def _read_request_line(
line: bytes) -> Tuple[str, int, bytes, bytes, bytes, bytes, bytes]:
try:
method, target, http_version = line.split()
port: Optional[int]
if target == b"*" or target.startswith(b"/"):
scheme, authority, path = b"", b"", target
host, port = "", 0
elif method == b"CONNECT":
scheme, authority, path = b"", target, b""
host, port = url.parse_authority(authority, check=True)
if not port:
raise ValueError
else:
scheme, rest = target.split(b"://", maxsplit=1)
authority, path_ = rest.split(b"/", maxsplit=1)
path = b"/" + path_
host, port = url.parse_authority(authority, check=True)
port = port or url.default_port(scheme)
if not port:
raise ValueError
# TODO: we can probably get rid of this check?
url.parse(target)
raise_if_http_version_unknown(http_version)
except ValueError as e:
raise ValueError(f"Bad HTTP request line: {line!r}") from e
return host, port, method, scheme, authority, path, http_version
def read_request_headers(self, flow):
self.request_message.arrived.wait()
self.raise_zombie()
if self.pushed:
flow.metadata['h2-pushed-stream'] = True
# pseudo header must be present, see https://http2.github.io/http2-spec/#rfc.section.8.1.2.3
authority = self.request_message.headers.pop(':authority', "")
method = self.request_message.headers.pop(':method')
scheme = self.request_message.headers.pop(':scheme')
path = self.request_message.headers.pop(':path')
host, port = url.parse_authority(authority, check=True)
port = port or url.default_port(scheme) or 0
return http.HTTPRequest(
host,
port,
method.encode(),
scheme.encode(),
authority.encode(),
path.encode(),
b"HTTP/2.0",
self.request_message.headers,
None,
None,
self.timestamp_start,
self.timestamp_end,
)
def _read_request_line(rfile):
try:
line = _get_first_line(rfile)
except exceptions.HttpReadDisconnect:
# We want to provide a better error message.
raise exceptions.HttpReadDisconnect("Client disconnected")
try:
method, target, http_version = line.split()
if target == b"*" or target.startswith(b"/"):
scheme, authority, path = b"", b"", target
host, port = "", 0
elif method == b"CONNECT":
scheme, authority, path = b"", target, b""
host, port = url.parse_authority(authority, check=True)
if not port:
raise ValueError
else:
scheme, rest = target.split(b"://", maxsplit=1)
authority, path_ = rest.split(b"/", maxsplit=1)
path = b"/" + path_
host, port = url.parse_authority(authority, check=True)
port = port or url.default_port(scheme)
if not port:
raise ValueError
# TODO: we can probably get rid of this check?
url.parse(target)
_check_http_version(http_version)
except ValueError:
raise exceptions.HttpSyntaxException(f"Bad HTTP request line: {line}")
return host, port, method, scheme, authority, path, http_version
def read_request(
self,
__rfile,
include_body=True,
body_size_limit=None,
allow_empty=False,
):
if body_size_limit is not None:
raise NotImplementedError()
self.perform_connection_preface()
timestamp_start = time.time()
if hasattr(self.tcp_handler.rfile, "reset_timestamps"):
self.tcp_handler.rfile.reset_timestamps()
stream_id, headers, body = self._receive_transmission(
include_body=include_body,
)
if hasattr(self.tcp_handler.rfile, "first_byte_timestamp"):
# more accurate timestamp_start
timestamp_start = self.tcp_handler.rfile.first_byte_timestamp
timestamp_end = time.time()
# pseudo header must be present, see https://http2.github.io/http2-spec/#rfc.section.8.1.2.3
authority = headers.pop(':authority', "")
method = headers.pop(':method', "")
scheme = headers.pop(':scheme', "")
path = headers.pop(':path', "")
host, port = url.parse_authority(authority, check=False)
port = port or url.default_port(scheme) or 0
request = mitmproxy.net.http.Request(
host=host,
port=port,
method=method.encode(),
scheme=scheme.encode(),
authority=authority.encode(),
path=path.encode(),
http_version=b"HTTP/2.0",
headers=headers,
content=body,
trailers=None,
timestamp_start=timestamp_start,
timestamp_end=timestamp_end,
)
request.stream_id = stream_id
return request
def pretty_url(self) -> str:
"""
Like :py:attr:`url`, but using :py:attr:`pretty_host` instead of :py:attr:`host`.
"""
if self.first_line_format == "authority":
return self.authority
host_header = self.host_header
if not host_header:
return self.url
pretty_host, pretty_port = url.parse_authority(host_header,
check=False)
pretty_port = pretty_port or url.default_port(self.scheme) or 443
return url.unparse(self.scheme, pretty_host, pretty_port, self.path)
def test_default_port():
assert url.default_port("http") == 80
assert url.default_port(b"https") == 443
assert url.default_port(b"qux") is None
