def test_paramiko_doc(self) -> None:
response = b'HTTP/1.1 304 Not Modified\r\nDate: Tue, 03 Dec 2019 02:31:55 GMT\r\nConnection: keep-alive' \
b'\r\nLast-Modified: Sun, 23 Jun 2019 22:58:21 GMT\r\nETag: "5d10040d-1af2c"' \
b'\r\nX-Cname-TryFiles: True\r\nX-Served: Nginx\r\nX-Deity: web02\r\nCF-Cache-Status: DYNAMIC' \
b'\r\nServer: cloudflare\r\nCF-RAY: 53f2208c6fef6c38-SJC\r\n\r\n'
self.parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
self.parser.parse(response)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def before_upstream_connection(
self, request: HttpParser) -> Optional[HttpParser]:
if request.has_header(b'Host'):
print(request.headers)
if request.headers[b'host'] == (b'Host', b'github.com'):
repo_name, _ = request.path.split(b'.git', 1)
repo_url = b'[email protected]:'+repo_name[1:]+b'.git'
print('repo name: ', repo_name)
if not os.path.exists((b'repos'+repo_name).decode()):
print('repo path: ', (b'repos'+repo_name).decode())
os.makedirs((b'repos'+repo_name).decode(), exist_ok=True)
print(*['git', 'clone', repo_url.decode(), '.'+repo_name.decode()], sep=' ')
subprocess.check_call(['git', '--bare', 'clone', repo_url.decode(), '.'+repo_name.decode()], cwd='repos')
subprocess.check_call(['mv', '.'+repo_name.decode()+'/.git', '.'+repo_name.decode()+'.git'], cwd='repos')
print('path: ', (b'repos'+repo_name+b'/.git').decode())
subprocess.check_call(['git', '--bare', 'update-server-info'], cwd=(b'repos'+repo_name+b'.git').decode())
# Redirect all non-https requests to inbuilt WebServer.
if request.method != httpMethods.CONNECT:
request.set_url(self.UPSTREAM_SERVER+request.url.path)
print(request.url)
# Update Host header too, otherwise upstream can reject our request
if request.has_header(b'Host'):
request.del_header(b'Host')
request.add_header(
b'Host', urlparse.urlsplit(
self.UPSTREAM_SERVER).netloc)
return request
async def test_man_in_the_middle_plugin(self) -> None:
await self.protocol_handler._run_once()
self.assertEqual(self.mock_sign_csr.call_count, 1)
self.assertEqual(self.mock_gen_csr.call_count, 1)
self.assertEqual(self.mock_gen_public_key.call_count, 1)
self.mock_server_conn.assert_called_once_with('uni.corn', 443)
self.server.connect.assert_called()
self.assertEqual(
self.protocol_handler.work.connection,
self.client_ssl_connection,
)
self.assertEqual(self.server.connection, self.server_ssl_connection)
self._conn.send.assert_called_with(
PROXY_TUNNEL_ESTABLISHED_RESPONSE_PKT, )
self.assertFalse(self.protocol_handler.work.has_buffer())
#
request = build_http_request(
b'GET',
b'/',
headers={
b'Host': b'uni.corn',
},
no_ua=True,
)
self.client_ssl_connection.recv.return_value = request
# Client read
await self.protocol_handler._run_once()
self.server.queue.assert_called_once_with(request)
# Server write
await self.protocol_handler._run_once()
self.server.flush.assert_called_once()
# Server read
self.server.recv.return_value = okResponse(
content=b'Original Response From Upstream', )
await self.protocol_handler._run_once()
response = HttpParser(httpParserTypes.RESPONSE_PARSER)
response.parse(self.protocol_handler.work.buffer[0])
assert response.body
self.assertEqual(
gzip.decompress(response.body),
b'Hello from man in the middle',
)
def before_upstream_connection(
self, request: HttpParser) -> Optional[HttpParser]:
"""Avoid upstream connection of server in the request.
Initialize, connection to upstream proxy.
"""
if 'Zeus-proxy' in request.headers:
[host, port, username,
password] = request.header['Zeus-proxy'].split(':')
del request.header['Zeus-proxy']
request.add_header('Proxy-Authorization',
_basic_auth_str(self.username, self.password))
self.conn = new_socket_connection((host, int(port)))
else:
raise Exception('Could not find Zeus-proxy header')
# Implement your own logic here e.g. round-robin, least connection etc.
return None
class EmojiInjectionPlugin(HttpProxyBasePlugin):
def __init__(self, *args: Any, **kwargs: Any) -> None:
super().__init__(*args, **kwargs)
self.response = HttpParser(httpParserTypes.RESPONSE_PARSER)
def before_upstream_connection(self, request: HttpParser) -> Optional[HttpParser]:
return request
def handle_client_request(self, request: HttpParser) -> Optional[HttpParser]:
return request
def handle_upstream_chunk(self, chunk: memoryview) -> memoryview:
self.response.parse(chunk.tobytes())
if self.response.state == httpParserStates.COMPLETE:
self.handle_response_data()
new_content_length = len(self.response.body)
bytes_header_name = b'content-length'
self.response.add_header(bytes_header_name, bytes_(new_content_length))
self.client.queue(memoryview(self.response.build_response()))
return memoryview(b'')
def on_upstream_connection_close(self) -> None:
pass
def handle_response_data(self) -> None:
if 'text/html' in self.response.header(b'content-type').decode():
response_handler = HtmlResponseHandler(self.response)
elif 'application/json' == self.response.header(b'content-type').decode():
response_handler = JsonResponseHandler(self.response)
else:
return
response_handler.insert_emoji_to_response_data()
def handle_request(self, request: HttpParser) -> None:
nspid = self.nspid
with netns.NetNS(nspid=nspid):
with socket_connection(self.upstream) as conn:
conn.send(request.build())
data = conn.recv(DEFAULT_BUFFER_SIZE)
while (data):
self.client.queue(memoryview(data))
data = conn.recv(DEFAULT_BUFFER_SIZE)
def assert_tunnel_response(self, mock_server_connection: mock.Mock,
server: mock.Mock) -> None:
self.protocol_handler.run_once()
self.assertTrue(
cast(HttpProxyPlugin, self.protocol_handler.
plugins['HttpProxyPlugin']).server is not None)
self.assertEqual(self.protocol_handler.client.buffer[0],
HttpProxyPlugin.PROXY_TUNNEL_ESTABLISHED_RESPONSE_PKT)
mock_server_connection.assert_called_once()
server.connect.assert_called_once()
server.queue.assert_not_called()
server.closed = False
parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
parser.parse(self.protocol_handler.client.buffer[0].tobytes())
self.assertEqual(parser.state, httpParserStates.COMPLETE)
assert parser.code is not None
self.assertEqual(int(parser.code), 200)
async def test_static_web_server_serves(self) -> None:
self._conn.recv.return_value = build_http_request(
b'GET', b'/index.html',
)
self.mock_selector.return_value.select.side_effect = [
[(
selectors.SelectorKey(
fileobj=self._conn.fileno(),
fd=self._conn.fileno(),
events=selectors.EVENT_READ,
data=None,
),
selectors.EVENT_READ,
)],
[(
selectors.SelectorKey(
fileobj=self._conn.fileno(),
fd=self._conn.fileno(),
events=selectors.EVENT_WRITE,
data=None,
),
selectors.EVENT_WRITE,
)],
]
await self.protocol_handler._run_once()
await self.protocol_handler._run_once()
self.assertEqual(self.mock_selector.return_value.select.call_count, 2)
self.assertEqual(self._conn.send.call_count, 1)
encoded_html_file_content = gzip.compress(self.html_file_content)
# parse response and verify
response = HttpParser(httpParserTypes.RESPONSE_PARSER)
response.parse(self._conn.send.call_args[0][0])
self.assertEqual(response.code, b'200')
self.assertEqual(response.header(b'content-type'), b'text/html')
self.assertEqual(response.header(b'cache-control'), b'max-age=86400')
self.assertEqual(response.header(b'content-encoding'), b'gzip')
self.assertEqual(response.header(b'connection'), b'close')
self.assertEqual(
response.header(b'content-length'),
bytes_(len(encoded_html_file_content)),
)
assert response.body
self.assertEqual(
gzip.decompress(response.body),
self.html_file_content,
)
def test_request_factory(self) -> None:
r = HttpParser.request(
b'POST http://localhost:12345 HTTP/1.1' + CRLF + b'key: value' +
CRLF + b'Content-Length: 13' + CRLF + CRLF + b'Hello from py', )
self.assertEqual(r.host, b'localhost')
self.assertEqual(r.port, 12345)
self.assertEqual(r.path, None)
self.assertEqual(r.header(b'key'), b'value')
self.assertEqual(r.header(b'KEY'), b'value')
self.assertEqual(r.header(b'content-length'), b'13')
self.assertEqual(r.body, b'Hello from py')
async def assert_tunnel_response(
self,
server: mock.Mock,
) -> None:
await self.protocol_handler._run_once()
self.assertTrue(
cast(
HttpProxyPlugin,
self.protocol_handler.plugin,
).upstream is not None, )
self.assertEqual(
self.protocol_handler.work.buffer[0],
PROXY_TUNNEL_ESTABLISHED_RESPONSE_PKT,
)
self.mock_server_connection.assert_called_once()
server.connect.assert_called_once()
server.queue.assert_not_called()
server.closed = False
parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
parser.parse(self.protocol_handler.work.buffer[0])
self.assertEqual(parser.state, httpParserStates.COMPLETE)
assert parser.code is not None
self.assertEqual(int(parser.code), 200)
def test_proxy_protocol(self) -> None:
r = HttpParser.request(
b'PROXY TCP4 192.168.0.1 192.168.0.11 56324 443' + CRLF +
b'GET / HTTP/1.1' + CRLF + b'Host: 192.168.0.11' + CRLF + CRLF,
enable_proxy_protocol=True,
)
self.assertTrue(r.protocol is not None)
assert r.protocol and r.protocol.version and \
r.protocol.family and \
r.protocol.source and \
r.protocol.destination
self.assertEqual(r.protocol.version, 1)
self.assertEqual(r.protocol.family, b'TCP4')
self.assertEqual(r.protocol.source, (b'192.168.0.1', 56324))
self.assertEqual(r.protocol.destination, (b'192.168.0.11', 443))
async def test_modify_post_data_plugin(self) -> None:
await self.protocol_handler._run_once()
self.assertEqual(self.mock_sign_csr.call_count, 1)
self.assertEqual(self.mock_gen_csr.call_count, 1)
self.assertEqual(self.mock_gen_public_key.call_count, 1)
self.mock_server_conn.assert_called_once_with('uni.corn', 443)
self.server.connect.assert_called()
self.assertEqual(
self.protocol_handler.work.connection,
self.client_ssl_connection,
)
self.assertEqual(self.server.connection, self.server_ssl_connection)
self._conn.send.assert_called_with(
PROXY_TUNNEL_ESTABLISHED_RESPONSE_PKT, )
self.assertFalse(self.protocol_handler.work.has_buffer())
#
original = b'{"key": "value"}'
modified = b'{"key": "modified"}'
self.client_ssl_connection.recv.return_value = build_http_request(
b'POST',
b'/',
headers={
b'Host': b'uni.corn',
b'Content-Length': bytes_(len(original)),
b'Content-Type': b'application/x-www-form-urlencoded',
},
body=original,
no_ua=True,
)
await self.protocol_handler._run_once()
self.server.queue.assert_called_once()
# pkt = build_http_request(
# b'POST', b'/',
# headers={
# b'Host': b'uni.corn',
# b'Content-Length': bytes_(len(modified)),
# b'Content-Type': b'application/json',
# },
# body=modified,
# )
response = HttpParser.response(
self.server.queue.call_args_list[0][0][0].tobytes(), )
self.assertEqual(response.body, modified)
def handle_client_request(self,
request: HttpParser) -> Optional[HttpParser]:
if request.method == httpMethods.GET:
path_str = text_(request.path)
downloaded = 0
match = re.search('(?:^|\\?|&)downloaded=(\\d+)(?:&|$)', path_str)
if match:
downloaded = int(match.group(1))
match = re.search('(?:^|\\?|&)uploaded=(\\d+)(?:&|$)', path_str)
if match:
uploaded = int(downloaded *
1.2 if downloaded > 0 else int(match.group(1)) *
1.2)
path_str = path_str[:match.start(1)] + str(
uploaded) + path_str[match.end(1):]
request.path = bytes_(path_str)
return request
def assert_pipeline_response(self, response: bytes) -> None:
self.parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
self.parser.parse(response + response)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
self.assertEqual(self.parser.body, b'{"key":"value"}')
self.assertEqual(self.parser.buffer, response)
# parse buffer
parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
parser.parse(self.parser.buffer)
self.assertEqual(parser.state, httpParserStates.COMPLETE)
self.assertEqual(parser.body, b'{"key":"value"}')
self.assertEqual(parser.buffer, b'')
def test_issue_398(self) -> None:
p = HttpParser(httpParserTypes.RESPONSE_PARSER)
p.parse(HTTP_1_0 + b' 200 OK' + CRLF)
self.assertEqual(p.version, HTTP_1_0)
self.assertEqual(p.code, b'200')
self.assertEqual(p.reason, b'OK')
self.assertEqual(p.state, httpParserStates.LINE_RCVD)
p.parse(
b'CP=CAO PSA OUR' + CRLF + b'Cache-Control:private,max-age=0;' +
CRLF + b'X-Frame-Options:SAMEORIGIN' + CRLF +
b'X-Content-Type-Options:nosniff' + CRLF +
b'X-XSS-Protection:1; mode=block' + CRLF +
b'Content-Security-Policy:default-src \'self\' \'unsafe-inline\' \'unsafe-eval\''
+ CRLF +
b'Strict-Transport-Security:max-age=2592000; includeSubdomains' +
CRLF + b'Set-Cookie: lang=eng; path=/;HttpOnly;' + CRLF +
b'Content-type:text/html;charset=UTF-8;' + CRLF + CRLF +
b'<!-- HTML RESPONSE HERE -->', )
self.assertEqual(p.body, b'<!-- HTML RESPONSE HERE -->')
self.assertEqual(p.state, httpParserStates.RCVING_BODY)
async def test_proposed_rest_api_plugin(self) -> None:
path = b'/v1/users/'
self._conn.recv.return_value = build_http_request(
b'GET',
b'http://%s%s' % (
ProposedRestApiPlugin.API_SERVER,
path,
),
headers={
b'Host': ProposedRestApiPlugin.API_SERVER,
},
)
self.mock_selector.return_value.select.side_effect = [
[(
selectors.SelectorKey(
fileobj=self._conn.fileno(),
fd=self._conn.fileno(),
events=selectors.EVENT_READ,
data=None,
),
selectors.EVENT_READ,
)],
]
await self.protocol_handler._run_once()
self.mock_server_conn.assert_not_called()
response = HttpParser(httpParserTypes.RESPONSE_PARSER)
response.parse(self.protocol_handler.work.buffer[0])
assert response.body
self.assertEqual(
response.header(b'content-type'),
b'application/json',
)
self.assertEqual(
response.header(b'content-encoding'),
b'gzip',
)
self.assertEqual(
gzip.decompress(response.body),
bytes_(json.dumps(ProposedRestApiPlugin.REST_API_SPEC[path], ), ),
)
class PluginBase(HttpProxyBasePlugin):
'''Modified HttpProxyBasePlugin from proxy.py. Provides more functionality'''
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.request = HttpParser(httpParserTypes.REQUEST_PARSER)
self.response = HttpParser(httpParserTypes.RESPONSE_PARSER)
def before_upstream_connection(self, connection):
return self.intercept_connection(connection)
def handle_client_request(self, request):
rtval = self.intercept_request(request)
self.request = rtval
return rtval
def handle_upstream_chunk(self, chunk):
self.response.parse(chunk.tobytes())
if self.response.state == httpParserStates.COMPLETE:
self.intercept_response(self.response)
return memoryview(self.intercept_chunk(chunk.tobytes()))
def on_upstream_connection_close(self):
self.close_connection()
def intercept_connection(self, conn):
'''
Intercept Connection. The `conn` argument represents
HttpParser class from proxy.http.parser.
Intercept the initial connection state and represent
the data before the connection is actuall established.
For TLS Interception, it would most probably be a
CONNECT request. You can check host, port and other initials
in this section.
Modified/Orignal `conn` must be returned in this function.
Returning `None` will drop the connection.
'''
return conn
def intercept_request(self, request):
'''
Intercept Request. The `request` argument represents
HttpParser class from proxy.http.parser
Intercept the orignal request sent to the server. The
request can be modified at this stage.
You will see GET/POST/HEAD and other method requests
here. You can modify the request body and other
necessary details here.
Modified/Orignal `request` must be returned in this function.
Returning `None` will not send the request to the server.
'''
return request
def intercept_chunk(self, chunk):
'''
Intercept chunks received from the server. The `chunk`
argument represents built-in `bytes` object from
python.
Intercept response in chunks received from server.
It is not in the final form yet. So, you can the received
response here
You will see mostly plain bytes in this section with what's
incoming. This chunk from this function is then appended
to response in `interept_response`.
Modified/Orignal `chunk` must be returned in this function.
Empty bytes object will append nothing to response
Return `None` will cancel the response at that very moment.
'''
return chunk
def intercept_response(self, response):
'''
Intercept response received from the server. The `response`
argument represents HttpParser from proxy.http.parser.
Intercept the response combined from the chunks received in
`intercept_chunk` method. The response can only be viewed
at this stage. It is not to be modified here
NOTE: Changing or returning response here won't have any effect
on the orignal response. For modifying response see
`intercept_response` method.
This function doesn't return anything.
'''
return response
def close_connection(self):
'''
This methods gets called when the connection has been closed.
It doesn't have arguments.
The request and response both can be analyzed at this part
and can be accessed at `self.request` and `self.response`
respectively.
The modifications in here won't mean anything. But you can
deduce other results in this section.
This function doesn't return anything and basically used for
analysis.
'''
pass
class TestHttpParser(unittest.TestCase):
def setUp(self) -> None:
self.parser = HttpParser(httpParserTypes.REQUEST_PARSER)
def test_issue_127(self) -> None:
with self.assertRaises(HttpProtocolException):
self.parser.parse(CRLF)
with self.assertRaises(HttpProtocolException):
raw = b'qwqrqw!@!#@!#ad adfad\r\n'
while True:
self.parser.parse(raw)
def test_issue_398(self) -> None:
p = HttpParser(httpParserTypes.RESPONSE_PARSER)
p.parse(HTTP_1_0 + b' 200 OK' + CRLF)
self.assertEqual(p.version, HTTP_1_0)
self.assertEqual(p.code, b'200')
self.assertEqual(p.reason, b'OK')
self.assertEqual(p.state, httpParserStates.LINE_RCVD)
p.parse(
b'CP=CAO PSA OUR' + CRLF + b'Cache-Control:private,max-age=0;' +
CRLF + b'X-Frame-Options:SAMEORIGIN' + CRLF +
b'X-Content-Type-Options:nosniff' + CRLF +
b'X-XSS-Protection:1; mode=block' + CRLF +
b'Content-Security-Policy:default-src \'self\' \'unsafe-inline\' \'unsafe-eval\''
+ CRLF +
b'Strict-Transport-Security:max-age=2592000; includeSubdomains' +
CRLF + b'Set-Cookie: lang=eng; path=/;HttpOnly;' + CRLF +
b'Content-type:text/html;charset=UTF-8;' + CRLF + CRLF +
b'<!-- HTML RESPONSE HERE -->', )
self.assertEqual(p.body, b'<!-- HTML RESPONSE HERE -->')
self.assertEqual(p.state, httpParserStates.RCVING_BODY)
def test_urlparse(self) -> None:
self.parser.parse(b'CONNECT httpbin.org:443 HTTP/1.1\r\n')
self.assertTrue(self.parser.is_https_tunnel)
self.assertFalse(self.parser.is_connection_upgrade)
self.assertTrue(self.parser.is_http_1_1_keep_alive)
self.assertFalse(self.parser.content_expected)
self.assertFalse(self.parser.body_expected)
self.assertEqual(self.parser.host, b'httpbin.org')
self.assertEqual(self.parser.port, 443)
self.assertNotEqual(self.parser.state, httpParserStates.COMPLETE)
def test_urlparse_on_invalid_connect_request(self) -> None:
self.parser.parse(b'CONNECT / HTTP/1.0\r\n\r\n')
self.assertTrue(self.parser.is_https_tunnel)
self.assertEqual(self.parser.host, None)
self.assertEqual(self.parser.port, 443)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_unicode_character_domain_connect(self) -> None:
self.parser.parse(bytes_('CONNECT ççç.org:443 HTTP/1.1\r\n'))
self.assertTrue(self.parser.is_https_tunnel)
self.assertEqual(self.parser.host, bytes_('ççç.org'))
self.assertEqual(self.parser.port, 443)
def test_invalid_ipv6_in_request_line(self) -> None:
self.parser.parse(
bytes_(
'CONNECT 2001:db8:3333:4444:CCCC:DDDD:EEEE:FFFF:443 HTTP/1.1\r\n'
), )
self.assertTrue(self.parser.is_https_tunnel)
self.assertEqual(
self.parser.host,
bytes_('[2001:db8:3333:4444:CCCC:DDDD:EEEE:FFFF]', ),
)
self.assertEqual(self.parser.port, 443)
def test_valid_ipv6_in_request_line(self) -> None:
self.parser.parse(
bytes_(
'CONNECT [2001:db8:3333:4444:CCCC:DDDD:EEEE:FFFF]:443 HTTP/1.1\r\n',
), )
self.assertTrue(self.parser.is_https_tunnel)
self.assertEqual(
self.parser.host,
bytes_('[2001:db8:3333:4444:CCCC:DDDD:EEEE:FFFF]', ),
)
self.assertEqual(self.parser.port, 443)
def test_build_request(self) -> None:
self.assertEqual(
build_http_request(
b'GET',
b'http://*****:*****@nm>;tag=root\r\nTo: <sip:nm2@nm2>\r\n' +
b'Call-ID: 50000\r\n' + b'CSeq: 42 OPTIONS\r\n' +
b'Max-Forwards: 70\r\n' + b'Content-Length: 0\r\n' +
b'Contact: <sip:nm@nm>\r\n' + b'Accept: application/sdp\r\n' +
b'\r\n', )
def setUp(self) -> None:
self.request = HttpParser(httpParserTypes.REQUEST_PARSER)
class TestHttpParser(unittest.TestCase):
def setUp(self) -> None:
self.parser = HttpParser(httpParserTypes.REQUEST_PARSER)
def test_urlparse(self) -> None:
self.parser.parse(b'CONNECT httpbin.org:443 HTTP/1.1\r\n')
self.assertEqual(self.parser.host, b'httpbin.org')
self.assertEqual(self.parser.port, 443)
def test_build_request(self) -> None:
self.assertEqual(
build_http_request(b'GET', b'http://localhost:12345', b'HTTP/1.1'),
CRLF.join([b'GET http://localhost:12345 HTTP/1.1', CRLF]))
self.assertEqual(
build_http_request(b'GET',
b'http://localhost:12345',
b'HTTP/1.1',
headers={b'key': b'value'}),
CRLF.join(
[b'GET http://localhost:12345 HTTP/1.1', b'key: value', CRLF]))
self.assertEqual(
build_http_request(b'GET',
b'http://localhost:12345',
b'HTTP/1.1',
headers={b'key': b'value'},
body=b'Hello from py'),
CRLF.join([
b'GET http://localhost:12345 HTTP/1.1', b'key: value', CRLF
]) + b'Hello from py')
def test_build_response(self) -> None:
self.assertEqual(
build_http_response(200,
reason=b'OK',
protocol_version=b'HTTP/1.1'),
CRLF.join([b'HTTP/1.1 200 OK', CRLF]))
self.assertEqual(
build_http_response(200,
reason=b'OK',
protocol_version=b'HTTP/1.1',
headers={b'key': b'value'}),
CRLF.join([b'HTTP/1.1 200 OK', b'key: value', CRLF]))
def test_build_response_adds_content_length_header(self) -> None:
body = b'Hello world!!!'
self.assertEqual(
build_http_response(200,
reason=b'OK',
protocol_version=b'HTTP/1.1',
headers={b'key': b'value'},
body=body),
CRLF.join([
b'HTTP/1.1 200 OK', b'key: value',
b'Content-Length: ' + bytes_(len(body)), CRLF
]) + body)
def test_build_header(self) -> None:
self.assertEqual(build_http_header(b'key', b'value'), b'key: value')
def test_header_raises(self) -> None:
with self.assertRaises(KeyError):
self.parser.header(b'not-found')
def test_has_header(self) -> None:
self.parser.add_header(b'key', b'value')
self.assertFalse(self.parser.has_header(b'not-found'))
self.assertTrue(self.parser.has_header(b'key'))
def test_set_host_port_raises(self) -> None:
with self.assertRaises(KeyError):
self.parser.set_line_attributes()
def test_find_line(self) -> None:
self.assertEqual(
find_http_line(b'CONNECT python.org:443 HTTP/1.0\r\n\r\n'),
(b'CONNECT python.org:443 HTTP/1.0', CRLF))
def test_find_line_returns_None(self) -> None:
self.assertEqual(find_http_line(b'CONNECT python.org:443 HTTP/1.0'),
(None, b'CONNECT python.org:443 HTTP/1.0'))
def test_connect_request_with_crlf_as_separate_chunk(self) -> None:
"""See https://github.com/abhinavsingh/py/issues/70 for background."""
raw = b'CONNECT pypi.org:443 HTTP/1.0\r\n'
self.parser.parse(raw)
self.assertEqual(self.parser.state, httpParserStates.LINE_RCVD)
self.parser.parse(CRLF)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_get_full_parse(self) -> None:
raw = CRLF.join([b'GET %s HTTP/1.1', b'Host: %s', CRLF])
pkt = raw % (b'https://example.com/path/dir/?a=b&c=d#p=q',
b'example.com')
self.parser.parse(pkt)
self.assertEqual(self.parser.total_size, len(pkt))
self.assertEqual(self.parser.build_path(), b'/path/dir/?a=b&c=d#p=q')
self.assertEqual(self.parser.method, b'GET')
assert self.parser.url
self.assertEqual(self.parser.url.hostname, b'example.com')
self.assertEqual(self.parser.url.port, None)
self.assertEqual(self.parser.version, b'HTTP/1.1')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
self.assertEqual(self.parser.headers[b'host'],
(b'Host', b'example.com'))
self.parser.del_headers([b'host'])
self.parser.add_headers([(b'Host', b'example.com')])
self.assertEqual(raw % (b'/path/dir/?a=b&c=d#p=q', b'example.com'),
self.parser.build())
def test_build_url_none(self) -> None:
self.assertEqual(self.parser.build_path(), b'/None')
def test_line_rcvd_to_rcving_headers_state_change(self) -> None:
pkt = b'GET http://localhost HTTP/1.1'
self.parser.parse(pkt)
self.assertEqual(self.parser.total_size, len(pkt))
self.assert_state_change_with_crlf(httpParserStates.INITIALIZED,
httpParserStates.LINE_RCVD,
httpParserStates.COMPLETE)
def test_get_partial_parse1(self) -> None:
pkt = CRLF.join([b'GET http://localhost:8080 HTTP/1.1'])
self.parser.parse(pkt)
self.assertEqual(self.parser.total_size, len(pkt))
self.assertEqual(self.parser.method, None)
self.assertEqual(self.parser.url, None)
self.assertEqual(self.parser.version, None)
self.assertEqual(self.parser.state, httpParserStates.INITIALIZED)
self.parser.parse(CRLF)
self.assertEqual(self.parser.total_size, len(pkt) + len(CRLF))
self.assertEqual(self.parser.method, b'GET')
assert self.parser.url
self.assertEqual(self.parser.url.hostname, b'localhost')
self.assertEqual(self.parser.url.port, 8080)
self.assertEqual(self.parser.version, b'HTTP/1.1')
self.assertEqual(self.parser.state, httpParserStates.LINE_RCVD)
host_hdr = b'Host: localhost:8080'
self.parser.parse(host_hdr)
self.assertEqual(self.parser.total_size,
len(pkt) + len(CRLF) + len(host_hdr))
self.assertDictEqual(self.parser.headers, dict())
self.assertEqual(self.parser.buffer, b'Host: localhost:8080')
self.assertEqual(self.parser.state, httpParserStates.LINE_RCVD)
self.parser.parse(CRLF * 2)
self.assertEqual(self.parser.total_size,
len(pkt) + (3 * len(CRLF)) + len(host_hdr))
self.assertEqual(self.parser.headers[b'host'],
(b'Host', b'localhost:8080'))
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_get_partial_parse2(self) -> None:
self.parser.parse(
CRLF.join([b'GET http://localhost:8080 HTTP/1.1', b'Host: ']))
self.assertEqual(self.parser.method, b'GET')
assert self.parser.url
self.assertEqual(self.parser.url.hostname, b'localhost')
self.assertEqual(self.parser.url.port, 8080)
self.assertEqual(self.parser.version, b'HTTP/1.1')
self.assertEqual(self.parser.buffer, b'Host: ')
self.assertEqual(self.parser.state, httpParserStates.LINE_RCVD)
self.parser.parse(b'localhost:8080' + CRLF)
self.assertEqual(self.parser.headers[b'host'],
(b'Host', b'localhost:8080'))
self.assertEqual(self.parser.buffer, b'')
self.assertEqual(self.parser.state, httpParserStates.RCVING_HEADERS)
self.parser.parse(b'Content-Type: text/plain' + CRLF)
self.assertEqual(self.parser.buffer, b'')
self.assertEqual(self.parser.headers[b'content-type'],
(b'Content-Type', b'text/plain'))
self.assertEqual(self.parser.state, httpParserStates.RCVING_HEADERS)
self.parser.parse(CRLF)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_post_full_parse(self) -> None:
raw = CRLF.join([
b'POST %s HTTP/1.1', b'Host: localhost', b'Content-Length: 7',
b'Content-Type: application/x-www-form-urlencoded' + CRLF,
b'a=b&c=d'
])
self.parser.parse(raw % b'http://localhost')
self.assertEqual(self.parser.method, b'POST')
assert self.parser.url
self.assertEqual(self.parser.url.hostname, b'localhost')
self.assertEqual(self.parser.url.port, None)
self.assertEqual(self.parser.version, b'HTTP/1.1')
self.assertEqual(
self.parser.headers[b'content-type'],
(b'Content-Type', b'application/x-www-form-urlencoded'))
self.assertEqual(self.parser.headers[b'content-length'],
(b'Content-Length', b'7'))
self.assertEqual(self.parser.body, b'a=b&c=d')
self.assertEqual(self.parser.buffer, b'')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
self.assertEqual(len(self.parser.build()), len(raw % b'/'))
def assert_state_change_with_crlf(self, initial_state: int,
next_state: int,
final_state: int) -> None:
self.assertEqual(self.parser.state, initial_state)
self.parser.parse(CRLF)
self.assertEqual(self.parser.state, next_state)
self.parser.parse(CRLF)
self.assertEqual(self.parser.state, final_state)
def test_post_partial_parse(self) -> None:
self.parser.parse(
CRLF.join([
b'POST http://localhost HTTP/1.1', b'Host: localhost',
b'Content-Length: 7',
b'Content-Type: application/x-www-form-urlencoded'
]))
self.assertEqual(self.parser.method, b'POST')
assert self.parser.url
self.assertEqual(self.parser.url.hostname, b'localhost')
self.assertEqual(self.parser.url.port, None)
self.assertEqual(self.parser.version, b'HTTP/1.1')
self.assert_state_change_with_crlf(httpParserStates.RCVING_HEADERS,
httpParserStates.RCVING_HEADERS,
httpParserStates.HEADERS_COMPLETE)
self.parser.parse(b'a=b')
self.assertEqual(self.parser.state, httpParserStates.RCVING_BODY)
self.assertEqual(self.parser.body, b'a=b')
self.assertEqual(self.parser.buffer, b'')
self.parser.parse(b'&c=d')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
self.assertEqual(self.parser.body, b'a=b&c=d')
self.assertEqual(self.parser.buffer, b'')
def test_connect_request_without_host_header_request_parse(self) -> None:
"""Case where clients can send CONNECT request without a Host header field.
Example:
1. pip3 --proxy http://localhost:8899 install <package name>
Uses HTTP/1.0, Host header missing with CONNECT requests
2. Android Emulator
Uses HTTP/1.1, Host header missing with CONNECT requests
See https://github.com/abhinavsingh/py/issues/5 for details.
"""
self.parser.parse(b'CONNECT pypi.org:443 HTTP/1.0\r\n\r\n')
self.assertEqual(self.parser.method, b'CONNECT')
self.assertEqual(self.parser.version, b'HTTP/1.0')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_request_parse_without_content_length(self) -> None:
"""Case when incoming request doesn't contain a content-length header.
From http://w3-org.9356.n7.nabble.com/POST-with-empty-body-td103965.html
'A POST with no content-length and no body is equivalent to a POST with Content-Length: 0
and nothing following, as could perfectly happen when you upload an empty file for instance.'
See https://github.com/abhinavsingh/py/issues/20 for details.
"""
self.parser.parse(
CRLF.join([
b'POST http://localhost HTTP/1.1', b'Host: localhost',
b'Content-Type: application/x-www-form-urlencoded', CRLF
]))
self.assertEqual(self.parser.method, b'POST')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_response_parse_without_content_length(self) -> None:
"""Case when server response doesn't contain a content-length header for non-chunk response types.
HttpParser by itself has no way to know if more data should be expected.
In example below, parser reaches state httpParserStates.HEADERS_COMPLETE
and it is responsibility of callee to change state to httpParserStates.COMPLETE
when server stream closes.
See https://github.com/abhinavsingh/proxy.py/issues/20 for details.
Post commit https://github.com/abhinavsingh/proxy.py/commit/269484df2e89bc659124177d339d4fc59f280cba
HttpParser would reach state COMPLETE also for RESPONSE_PARSER types and no longer
it is callee responsibility to change state on stream close. This was important because
pipelined responses not trigger stream close but may receive multiple responses.
"""
self.parser.type = httpParserTypes.RESPONSE_PARSER
self.parser.parse(b'HTTP/1.0 200 OK' + CRLF)
self.assertEqual(self.parser.code, b'200')
self.assertEqual(self.parser.version, b'HTTP/1.0')
self.assertEqual(self.parser.state, httpParserStates.LINE_RCVD)
self.parser.parse(
CRLF.join([
b'Server: BaseHTTP/0.3 Python/2.7.10',
b'Date: Thu, 13 Dec 2018 16:24:09 GMT', CRLF
]))
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_response_parse(self) -> None:
self.parser.type = httpParserTypes.RESPONSE_PARSER
self.parser.parse(b''.join([
b'HTTP/1.1 301 Moved Permanently\r\n',
b'Location: http://www.google.com/\r\n',
b'Content-Type: text/html; charset=UTF-8\r\n',
b'Date: Wed, 22 May 2013 14:07:29 GMT\r\n',
b'Expires: Fri, 21 Jun 2013 14:07:29 GMT\r\n',
b'Cache-Control: public, max-age=2592000\r\n', b'Server: gws\r\n',
b'Content-Length: 219\r\n', b'X-XSS-Protection: 1; mode=block\r\n',
b'X-Frame-Options: SAMEORIGIN\r\n\r\n',
b'<HTML><HEAD><meta http-equiv="content-type" content="text/html;charset=utf-8">\n'
+ b'<TITLE>301 Moved</TITLE></HEAD>',
b'<BODY>\n<H1>301 Moved</H1>\nThe document has moved\n' +
b'<A HREF="http://www.google.com/">here</A>.\r\n</BODY></HTML>\r\n'
]))
self.assertEqual(self.parser.code, b'301')
self.assertEqual(self.parser.reason, b'Moved Permanently')
self.assertEqual(self.parser.version, b'HTTP/1.1')
self.assertEqual(
self.parser.body,
b'<HTML><HEAD><meta http-equiv="content-type" content="text/html;charset=utf-8">\n'
+
b'<TITLE>301 Moved</TITLE></HEAD><BODY>\n<H1>301 Moved</H1>\nThe document has moved\n'
+
b'<A HREF="http://www.google.com/">here</A>.\r\n</BODY></HTML>\r\n'
)
self.assertEqual(self.parser.headers[b'content-length'],
(b'Content-Length', b'219'))
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_response_partial_parse(self) -> None:
self.parser.type = httpParserTypes.RESPONSE_PARSER
self.parser.parse(b''.join([
b'HTTP/1.1 301 Moved Permanently\r\n',
b'Location: http://www.google.com/\r\n',
b'Content-Type: text/html; charset=UTF-8\r\n',
b'Date: Wed, 22 May 2013 14:07:29 GMT\r\n',
b'Expires: Fri, 21 Jun 2013 14:07:29 GMT\r\n',
b'Cache-Control: public, max-age=2592000\r\n', b'Server: gws\r\n',
b'Content-Length: 219\r\n', b'X-XSS-Protection: 1; mode=block\r\n',
b'X-Frame-Options: SAMEORIGIN\r\n'
]))
self.assertEqual(self.parser.headers[b'x-frame-options'],
(b'X-Frame-Options', b'SAMEORIGIN'))
self.assertEqual(self.parser.state, httpParserStates.RCVING_HEADERS)
self.parser.parse(b'\r\n')
self.assertEqual(self.parser.state, httpParserStates.HEADERS_COMPLETE)
self.parser.parse(
b'<HTML><HEAD><meta http-equiv="content-type" content="text/html;charset=utf-8">\n'
+ b'<TITLE>301 Moved</TITLE></HEAD>')
self.assertEqual(self.parser.state, httpParserStates.RCVING_BODY)
self.parser.parse(
b'<BODY>\n<H1>301 Moved</H1>\nThe document has moved\n' +
b'<A HREF="http://www.google.com/">here</A>.\r\n</BODY></HTML>\r\n'
)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_chunked_response_parse(self) -> None:
self.parser.type = httpParserTypes.RESPONSE_PARSER
self.parser.parse(b''.join([
b'HTTP/1.1 200 OK\r\n', b'Content-Type: application/json\r\n',
b'Date: Wed, 22 May 2013 15:08:15 GMT\r\n',
b'Server: gunicorn/0.16.1\r\n', b'transfer-encoding: chunked\r\n',
b'Connection: keep-alive\r\n\r\n', b'4\r\n', b'Wiki\r\n', b'5\r\n',
b'pedia\r\n', b'E\r\n', b' in\r\n\r\nchunks.\r\n', b'0\r\n',
b'\r\n'
]))
self.assertEqual(self.parser.body, b'Wikipedia in\r\n\r\nchunks.')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_pipelined_response_parse(self) -> None:
response = build_http_response(
httpStatusCodes.OK,
reason=b'OK',
headers={b'Content-Length': b'15'},
body=b'{"key":"value"}',
)
self.assert_pipeline_response(response)
def test_pipelined_chunked_response_parse(self) -> None:
response = build_http_response(
httpStatusCodes.OK,
reason=b'OK',
headers={
b'Transfer-Encoding': b'chunked',
b'Content-Type': b'application/json',
},
body=b'f\r\n{"key":"value"}\r\n0\r\n\r\n')
self.assert_pipeline_response(response)
def assert_pipeline_response(self, response: bytes) -> None:
self.parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
self.parser.parse(response + response)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
self.assertEqual(self.parser.body, b'{"key":"value"}')
self.assertEqual(self.parser.buffer, response)
# parse buffer
parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
parser.parse(self.parser.buffer)
self.assertEqual(parser.state, httpParserStates.COMPLETE)
self.assertEqual(parser.body, b'{"key":"value"}')
self.assertEqual(parser.buffer, b'')
def test_chunked_request_parse(self) -> None:
self.parser.parse(
build_http_request(httpMethods.POST,
b'http://example.org/',
headers={
b'Transfer-Encoding': b'chunked',
b'Content-Type': b'application/json',
},
body=b'f\r\n{"key":"value"}\r\n0\r\n\r\n'))
self.assertEqual(self.parser.body, b'{"key":"value"}')
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
self.assertEqual(
self.parser.build(),
build_http_request(httpMethods.POST,
b'/',
headers={
b'Transfer-Encoding': b'chunked',
b'Content-Type': b'application/json',
},
body=b'f\r\n{"key":"value"}\r\n0\r\n\r\n'))
def test_is_http_1_1_keep_alive(self) -> None:
self.parser.parse(build_http_request(httpMethods.GET, b'/'))
self.assertTrue(self.parser.is_http_1_1_keep_alive())
def test_is_http_1_1_keep_alive_with_non_close_connection_header(
self) -> None:
self.parser.parse(
build_http_request(httpMethods.GET,
b'/',
headers={
b'Connection': b'keep-alive',
}))
self.assertTrue(self.parser.is_http_1_1_keep_alive())
def test_is_not_http_1_1_keep_alive_with_close_header(self) -> None:
self.parser.parse(
build_http_request(httpMethods.GET,
b'/',
headers={
b'Connection': b'close',
}))
self.assertFalse(self.parser.is_http_1_1_keep_alive())
def test_is_not_http_1_1_keep_alive_for_http_1_0(self) -> None:
self.parser.parse(
build_http_request(
httpMethods.GET,
b'/',
protocol_version=b'HTTP/1.0',
))
self.assertFalse(self.parser.is_http_1_1_keep_alive())
def test_paramiko_doc(self) -> None:
response = b'HTTP/1.1 304 Not Modified\r\nDate: Tue, 03 Dec 2019 02:31:55 GMT\r\nConnection: keep-alive' \
b'\r\nLast-Modified: Sun, 23 Jun 2019 22:58:21 GMT\r\nETag: "5d10040d-1af2c"' \
b'\r\nX-Cname-TryFiles: True\r\nX-Served: Nginx\r\nX-Deity: web02\r\nCF-Cache-Status: DYNAMIC' \
b'\r\nServer: cloudflare\r\nCF-RAY: 53f2208c6fef6c38-SJC\r\n\r\n'
self.parser = HttpParser(httpParserTypes.RESPONSE_PARSER)
self.parser.parse(response)
self.assertEqual(self.parser.state, httpParserStates.COMPLETE)
def test_response_factory(self) -> None:
r = HttpParser.response(b'HTTP/1.1 200 OK\r\nkey: value\r\n\r\n', )
self.assertEqual(r.code, b'200')
self.assertEqual(r.reason, b'OK')
self.assertEqual(r.header(b'key'), b'value')
async def test_man_in_the_middle_plugin(self) -> None:
request = build_http_request(
b'GET',
b'http://super.secure/',
headers={
b'Host': b'super.secure',
},
no_ua=True,
)
self._conn.recv.return_value = request
server = self.mock_server_conn.return_value
server.connect.return_value = True
def has_buffer() -> bool:
return cast(bool, server.queue.called)
def closed() -> bool:
return not server.connect.called
server.has_buffer.side_effect = has_buffer
type(server).closed = mock.PropertyMock(side_effect=closed)
self.mock_selector.return_value.select.side_effect = [
[(
selectors.SelectorKey(
fileobj=self._conn.fileno(),
fd=self._conn.fileno(),
events=selectors.EVENT_READ,
data=None,
),
selectors.EVENT_READ,
)],
[(
selectors.SelectorKey(
fileobj=server.connection.fileno(),
fd=server.connection.fileno(),
events=selectors.EVENT_WRITE,
data=None,
),
selectors.EVENT_WRITE,
)],
[(
selectors.SelectorKey(
fileobj=server.connection.fileno(),
fd=server.connection.fileno(),
events=selectors.EVENT_READ,
data=None,
),
selectors.EVENT_READ,
)],
]
# Client read
await self.protocol_handler._run_once()
self.mock_server_conn.assert_called_with(
'super.secure',
DEFAULT_HTTP_PORT,
)
server.connect.assert_called_once()
queued_request = \
build_http_request(
b'GET', b'/',
headers={
b'Host': b'super.secure',
b'Via': b'1.1 %s' % PROXY_AGENT_HEADER_VALUE,
},
no_ua=True,
)
server.queue.assert_called_once()
print(server.queue.call_args_list[0][0][0].tobytes())
print(queued_request)
self.assertEqual(server.queue.call_args_list[0][0][0], queued_request)
# Server write
await self.protocol_handler._run_once()
server.flush.assert_called_once()
# Server read
server.recv.return_value = \
build_http_response(
httpStatusCodes.OK,
reason=b'OK',
body=b'Original Response From Upstream',
)
await self.protocol_handler._run_once()
response = HttpParser(httpParserTypes.RESPONSE_PARSER)
response.parse(self.protocol_handler.work.buffer[0])
assert response.body
self.assertEqual(
gzip.decompress(response.body),
b'Hello from man in the middle',
)
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.request = HttpParser(httpParserTypes.REQUEST_PARSER)
self.response = HttpParser(httpParserTypes.RESPONSE_PARSER)
def __init__(self, *args: Any, **kwargs: Any) -> None:
super().__init__(*args, **kwargs)
self.response = HttpParser(httpParserTypes.RESPONSE_PARSER)
def test_proxy_protocol_not_for_response_parser(self) -> None:
with self.assertRaises(AssertionError):
HttpParser(
httpParserTypes.RESPONSE_PARSER,
enable_proxy_protocol=True,
)
