_SIZE)
recved = self.stream.readinto(b)
if recved is None:
raise NoMoreData("no more data")
del b[recved:]
# parse data
nparsed = self.parser.execute(bytes(b), recved)
if nparsed != recved and not self.parser.is_message_complete():
raise ParserError("nparsed != recved")
if recved == 0:
raise StopIteration
return bytes(b)
def __call__(self, addr):
"""Since an instance of HttpServer is passed to the Service
class (with appropriate request_handler established during
initialization), this __call__ method is what's actually
invoked by diesel.
"""
data = None
while True:
try:
h = HttpParser()
body = []
while True:
if data:
used = h.execute(data, len(data))
if h.is_headers_complete():
body.append(h.recv_body())
if h.is_message_complete():
data = data[used:]
break
data = receive()
env = h.get_wsgi_environ()
if "HTTP_CONTENT_LENGTH" in env:
env["CONTENT_LENGTH"] = env.pop("HTTP_CONTENT_LENGTH")
if "HTTP_CONTENT_TYPE" in env:
env["CONTENT_TYPE"] = env.pop("HTTP_CONTENT_TYPE")
env.update(
{
"wsgi.version": (1, 0),
"wsgi.url_scheme": "http", # XXX incomplete
"wsgi.input": cStringIO.StringIO("".join(body)),
"wsgi.errors": FileLikeErrorLogger(hlog),
"wsgi.multithread": False,
"wsgi.multiprocess": False,
"wsgi.run_once": False,
"REMOTE_ADDR": addr[0],
"SERVER_NAME": HOSTNAME,
"SERVER_PORT": str(self.port),
}
)
req = Request(env)
if req.headers.get("Connection", "").lower() == "upgrade":
req.data = data
resp = self.request_handler(req)
if "Server" not in resp.headers:
resp.headers.add("Server", SERVER_TAG)
if "Date" not in resp.headers:
resp.headers.add("Date", utcnow().strftime("%a, %d %b %Y %H:%M:%S UTC"))
assert resp, "HTTP request handler _must_ return a response"
self.send_response(resp, version=h.get_version())
if (
(not h.should_keep_alive())
or resp.headers.get("Connection", "").lower() == "close"
or resp.headers.get("Content-Length") == None
):
return
except ConnectionClosed:
break
class HttpProxyProtocol(asyncio.Protocol):
''' Implement HTTP(S) proxy behavior. '''
def __init__(self, loop, config, token_store):
''' Constructor. '''
self._parser = HttpParser()
self._body = b''
self._config = config
self._loop = loop
self._mitm = None
self._mitm_host = None
self._token_store = token_store
self._instagram = InstagramApi(
client_id=config['Instagram']['ClientID'],
client_secret=config['Instagram']['ClientSecret'],
)
self._twitter = TwitterApi(
consumer_key=config['Twitter']['ConsumerKey'],
consumer_secret=config['Twitter']['ConsumerSecret'],
app_token=config['Twitter']['AppToken'],
app_token_secret=config['Twitter']['AppTokenSecret'],
)
def connection_made(self, transport):
''' Save a reference to the transport so that we can send a reply. '''
log.debug('Connection opened.')
self._transport = transport
def connection_lost(self, exc):
log.debug('Connection closed.')
def data_received(self, data):
''' Parse incoming HTTP request. '''
log.debug('Data received: {}'.format(data))
self._parser.execute(data, len(data))
if self._parser.is_partial_body():
self._body += self._parser.recv_body()
if self._parser.is_message_complete():
method = self._parser.get_method()
uri = self._parser.get_url()
version = self._parser.get_version()
headers = self._parser.get_headers()
content_type = headers.get('Content-type', '')
charset = _get_charset(content_type)
body = self._body.decode(charset)
log.debug('Client charset: {}'.format(charset))
log.debug('Client status: method={} uri={} version={}' \
.format(method, uri, version))
log.debug('Client headers: {}'.format(headers))
log.debug('Client body: {}...'.format(body[:1000]))
if method == 'CONNECT':
asyncio.async(self._start_mitm(uri, version))
self._parser = HttpParser()
else:
asyncio.async(
self._request_upstream(
method,
uri,
version,
headers,
body
)
)
def start_tls(self, version):
'''
Initiate TLS session with the client.
This part is completely hacky! We mess around with the
transport's internals in order to wrap the current transport in TLS.
Python doesn't have an official way to do this, although it *might*
get fixed in 3.6: http://bugs.python.org/issue23749
'''
log.debug('The proxy is starting TLS with its client.')
status_line = 'HTTP/{}.{} {} {}\r\n\r\n' \
.format(version[0], version[1], 200, 'OK')
self._transport.write(status_line.encode('ascii'))
ssl_context = ssl.SSLContext(ssl.PROTOCOL_TLSv1)
ssl_context.set_ciphers('HIGH:!aNull:!eNull')
ssl_context.load_cert_chain('ssl/server.crt', 'ssl/server.key')
original_socket = self._transport._sock
self._transport = self._loop._make_ssl_transport(
original_socket,
self,
ssl_context,
server_side=True
)
@asyncio.coroutine
def _request_upstream(self, method, uri, version, headers, body):
''' Forward the request to the upstream server. '''
try:
yield from self._request_upstream_helper(
method, uri, version, headers, body
)
except Exception:
charset = _get_charset(headers.get('Content-type', ''))
response = render_http_response(
version,
500,
'PROXY ERROR',
{'Content-type': 'text/plain; charset={}'.format(charset)},
traceback.format_exc().encode(charset)
)
self._transport.write(response)
self._transport.close()
raise
@asyncio.coroutine
def _request_upstream_helper(self, method, uri, version, headers, body):
''' Forward the request to the upstream server. '''
log.debug('_request_upstream(): method={}, uri={}' \
.format(method, uri))
if self._mitm_host:
parsed = urlparse(uri)
url = 'https://{}{}'.format(self._mitm_host, parsed.path)
else:
url = uri
token, remaining = self._token_store.dispense(url)
log.debug('Signing request with {} token: {}.'
.format(token.site, token.public))
if 'instagram' in url:
qp = parse_qs(parsed.query)
qp['access_token'] = token.public
qp['sig'] = self._instagram.oauth_sign(
method=method,
url=url,
token=token,
query_params=qp,
body_params=parse_qs(body)
)
params = ['{}={}'.format(quote(k.encode('utf8')), quote(v.encode('utf8'))) for k,v in qp.items()]
uri = '{}?{}'.format(parsed.path, '&'.join(params))
log.debug('Signed instagram URL: {}'.format(uri))
elif 'twitter' in url:
headers['Authorization'] = self._twitter.oauth_sign(
method=method,
url=url,
token=token.public,
token_secret=token.secret,
query_params=parse_qs(parsed.query),
body_params=parse_qs(body)
)
else:
raise ValueError('No signing algorithm known for URL: {}'
.format(url))
if self._mitm is None:
url = urlparse(uri)
host = url.hostname
port = url.port
if port is None:
port = 80 if url.scheme == 'http' else 443
log.debug('Connecting to upstream (plaintext).')
upstream = yield from asyncio.open_connection(host, port)
upstream_reader, upstream_writer = upstream
request = render_http_request(method, uri, version, headers, body)
upstream_writer.write(request)
response = b''
parser = HttpParser()
while True:
if not parser.is_headers_complete():
data = yield from upstream_reader.readline()
else:
data = yield from upstream_reader.read(
int(parser.get_headers()['Content-Length'])
)
log.debug('Received plaintext from upstream: {}'.format(data))
parser.execute(data, len(data))
if parser.is_partial_body():
body += parser.recv_body()
if parser.is_message_complete():
version = self._parser.get_version()
status = self._parser.get_status_code()
reason = None # For some reason, the parser doesn't expose this :(
headers = self._parser.get_headers()
if status == 200:
self._token_store.update_rate_limit(url, headers)
log.debug('Plaintext upstream status: {}'.format(status))
log.debug('Plaintext upstream headers: {}'.format(headers))
log.debug('Plaintext upstream body: {}...'.format(body[:1000]))
response = render_http_response(
version, status, reason, headers, body
)
break
upstream_writer.close()
else:
upstream_write = self._mitm.forward
request = render_http_request(method, uri, version, headers, body)
upstream_write(request)
response = yield from self._mitm.receive()
version, status, reason, headers, body = response
if status == 200:
self._token_store.update_rate_limit(token, url, headers)
response = render_http_response(
version, status, reason, headers, body
)
# Forward the upstream response to the client.
self._transport.write(response)
self._transport.close()
def _set_header(self, key, value):
''' Set a header value. '''
key = key.strip().upper()
value = value.strip()
self._headers[key] = value
@asyncio.coroutine
def _start_mitm(self, uri, version):
''' MITM a connection to the upstream server. '''
log.debug('The proxy is starting an MITM connection.')
host, port = uri.split(':')
port = int(port)
self._mitm_host = host
_, self._mitm = yield from self._loop.create_connection(
lambda: MitmProtocol(self._loop, version, self),
host,
port,
ssl = ssl.create_default_context()
)
class MitmProtocol(asyncio.Protocol):
''' Handles details of MITMing a TLS connection. '''
def __init__(self, loop, http_version, proxy):
''' Constructor. '''
self._http_version = http_version
self._loop = loop
self._parser = HttpParser()
self._proxy = proxy
self._received = asyncio.Future()
self._body = b''
def connection_made(self, transport):
''' Save a reference to the transport. '''
log.debug('MITM connection opened.')
self._transport = transport
cert = self._transport.get_extra_info('peercert')
log.debug('MITM upstream certificate: {}'.format(cert))
self._loop.call_soon(self._proxy.start_tls, self._http_version)
def connection_lost(self, exc):
log.debug('MITM connection closed.')
self._received.cancel()
def data_received(self, data):
''' Accumulate request data. '''
log.debug('MITM data received: {}'.format(data))
self._parser.execute(data, len(data))
if self._parser.is_partial_body():
self._body += self._parser.recv_body()
if self._parser.is_message_complete():
version = self._parser.get_version()
status = self._parser.get_status_code()
reason = None # For some reason, the parser doesn't expose this :(
headers = self._parser.get_headers()
log.debug('MITM upstream status: {}'.format(status))
log.debug('MITM upstream headers: {}'.format(headers))
log.debug('MITM upstream body: {}...'.format(self._body[:1000]))
self._received.set_result(
(version, status, reason, headers, self._body)
)
self._transport.close()
def forward(self, data):
''' Forward data to upstream host. '''
log.debug('MITM sending data: {}'.format(data))
self._transport.write(data)
@asyncio.coroutine
def receive(self):
''' Read data received by this MITM instance. '''
response = yield from self._received
return response
def parse(self):
data = [{
'label':
'以太网帧头部 / Ethernet Headers',
'value':
'',
'bold':
True,
'children': [{
'label': '目的端 MAC 地址',
'value': self.ethHeader.destMac
}, {
'label': '发送端 MAC 地址',
'value': self.ethHeader.sourceMac
}, {
'label':
'帧类型',
'value':
'%s (0x%s)' % (self.ethHeader.type, self.ethHeader.type_code)
}]
}]
if self.protocol == 'ARP':
data.append({
'label':
'ARP 消息 / Address Resolution Protocol',
'value':
'',
'bold':
True,
'children': [{
'label':
'硬件类型',
'value':
'%s (%s)' % (self.arpBody.hardware_type,
self.arpBody.hardware_type_code)
}, {
'label':
'协议类型',
'value':
'%s (0x%s)' % (self.arpBody.protocol_type,
self.arpBody.protocol_type_code)
}, {
'label': '硬件地址长度',
'value': str(self.arpBody.hardware_size)
}, {
'label': '协议地址长度',
'value': str(self.arpBody.protocol_size)
}, {
'label':
'操作码',
'value':
'%s (%s)' %
(self.arpBody.operation, self.arpBody.operation_code)
}, {
'label': '发送端 MAC 地址',
'value': self.arpBody.sender_mac_address
}, {
'label': '发送端 IP 地址',
'value': self.arpBody.sender_ip_address
}, {
'label': '目的端 MAC 地址',
'value': self.arpBody.target_mac_address
}, {
'label': '目的端 IP 地址',
'value': self.arpBody.target_ip_address
}]
})
else:
if self.ipHeader.version == 4:
self.ipHeader.verifyChecksum = verifyChecksum(
self.ipHeader.header_raw, [], '').verifyChecksum
data.append({
'label':
'IPv4 头部 / IPv4 Header',
'value':
'',
'bold':
True,
'children': [{
'label': '协议版本',
'value': self.ipHeader.version
}, {
'label':
'头部长度',
'value':
str(self.ipHeader.header_length) + ' Bytes'
}, {
'label':
'服务类型',
'value':
'0x%s' % (self.ipHeader.differentiated_services)
}, {
'label': '来源 IP',
'value': self.ipHeader.source_ip
}, {
'label': '目标 IP',
'value': self.ipHeader.dest_ip
}, {
'label': '总长度',
'value': self.ipHeader.total_length
}, {
'label':
'标识',
'value':
'0x%s (%s)' % (self.ipHeader.identification,
self.ipHeader.identification_int)
}, {
'label':
'标志',
'value':
'%s' % (self.ipHeader.flags.raw),
'children': [{
'label':
'保留位',
'value':
'%s | %s... .... .... ....' %
(self.ipHeader.flags.reserved,
int(self.ipHeader.flags.reserved))
}, {
'label':
'Don\'t fragment',
'value':
'%s | .%s.. .... .... ....' %
(self.ipHeader.flags.fragment,
int(self.ipHeader.flags.fragment))
}, {
'label':
'More fragments',
'value':
'%s | ..%s. .... .... ....' %
(self.ipHeader.flags.more_fragment,
int(self.ipHeader.flags.more_fragment))
}, {
'label':
'分段偏移',
'value':
'%s | ...%s' %
(self.ipHeader.flags.fragment_offset,
self.ipHeader.flags.fragment_offset_bin)
}]
}, {
'label': '生存期',
'value': self.ipHeader.time_to_live
}, {
'label':
'协议',
'value':
'%s (%s)' %
(self.ipHeader.protocol, self.ipHeader.protocol_code)
}, {
'label':
'校验和',
'value':
'0x%s (%s)' % (self.ipHeader.origin_checksum, '校验' + {
True: '通过',
False: '失败'
}[self.ipHeader.verifyChecksum])
}]
})
else:
ipv6_header = {
'label':
'IPv6 头部 / IPv6 Header',
'value':
'',
'bold':
True,
'children': [{
'label': '协议版本',
'value': self.ipHeader.version
}, {
'label': '通信分类',
'value': '0x%s' % (self.ipHeader._class)
}, {
'label': '流标签',
'value': '0x%s' % (self.ipHeader.float_label)
}, {
'label': '有效载荷长度',
'value': self.ipHeader.payload_length
}, {
'label':
'下一头部类型',
'value':
'%s (%s)' % (self.ipHeader.next_header,
self.ipHeader.next_header_code)
}, {
'label': '跳数限制',
'value': self.ipHeader.hop_limit
}, {
'label': '源 IP',
'value': self.ipHeader.source_ip
}, {
'label': '目的 IP',
'value': self.ipHeader.dest_ip
}]
}
for option in self.ipHeader.options:
ipv6_header['children'].append({
'label':
consts.protocol_types[str(option['code'])],
'value':
'0x' + option['value'],
'children': [{
'label':
'下一头部类型',
'value':
'%s (%s)' % (consts.protocol_types[str(
option['next_header'])], option['next_header'])
}]
})
data.append(ipv6_header)
if self.ipHeader.version == 4 and self.ipHeader.flags.more_fragment == True:
# print('Waiting for more fragments.')
ids = self.ip_ids[self.ipHeader.identification_int]
slicing = {
'label': 'IP 分片',
'value': '共 %s 个数据包' % len(ids),
'bold': True,
'children': []
}
for id in ids:
slicing['children'].append({
'label':
'#%s' % id,
'value':
'%s Bytes' % (self.ip_packets[id].length / 8)
})
data.append(slicing)
else:
if self.ipHeader.protocol == 'TCP':
self.ipBody.tcpHeader.verifyChecksum = verifyChecksum(
self.ipBody.parameters[0], self.ipBody.parameters[1],
self.ipHeader.protocol).verifyChecksum
self.ipBody.tcpHeader.options = tcpOptions(
BitArray(self.ipBodyRaw)
[160:self.ipBody.tcpHeader.header_length * 8]).options
tcp_header = {
'label':
'TCP 头部 / Transmission Control Protocol Header',
'value':
'',
'bold':
True,
'children': [{
'label': '源端口',
'value': self.ipBody.tcpHeader.source_port
}, {
'label':
'目的端口',
'value':
self.ipBody.tcpHeader.destination_port
}, {
'label':
'数据序号 (seq)',
'value':
self.ipBody.tcpHeader.sequence_number
}, {
'label':
'确认序号 (ack)',
'value':
self.ipBody.tcpHeader.acknowledge_number
}, {
'label':
'首部长度',
'value':
self.ipBody.tcpHeader.header_length
}, {
'label':
'标志位',
'value':
'0x' + self.ipBody.tcpHeader.flags_raw,
'children': [{
'label':
'Reserved',
'value':
'%s | %s. .... ....' %
(self.ipBody.tcpHeader.flags.reserved.uint,
self.ipBody.tcpHeader.flags.reserved.bin)
}, {
'label':
'Nonce',
'value':
'%s | ...%d .... ....' %
(self.ipBody.tcpHeader.flags.nonce,
self.ipBody.tcpHeader.flags.nonce)
}, {
'label':
'Congestion Window Reduced',
'value':
'%s | .... %d... ....' %
(self.ipBody.tcpHeader.flags.cwr,
self.ipBody.tcpHeader.flags.cwr)
}, {
'label':
'ECN-Echo',
'value':
'%s | .... .%d.. ....' %
(self.ipBody.tcpHeader.flags.ecn_echo,
self.ipBody.tcpHeader.flags.ecn_echo)
}, {
'label':
'Urgent',
'value':
'%s | .... ..%d. ....' %
(self.ipBody.tcpHeader.flags.urgent,
self.ipBody.tcpHeader.flags.urgent)
}, {
'label':
'Acknowledgment',
'value':
'%s | .... ...%d ....' %
(self.ipBody.tcpHeader.flags.acknowledgement,
self.ipBody.tcpHeader.flags.acknowledgement)
}, {
'label':
'Push',
'value':
'%s | .... .... %d...' %
(self.ipBody.tcpHeader.flags.push,
self.ipBody.tcpHeader.flags.push)
}, {
'label':
'Reset',
'value':
'%s | .... .... .%d..' %
(self.ipBody.tcpHeader.flags.reset,
self.ipBody.tcpHeader.flags.reset)
}, {
'label':
'Syn',
'value':
'%s | .... .... ..%d.' %
(self.ipBody.tcpHeader.flags.syn,
self.ipBody.tcpHeader.flags.syn)
}, {
'label':
'Fin',
'value':
'%s | .... .... ...%d' %
(self.ipBody.tcpHeader.flags.fin,
self.ipBody.tcpHeader.flags.fin)
}]
}, {
'label': '窗口大小',
'value': self.ipBody.tcpHeader.window_size
}, {
'label':
'校验和',
'value':
'0x%s (%s)' %
(self.ipBody.tcpHeader.checksum, '校验' + {
True: '通过',
False: '失败'
}[self.ipBody.tcpHeader.verifyChecksum])
}]
}
options = []
if self.ipBody.tcpHeader.options:
for idx in range(len(self.ipBody.tcpHeader.options)):
option = {
'label':
self.ipBody.tcpHeader.options[idx][0]['label'],
'value':
'(%s)' %
self.ipBody.tcpHeader.options[idx][0]['value'],
'children':
self.ipBody.tcpHeader.options[idx][1:]
}
options.append(option)
if options:
tcp_header['children'].append({
'label': '选项',
'value': '',
'children': options
})
data.append(tcp_header)
print(self.id)
print(tcp_bodies)
if self.id in packet_id_struct:
tmp = []
http_payload = None
for p_id in packet_id_struct[self.id]:
tmp.append({'value': '', 'label': '#%s' % p_id})
if self.id in tcp_bodies:
# print(tcp_bodies[self.id]['data'].decode('utf-8', 'ignore'))
children = [{
'label':
'该包是 TCP 分段的最后一段, 可以通过右下角按钮「导出 TCP 分段数据」.',
'value': '',
'bold': True
}, {
'label': '共 %s 个分段' % len(tmp),
'value': '',
'bold': True,
'children': tmp
}]
try:
p = HttpParser()
recved = len(tcp_bodies[self.id]['data'])
nparsed = p.execute(
tcp_bodies[self.id]['data'], recved)
assert nparsed == recved
headers = []
for header in p.get_headers():
headers.append({
'label':
header,
'value':
p.get_headers()[header]
})
print(p.get_path(), p.get_url(),
p.get_fragment(), p.get_method(),
p.get_query_string(),
p.get_status_code(),
p.get_wsgi_environ())
http_payload = [{
'label':
'HTTP 版本',
'value':
'%s.%s' %
(p.get_version()[0], p.get_version()[1])
}, {
'label': 'HTTP 头部',
'value': '',
'children': headers
}]
if len(p.get_url()) != 0:
http_payload.append({
'label': '请求方式',
'value': p.get_method()
})
http_payload.append({
'label': '路径',
'value': p.get_url()
})
http_payload.append({
'label':
'请求参数',
'value':
p.get_query_string()
})
http_payload.append({
'label':
'主机名',
'value':
p.get_wsgi_environ()['HTTP_HOST']
})
else:
http_payload.append({
'label':
'状态码',
'value':
p.get_status_code()
})
except AssertionError:
pass
else:
children = [{
'label': '共 %s 个分段' % len(tmp),
'value': '',
'bold': True,
'children': tmp
}]
data.append({
'label': 'TCP 数据 / TCP Payload',
'value': '',
'bold': True,
'children': children
})
if http_payload != None:
data.append({
'label': 'HTTP 数据 / HTTP Data',
'value': '',
'bold': True,
'children': http_payload
})
'''
if self.ipBody.tcpBody.has_body:
try:
p = HttpParser()
recved = len(self.ipBody.tcpBody.buf)
nparsed = p.execute(self.ipBody.tcpBody.buf, recved)
assert nparsed == recved
print(p.get_headers())
except AssertionError:
print('NOT HTTP')
data.append({
'label': 'TCP 数据 / Data',
'value': '',
'bold': True,
'children': [
{
'label': '数据',
'value': self.ipBody.tcpBody.raw
}
]
})
'''
elif self.ipHeader.protocol == 'UDP':
self.ipBody.udpHeader.verifyChecksum = verifyChecksum(
self.ipBody.parameters[0], self.ipBody.parameters[1],
self.ipHeader.protocol).verifyChecksum
data.append({
'label':
'UDP 头部 / User Datagram Protocol Header',
'value':
'',
'bold':
True,
'children': [{
'label': '源端口',
'value': self.ipBody.udpHeader.source_port
}, {
'label':
'目的端口',
'value':
self.ipBody.udpHeader.destination_port
}, {
'label': '长度',
'value': self.ipBody.udpHeader.length
}, {
'label':
'校验和',
'value':
'0x%s (%s)' %
(self.ipBody.udpHeader.checksum, '校验' + {
True: '通过',
False: '失败'
}[self.ipBody.udpHeader.verifyChecksum])
}]
})
if self.ipBody.udpHeader.source_port == 53 or self.ipBody.udpHeader.destination_port == 53: # DNS
children = [{
'label': '会话标识',
'value': self.ipBody.dnsBody.transaction_id
}, {
'label':
'标志',
'value':
'0x' + self.ipBody.dnsBody.transaction_id
}, {
'label': '问题数',
'value': self.ipBody.dnsBody.questions
}, {
'label': '回答资源记录数',
'value': self.ipBody.dnsBody.answer_rrs
}, {
'label': '授权资源记录数',
'value': self.ipBody.dnsBody.authority_rrs
}, {
'label': '附加资源记录数',
'value': self.ipBody.dnsBody.additional_rrs
}]
if len(self.ipBody.dnsBody.queries) > 0:
queries = []
for query in self.ipBody.dnsBody.queries:
queries.append({
'label':
str(query.qname),
'value':
'',
'bold':
True,
'children': [{
'label': '域名',
'value': str(query.qname)
}, {
'label':
'Type',
'value':
'%s (%s)' %
(consts.dns_types[query.qtype],
query.qtype)
}, {
'label':
'Class',
'value':
'%s (%s)' %
(consts.dns_classes[query.qclass],
query.qclass)
}]
})
children.append({
'label': '查询问题',
'value': '',
'bold': True,
'children': queries
})
if len(self.ipBody.dnsBody.answers) > 0:
answers = []
for answer in self.ipBody.dnsBody.answers:
answers.append({
'label':
str(answer.rname),
'value':
'',
'bold':
True,
'children': [{
'label': '域名',
'value': str(answer.rname)
}, {
'label':
'Type',
'value':
'%s (%s)' %
(consts.dns_types[answer.rtype],
answer.rtype)
}, {
'label':
'Class',
'value':
'%s (%s)' %
(consts.dns_classes[answer.rclass],
answer.rclass)
}, {
'label': '生存时间 (ttl)',
'value': str(answer.ttl)
}, {
'label': '数据',
'value': str(answer.rdata)
}]
})
children.append({
'label': '回答',
'value': '',
'bold': True,
'children': answers
})
data.append({
'label': 'DNS / Domain Name System',
'value': '',
'bold': True,
'children': children
})
elif 'ICMP' in self.ipHeader.protocol:
if 'IPv6' in self.ipHeader.protocol:
self.ipBody.icmpHeader.verifyChecksum = verifyChecksum(
self.ipBody.parameters[0],
self.ipBody.parameters[1],
self.ipHeader.protocol).verifyChecksum
else:
self.ipBody.icmpHeader.verifyChecksum = verifyChecksum(
self.ipBody.parameters, [], '').verifyChecksum
data.append({
'label':
'ICMP 头部 / Internet Control Message Protocol Headers',
'value':
'',
'bold':
True,
'children': [{
'label':
'类型',
'value':
'%s (%s)' % (self.ipBody.icmpHeader.type,
self.ipBody.icmpHeader.type_name)
}, {
'label': '代码',
'value': self.ipBody.icmpHeader.code
}, {
'label':
'校验和',
'value':
'0x%s (%s)' %
(self.ipBody.icmpHeader.checksum, '校验' + {
True: '通过',
False: '失败'
}[self.ipBody.icmpHeader.verifyChecksum])
}]
})
elif 'IGMP' in self.ipHeader.protocol:
if self.ipHeader.payload_length == 8:
self.ipBody.igmpHeader.verifyChecksum = verifyChecksum(
self.ipBody.parameters, [], '').verifyChecksum
data.append({
'label':
'IGMP 头部 / Internet Group Management Protocol Headers',
'value':
'',
'bold':
True,
'children': [{
'label':
'类型',
'value':
'0x%s(%s)' % (self.ipBody.igmpHeader.type,
self.ipBody.igmpHeader.type_name)
}, {
'label':
'最大响应时延',
'value':
'%s 秒(0x%s)' %
(self.ipBody.igmpHeader.maxRespTime,
self.ipBody.igmpHeader.maxRespTimeHex)
}, {
'label':
'校验和',
'value':
'0x%s(%s)' %
(self.ipBody.igmpHeader.checksum, '校验' + {
True: '通过',
False: '失败'
}[self.ipBody.igmpHeader.verifyChecksum])
}, {
'label':
'组地址',
'value':
self.ipBody.igmpHeader.groupAddress
}]
})
else:
self.ipBody.igmpv3Header.verifyChecksum = verifyChecksum(
self.ipBody.parameters, [], '').verifyChecksum
data.append({
'label':
'IGMPv3 头部 / Internet Group Management Protocol Version 3 Headers',
'value':
'',
'bold':
True,
'children': [{
'label':
'类型',
'value':
'0x%s' % self.ipBody.igmpv3Header.type
}, {
'label':
'校验和',
'value':
'0x%s(%s)' %
(self.ipBody.igmpv3Header.checksum, '校验' + {
True: '通过',
False: '失败'
}[self.ipBody.igmpv3Header.verifyChecksum])
}]
})
return data
