Update request urllib3 (#464)

* requests: updated to version 2.27.1

* urllib3: updated to version 1.26.18
This commit is contained in:
ilmich
2024-01-29 20:12:20 +01:00
committed by GitHub
parent 349752a380
commit b96b63448c
51 changed files with 2414 additions and 801 deletions
+223 -99
View File
@@ -1,13 +1,18 @@
from __future__ import absolute_import
import datetime
import logging
import os
import re
import socket
from socket import error as SocketError, timeout as SocketTimeout
import warnings
from socket import error as SocketError
from socket import timeout as SocketTimeout
from .packages import six
from .packages.six.moves.http_client import HTTPConnection as _HTTPConnection
from .packages.six.moves.http_client import HTTPException # noqa: F401
from .util.proxy import create_proxy_ssl_context
try: # Compiled with SSL?
import ssl
@@ -29,26 +34,33 @@ except NameError:
pass
try: # Python 3:
# Not a no-op, we're adding this to the namespace so it can be imported.
BrokenPipeError = BrokenPipeError
except NameError: # Python 2:
class BrokenPipeError(Exception):
pass
from ._collections import HTTPHeaderDict # noqa (historical, removed in v2)
from ._version import __version__
from .exceptions import (
NewConnectionError,
ConnectTimeoutError,
NewConnectionError,
SubjectAltNameWarning,
SystemTimeWarning,
)
from .packages.ssl_match_hostname import match_hostname, CertificateError
from .util import SKIP_HEADER, SKIPPABLE_HEADERS, connection
from .util.ssl_ import (
resolve_cert_reqs,
resolve_ssl_version,
assert_fingerprint,
create_urllib3_context,
is_ipaddress,
resolve_cert_reqs,
resolve_ssl_version,
ssl_wrap_socket,
)
from .util import connection
from ._collections import HTTPHeaderDict
from .util.ssl_match_hostname import CertificateError, match_hostname
log = logging.getLogger(__name__)
@@ -56,37 +68,35 @@ port_by_scheme = {"http": 80, "https": 443}
# When it comes time to update this value as a part of regular maintenance
# (ie test_recent_date is failing) update it to ~6 months before the current date.
RECENT_DATE = datetime.date(2019, 1, 1)
RECENT_DATE = datetime.date(2022, 1, 1)
class DummyConnection(object):
"""Used to detect a failed ConnectionCls import."""
pass
_CONTAINS_CONTROL_CHAR_RE = re.compile(r"[^-!#$%&'*+.^_`|~0-9a-zA-Z]")
class HTTPConnection(_HTTPConnection, object):
"""
Based on httplib.HTTPConnection but provides an extra constructor
Based on :class:`http.client.HTTPConnection` but provides an extra constructor
backwards-compatibility layer between older and newer Pythons.
Additional keyword parameters are used to configure attributes of the connection.
Accepted parameters include:
- ``strict``: See the documentation on :class:`urllib3.connectionpool.HTTPConnectionPool`
- ``source_address``: Set the source address for the current connection.
- ``socket_options``: Set specific options on the underlying socket. If not specified, then
defaults are loaded from ``HTTPConnection.default_socket_options`` which includes disabling
Nagle's algorithm (sets TCP_NODELAY to 1) unless the connection is behind a proxy.
- ``strict``: See the documentation on :class:`urllib3.connectionpool.HTTPConnectionPool`
- ``source_address``: Set the source address for the current connection.
- ``socket_options``: Set specific options on the underlying socket. If not specified, then
defaults are loaded from ``HTTPConnection.default_socket_options`` which includes disabling
Nagle's algorithm (sets TCP_NODELAY to 1) unless the connection is behind a proxy.
For example, if you wish to enable TCP Keep Alive in addition to the defaults,
you might pass::
For example, if you wish to enable TCP Keep Alive in addition to the defaults,
you might pass:
HTTPConnection.default_socket_options + [
(socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1),
]
.. code-block:: python
Or you may want to disable the defaults by passing an empty list (e.g., ``[]``).
HTTPConnection.default_socket_options + [
(socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1),
]
Or you may want to disable the defaults by passing an empty list (e.g., ``[]``).
"""
default_port = port_by_scheme["http"]
@@ -98,6 +108,10 @@ class HTTPConnection(_HTTPConnection, object):
#: Whether this connection verifies the host's certificate.
is_verified = False
#: Whether this proxy connection (if used) verifies the proxy host's
#: certificate.
proxy_is_verified = None
def __init__(self, *args, **kw):
if not six.PY2:
kw.pop("strict", None)
@@ -109,6 +123,10 @@ class HTTPConnection(_HTTPConnection, object):
#: provided, we use the default options.
self.socket_options = kw.pop("socket_options", self.default_socket_options)
# Proxy options provided by the user.
self.proxy = kw.pop("proxy", None)
self.proxy_config = kw.pop("proxy_config", None)
_HTTPConnection.__init__(self, *args, **kw)
@property
@@ -141,7 +159,7 @@ class HTTPConnection(_HTTPConnection, object):
self._dns_host = value
def _new_conn(self):
""" Establish a socket connection and set nodelay settings on it.
"""Establish a socket connection and set nodelay settings on it.
:return: New socket connection.
"""
@@ -171,10 +189,13 @@ class HTTPConnection(_HTTPConnection, object):
return conn
def _is_using_tunnel(self):
# Google App Engine's httplib does not define _tunnel_host
return getattr(self, "_tunnel_host", None)
def _prepare_conn(self, conn):
self.sock = conn
# Google App Engine's httplib does not define _tunnel_host
if getattr(self, "_tunnel_host", None):
if self._is_using_tunnel():
# TODO: Fix tunnel so it doesn't depend on self.sock state.
self._tunnel()
# Mark this connection as not reusable
@@ -184,20 +205,61 @@ class HTTPConnection(_HTTPConnection, object):
conn = self._new_conn()
self._prepare_conn(conn)
def putrequest(self, method, url, *args, **kwargs):
""" """
# Empty docstring because the indentation of CPython's implementation
# is broken but we don't want this method in our documentation.
match = _CONTAINS_CONTROL_CHAR_RE.search(method)
if match:
raise ValueError(
"Method cannot contain non-token characters %r (found at least %r)"
% (method, match.group())
)
return _HTTPConnection.putrequest(self, method, url, *args, **kwargs)
def putheader(self, header, *values):
""" """
if not any(isinstance(v, str) and v == SKIP_HEADER for v in values):
_HTTPConnection.putheader(self, header, *values)
elif six.ensure_str(header.lower()) not in SKIPPABLE_HEADERS:
raise ValueError(
"urllib3.util.SKIP_HEADER only supports '%s'"
% ("', '".join(map(str.title, sorted(SKIPPABLE_HEADERS))),)
)
def request(self, method, url, body=None, headers=None):
# Update the inner socket's timeout value to send the request.
# This only triggers if the connection is re-used.
if getattr(self, "sock", None) is not None:
self.sock.settimeout(self.timeout)
if headers is None:
headers = {}
else:
# Avoid modifying the headers passed into .request()
headers = headers.copy()
if "user-agent" not in (six.ensure_str(k.lower()) for k in headers):
headers["User-Agent"] = _get_default_user_agent()
super(HTTPConnection, self).request(method, url, body=body, headers=headers)
def request_chunked(self, method, url, body=None, headers=None):
"""
Alternative to the common request method, which sends the
body with chunked encoding and not as one block
"""
headers = HTTPHeaderDict(headers if headers is not None else {})
skip_accept_encoding = "accept-encoding" in headers
skip_host = "host" in headers
headers = headers or {}
header_keys = set([six.ensure_str(k.lower()) for k in headers])
skip_accept_encoding = "accept-encoding" in header_keys
skip_host = "host" in header_keys
self.putrequest(
method, url, skip_accept_encoding=skip_accept_encoding, skip_host=skip_host
)
if "user-agent" not in header_keys:
self.putheader("User-Agent", _get_default_user_agent())
for header, value in headers.items():
self.putheader(header, value)
if "transfer-encoding" not in headers:
if "transfer-encoding" not in header_keys:
self.putheader("Transfer-Encoding", "chunked")
self.endheaders()
@@ -211,19 +273,31 @@ class HTTPConnection(_HTTPConnection, object):
if not isinstance(chunk, bytes):
chunk = chunk.encode("utf8")
len_str = hex(len(chunk))[2:]
self.send(len_str.encode("utf-8"))
self.send(b"\r\n")
self.send(chunk)
self.send(b"\r\n")
to_send = bytearray(len_str.encode())
to_send += b"\r\n"
to_send += chunk
to_send += b"\r\n"
self.send(to_send)
# After the if clause, to always have a closed body
self.send(b"0\r\n\r\n")
class HTTPSConnection(HTTPConnection):
"""
Many of the parameters to this constructor are passed to the underlying SSL
socket by means of :py:func:`urllib3.util.ssl_wrap_socket`.
"""
default_port = port_by_scheme["https"]
cert_reqs = None
ca_certs = None
ca_cert_dir = None
ca_cert_data = None
ssl_version = None
assert_fingerprint = None
tls_in_tls_required = False
def __init__(
self,
@@ -251,53 +325,6 @@ class HTTPSConnection(HTTPConnection):
# HTTPS requests to go out as HTTP. (See Issue #356)
self._protocol = "https"
def connect(self):
conn = self._new_conn()
self._prepare_conn(conn)
# Wrap socket using verification with the root certs in
# trusted_root_certs
default_ssl_context = False
if self.ssl_context is None:
default_ssl_context = True
self.ssl_context = create_urllib3_context(
ssl_version=resolve_ssl_version(self.ssl_version),
cert_reqs=resolve_cert_reqs(self.cert_reqs),
)
# Try to load OS default certs if none are given.
# Works well on Windows (requires Python3.4+)
context = self.ssl_context
if (
not self.ca_certs
and not self.ca_cert_dir
and default_ssl_context
and hasattr(context, "load_default_certs")
):
context.load_default_certs()
self.sock = ssl_wrap_socket(
sock=conn,
keyfile=self.key_file,
certfile=self.cert_file,
key_password=self.key_password,
ssl_context=self.ssl_context,
server_hostname=self.server_hostname,
)
class VerifiedHTTPSConnection(HTTPSConnection):
"""
Based on httplib.HTTPSConnection but wraps the socket with
SSL certification.
"""
cert_reqs = None
ca_certs = None
ca_cert_dir = None
ssl_version = None
assert_fingerprint = None
def set_cert(
self,
key_file=None,
@@ -308,6 +335,7 @@ class VerifiedHTTPSConnection(HTTPSConnection):
assert_hostname=None,
assert_fingerprint=None,
ca_cert_dir=None,
ca_cert_data=None,
):
"""
This method should only be called once, before the connection is used.
@@ -328,15 +356,19 @@ class VerifiedHTTPSConnection(HTTPSConnection):
self.assert_fingerprint = assert_fingerprint
self.ca_certs = ca_certs and os.path.expanduser(ca_certs)
self.ca_cert_dir = ca_cert_dir and os.path.expanduser(ca_cert_dir)
self.ca_cert_data = ca_cert_data
def connect(self):
# Add certificate verification
conn = self._new_conn()
self.sock = conn = self._new_conn()
hostname = self.host
tls_in_tls = False
if self._is_using_tunnel():
if self.tls_in_tls_required:
self.sock = conn = self._connect_tls_proxy(hostname, conn)
tls_in_tls = True
# Google App Engine's httplib does not define _tunnel_host
if getattr(self, "_tunnel_host", None):
self.sock = conn
# Calls self._set_hostport(), so self.host is
# self._tunnel_host below.
self._tunnel()
@@ -378,6 +410,7 @@ class VerifiedHTTPSConnection(HTTPSConnection):
if (
not self.ca_certs
and not self.ca_cert_dir
and not self.ca_cert_data
and default_ssl_context
and hasattr(context, "load_default_certs")
):
@@ -390,10 +423,29 @@ class VerifiedHTTPSConnection(HTTPSConnection):
key_password=self.key_password,
ca_certs=self.ca_certs,
ca_cert_dir=self.ca_cert_dir,
ca_cert_data=self.ca_cert_data,
server_hostname=server_hostname,
ssl_context=context,
tls_in_tls=tls_in_tls,
)
# If we're using all defaults and the connection
# is TLSv1 or TLSv1.1 we throw a DeprecationWarning
# for the host.
if (
default_ssl_context
and self.ssl_version is None
and hasattr(self.sock, "version")
and self.sock.version() in {"TLSv1", "TLSv1.1"}
):
warnings.warn(
"Negotiating TLSv1/TLSv1.1 by default is deprecated "
"and will be disabled in urllib3 v2.0.0. Connecting to "
"'%s' with '%s' can be enabled by explicitly opting-in "
"with 'ssl_version'" % (self.host, self.sock.version()),
DeprecationWarning,
)
if self.assert_fingerprint:
assert_fingerprint(
self.sock.getpeercert(binary_form=True), self.assert_fingerprint
@@ -412,7 +464,7 @@ class VerifiedHTTPSConnection(HTTPSConnection):
(
"Certificate for {0} has no `subjectAltName`, falling back to check for a "
"`commonName` for now. This feature is being removed by major browsers and "
"deprecated by RFC 2818. (See https://github.com/shazow/urllib3/issues/497 "
"deprecated by RFC 2818. (See https://github.com/urllib3/urllib3/issues/497 "
"for details.)".format(hostname)
),
SubjectAltNameWarning,
@@ -424,13 +476,76 @@ class VerifiedHTTPSConnection(HTTPSConnection):
or self.assert_fingerprint is not None
)
def _connect_tls_proxy(self, hostname, conn):
"""
Establish a TLS connection to the proxy using the provided SSL context.
"""
proxy_config = self.proxy_config
ssl_context = proxy_config.ssl_context
if ssl_context:
# If the user provided a proxy context, we assume CA and client
# certificates have already been set
return ssl_wrap_socket(
sock=conn,
server_hostname=hostname,
ssl_context=ssl_context,
)
ssl_context = create_proxy_ssl_context(
self.ssl_version,
self.cert_reqs,
self.ca_certs,
self.ca_cert_dir,
self.ca_cert_data,
)
# If no cert was provided, use only the default options for server
# certificate validation
socket = ssl_wrap_socket(
sock=conn,
ca_certs=self.ca_certs,
ca_cert_dir=self.ca_cert_dir,
ca_cert_data=self.ca_cert_data,
server_hostname=hostname,
ssl_context=ssl_context,
)
if ssl_context.verify_mode != ssl.CERT_NONE and not getattr(
ssl_context, "check_hostname", False
):
# While urllib3 attempts to always turn off hostname matching from
# the TLS library, this cannot always be done. So we check whether
# the TLS Library still thinks it's matching hostnames.
cert = socket.getpeercert()
if not cert.get("subjectAltName", ()):
warnings.warn(
(
"Certificate for {0} has no `subjectAltName`, falling back to check for a "
"`commonName` for now. This feature is being removed by major browsers and "
"deprecated by RFC 2818. (See https://github.com/urllib3/urllib3/issues/497 "
"for details.)".format(hostname)
),
SubjectAltNameWarning,
)
_match_hostname(cert, hostname)
self.proxy_is_verified = ssl_context.verify_mode == ssl.CERT_REQUIRED
return socket
def _match_hostname(cert, asserted_hostname):
# Our upstream implementation of ssl.match_hostname()
# only applies this normalization to IP addresses so it doesn't
# match DNS SANs so we do the same thing!
stripped_hostname = asserted_hostname.strip("u[]")
if is_ipaddress(stripped_hostname):
asserted_hostname = stripped_hostname
try:
match_hostname(cert, asserted_hostname)
except CertificateError as e:
log.warning(
"Certificate did not match expected hostname: %s. " "Certificate: %s",
"Certificate did not match expected hostname: %s. Certificate: %s",
asserted_hostname,
cert,
)
@@ -440,9 +555,18 @@ def _match_hostname(cert, asserted_hostname):
raise
if ssl:
# Make a copy for testing.
UnverifiedHTTPSConnection = HTTPSConnection
HTTPSConnection = VerifiedHTTPSConnection
else:
HTTPSConnection = DummyConnection
def _get_default_user_agent():
return "python-urllib3/%s" % __version__
class DummyConnection(object):
"""Used to detect a failed ConnectionCls import."""
pass
if not ssl:
HTTPSConnection = DummyConnection # noqa: F811
VerifiedHTTPSConnection = HTTPSConnection