2021-06-03 09:43:42 +00:00
|
|
|
#!/usr/bin/env python3
|
2022-06-24 11:06:16 +00:00
|
|
|
|
2015-01-30 01:57:37 +00:00
|
|
|
# Allow direct execution
|
|
|
|
|
import os
|
|
|
|
|
import sys
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
import pytest
|
2022-04-11 22:32:57 +00:00
|
|
|
|
2015-01-30 01:57:37 +00:00
|
|
|
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
|
|
|
|
|
|
2023-05-27 07:06:13 +00:00
|
|
|
import gzip
|
2023-07-15 10:25:23 +00:00
|
|
|
import http.client
|
2023-05-27 07:06:13 +00:00
|
|
|
import http.cookiejar
|
2022-06-24 11:06:16 +00:00
|
|
|
import http.server
|
2023-05-27 07:06:13 +00:00
|
|
|
import io
|
2024-02-17 22:32:34 +00:00
|
|
|
import logging
|
2023-05-27 07:06:13 +00:00
|
|
|
import pathlib
|
2023-07-15 10:25:23 +00:00
|
|
|
import random
|
2022-04-11 22:32:57 +00:00
|
|
|
import ssl
|
2023-05-27 07:06:13 +00:00
|
|
|
import tempfile
|
2022-04-11 22:32:57 +00:00
|
|
|
import threading
|
2023-07-15 10:25:23 +00:00
|
|
|
import time
|
2023-05-27 07:06:13 +00:00
|
|
|
import urllib.error
|
2022-06-24 08:10:17 +00:00
|
|
|
import urllib.request
|
2023-07-15 10:25:23 +00:00
|
|
|
import warnings
|
2023-05-27 10:40:05 +00:00
|
|
|
import zlib
|
2023-07-15 10:25:23 +00:00
|
|
|
from email.message import Message
|
|
|
|
|
from http.cookiejar import CookieJar
|
2022-04-11 22:32:57 +00:00
|
|
|
|
2024-03-17 03:47:56 +00:00
|
|
|
from test.conftest import validate_and_send
|
2024-01-19 21:39:49 +00:00
|
|
|
from test.helper import FakeYDL, http_server_port, verify_address_availability
|
2023-07-29 22:06:17 +00:00
|
|
|
from yt_dlp.cookies import YoutubeDLCookieJar
|
2024-03-17 03:52:38 +00:00
|
|
|
from yt_dlp.dependencies import brotli, curl_cffi, requests, urllib3
|
2023-07-15 10:25:23 +00:00
|
|
|
from yt_dlp.networking import (
|
|
|
|
|
HEADRequest,
|
|
|
|
|
PUTRequest,
|
|
|
|
|
Request,
|
|
|
|
|
RequestDirector,
|
|
|
|
|
RequestHandler,
|
|
|
|
|
Response,
|
|
|
|
|
)
|
|
|
|
|
from yt_dlp.networking._urllib import UrllibRH
|
|
|
|
|
from yt_dlp.networking.exceptions import (
|
|
|
|
|
CertificateVerifyError,
|
|
|
|
|
HTTPError,
|
|
|
|
|
IncompleteRead,
|
|
|
|
|
NoSupportingHandlers,
|
2023-10-13 23:33:00 +00:00
|
|
|
ProxyError,
|
2023-07-15 10:25:23 +00:00
|
|
|
RequestError,
|
|
|
|
|
SSLError,
|
|
|
|
|
TransportError,
|
|
|
|
|
UnsupportedRequest,
|
|
|
|
|
)
|
2024-03-17 03:47:56 +00:00
|
|
|
from yt_dlp.networking.impersonate import (
|
|
|
|
|
ImpersonateRequestHandler,
|
|
|
|
|
ImpersonateTarget,
|
|
|
|
|
)
|
|
|
|
|
from yt_dlp.utils import YoutubeDLError
|
2023-07-15 10:25:23 +00:00
|
|
|
from yt_dlp.utils._utils import _YDLLogger as FakeLogger
|
2024-03-17 03:52:38 +00:00
|
|
|
from yt_dlp.utils.networking import HTTPHeaderDict, std_headers
|
2015-01-30 01:57:37 +00:00
|
|
|
|
|
|
|
|
TEST_DIR = os.path.dirname(os.path.abspath(__file__))
|
|
|
|
|
|
2015-01-30 02:06:40 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def _build_proxy_handler(name):
|
|
|
|
|
class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
|
|
|
|
|
proxy_name = name
|
|
|
|
|
|
|
|
|
|
def log_message(self, format, *args):
|
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
|
def do_GET(self):
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/plain; charset=utf-8')
|
|
|
|
|
self.end_headers()
|
2024-03-10 15:18:44 +00:00
|
|
|
self.wfile.write(f'{self.proxy_name}: {self.path}'.encode())
|
2023-07-15 10:25:23 +00:00
|
|
|
return HTTPTestRequestHandler
|
|
|
|
|
|
|
|
|
|
|
2022-06-24 08:10:17 +00:00
|
|
|
class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
|
2023-05-27 07:06:13 +00:00
|
|
|
protocol_version = 'HTTP/1.1'
|
2024-03-17 03:52:38 +00:00
|
|
|
default_request_version = 'HTTP/1.1'
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2015-01-30 01:57:37 +00:00
|
|
|
def log_message(self, format, *args):
|
|
|
|
|
pass
|
|
|
|
|
|
2023-05-27 07:06:13 +00:00
|
|
|
def _headers(self):
|
2023-07-15 10:25:23 +00:00
|
|
|
payload = str(self.headers).encode()
|
2023-05-27 07:06:13 +00:00
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'application/json')
|
|
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
|
|
|
|
|
def _redirect(self):
|
|
|
|
|
self.send_response(int(self.path[len('/redirect_'):]))
|
|
|
|
|
self.send_header('Location', '/method')
|
|
|
|
|
self.send_header('Content-Length', '0')
|
|
|
|
|
self.end_headers()
|
|
|
|
|
|
|
|
|
|
def _method(self, method, payload=None):
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Length', str(len(payload or '')))
|
|
|
|
|
self.send_header('Method', method)
|
|
|
|
|
self.end_headers()
|
|
|
|
|
if payload:
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
|
|
|
|
|
def _status(self, status):
|
|
|
|
|
payload = f'<html>{status} NOT FOUND</html>'.encode()
|
|
|
|
|
self.send_response(int(status))
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
|
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
|
|
|
|
|
def _read_data(self):
|
|
|
|
|
if 'Content-Length' in self.headers:
|
|
|
|
|
return self.rfile.read(int(self.headers['Content-Length']))
|
2024-03-17 03:52:38 +00:00
|
|
|
else:
|
|
|
|
|
return b''
|
2023-05-27 07:06:13 +00:00
|
|
|
|
|
|
|
|
def do_POST(self):
|
2023-07-15 10:25:23 +00:00
|
|
|
data = self._read_data() + str(self.headers).encode()
|
2023-05-27 07:06:13 +00:00
|
|
|
if self.path.startswith('/redirect_'):
|
|
|
|
|
self._redirect()
|
|
|
|
|
elif self.path.startswith('/method'):
|
|
|
|
|
self._method('POST', data)
|
|
|
|
|
elif self.path.startswith('/headers'):
|
|
|
|
|
self._headers()
|
|
|
|
|
else:
|
|
|
|
|
self._status(404)
|
|
|
|
|
|
|
|
|
|
def do_HEAD(self):
|
|
|
|
|
if self.path.startswith('/redirect_'):
|
|
|
|
|
self._redirect()
|
|
|
|
|
elif self.path.startswith('/method'):
|
|
|
|
|
self._method('HEAD')
|
|
|
|
|
else:
|
|
|
|
|
self._status(404)
|
|
|
|
|
|
|
|
|
|
def do_PUT(self):
|
2023-07-15 10:25:23 +00:00
|
|
|
data = self._read_data() + str(self.headers).encode()
|
2023-05-27 07:06:13 +00:00
|
|
|
if self.path.startswith('/redirect_'):
|
|
|
|
|
self._redirect()
|
|
|
|
|
elif self.path.startswith('/method'):
|
|
|
|
|
self._method('PUT', data)
|
|
|
|
|
else:
|
|
|
|
|
self._status(404)
|
|
|
|
|
|
2015-01-30 01:57:37 +00:00
|
|
|
def do_GET(self):
|
|
|
|
|
if self.path == '/video.html':
|
2023-05-27 07:06:13 +00:00
|
|
|
payload = b'<html><video src="/vid.mp4" /></html>'
|
2015-01-30 01:57:37 +00:00
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
2023-07-15 10:25:23 +00:00
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
2015-01-30 01:57:37 +00:00
|
|
|
self.end_headers()
|
2023-05-27 07:06:13 +00:00
|
|
|
self.wfile.write(payload)
|
2015-01-30 01:57:37 +00:00
|
|
|
elif self.path == '/vid.mp4':
|
2023-05-27 07:06:13 +00:00
|
|
|
payload = b'\x00\x00\x00\x00\x20\x66\x74[video]'
|
2015-01-30 01:57:37 +00:00
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'video/mp4')
|
2023-05-27 07:06:13 +00:00
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
2015-01-30 01:57:37 +00:00
|
|
|
self.end_headers()
|
2023-05-27 07:06:13 +00:00
|
|
|
self.wfile.write(payload)
|
2016-05-26 09:24:40 +00:00
|
|
|
elif self.path == '/%E4%B8%AD%E6%96%87.html':
|
2023-05-27 07:06:13 +00:00
|
|
|
payload = b'<html><video src="/vid.mp4" /></html>'
|
2016-05-26 09:24:40 +00:00
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
2023-05-27 07:06:13 +00:00
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
elif self.path == '/%c7%9f':
|
|
|
|
|
payload = b'<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
|
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
2023-07-15 10:25:23 +00:00
|
|
|
elif self.path.startswith('/redirect_loop'):
|
|
|
|
|
self.send_response(301)
|
|
|
|
|
self.send_header('Location', self.path)
|
|
|
|
|
self.send_header('Content-Length', '0')
|
|
|
|
|
self.end_headers()
|
2023-07-28 22:40:20 +00:00
|
|
|
elif self.path == '/redirect_dotsegments':
|
|
|
|
|
self.send_response(301)
|
|
|
|
|
# redirect to /headers but with dot segments before
|
|
|
|
|
self.send_header('Location', '/a/b/./../../headers')
|
|
|
|
|
self.send_header('Content-Length', '0')
|
|
|
|
|
self.end_headers()
|
2024-01-20 21:03:33 +00:00
|
|
|
elif self.path == '/redirect_dotsegments_absolute':
|
|
|
|
|
self.send_response(301)
|
|
|
|
|
# redirect to /headers but with dot segments before - absolute url
|
|
|
|
|
self.send_header('Location', f'http://127.0.0.1:{http_server_port(self.server)}/a/b/./../../headers')
|
|
|
|
|
self.send_header('Content-Length', '0')
|
|
|
|
|
self.end_headers()
|
2023-05-27 07:06:13 +00:00
|
|
|
elif self.path.startswith('/redirect_'):
|
|
|
|
|
self._redirect()
|
|
|
|
|
elif self.path.startswith('/method'):
|
2023-07-15 10:25:23 +00:00
|
|
|
self._method('GET', str(self.headers).encode())
|
2023-05-27 07:06:13 +00:00
|
|
|
elif self.path.startswith('/headers'):
|
|
|
|
|
self._headers()
|
2023-06-06 08:44:51 +00:00
|
|
|
elif self.path.startswith('/308-to-headers'):
|
|
|
|
|
self.send_response(308)
|
2024-03-17 03:52:38 +00:00
|
|
|
# redirect to "localhost" for testing cookie redirection handling
|
|
|
|
|
self.send_header('Location', f'http://localhost:{self.connection.getsockname()[1]}/headers')
|
2023-06-06 08:44:51 +00:00
|
|
|
self.send_header('Content-Length', '0')
|
|
|
|
|
self.end_headers()
|
2023-05-27 07:06:13 +00:00
|
|
|
elif self.path == '/trailing_garbage':
|
|
|
|
|
payload = b'<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
|
|
|
|
self.send_header('Content-Encoding', 'gzip')
|
|
|
|
|
buf = io.BytesIO()
|
|
|
|
|
with gzip.GzipFile(fileobj=buf, mode='wb') as f:
|
|
|
|
|
f.write(payload)
|
|
|
|
|
compressed = buf.getvalue() + b'trailing garbage'
|
|
|
|
|
self.send_header('Content-Length', str(len(compressed)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(compressed)
|
|
|
|
|
elif self.path == '/302-non-ascii-redirect':
|
|
|
|
|
new_url = f'http://127.0.0.1:{http_server_port(self.server)}/中文.html'
|
|
|
|
|
self.send_response(301)
|
|
|
|
|
self.send_header('Location', new_url)
|
|
|
|
|
self.send_header('Content-Length', '0')
|
2016-05-26 09:24:40 +00:00
|
|
|
self.end_headers()
|
2023-05-27 10:40:05 +00:00
|
|
|
elif self.path == '/content-encoding':
|
|
|
|
|
encodings = self.headers.get('ytdl-encoding', '')
|
|
|
|
|
payload = b'<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
for encoding in filter(None, (e.strip() for e in encodings.split(','))):
|
|
|
|
|
if encoding == 'br' and brotli:
|
|
|
|
|
payload = brotli.compress(payload)
|
|
|
|
|
elif encoding == 'gzip':
|
|
|
|
|
buf = io.BytesIO()
|
|
|
|
|
with gzip.GzipFile(fileobj=buf, mode='wb') as f:
|
|
|
|
|
f.write(payload)
|
|
|
|
|
payload = buf.getvalue()
|
|
|
|
|
elif encoding == 'deflate':
|
|
|
|
|
payload = zlib.compress(payload)
|
|
|
|
|
elif encoding == 'unsupported':
|
|
|
|
|
payload = b'raw'
|
|
|
|
|
break
|
|
|
|
|
else:
|
|
|
|
|
self._status(415)
|
|
|
|
|
return
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Encoding', encodings)
|
|
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
2023-07-15 10:25:23 +00:00
|
|
|
elif self.path.startswith('/gen_'):
|
|
|
|
|
payload = b'<html></html>'
|
|
|
|
|
self.send_response(int(self.path[len('/gen_'):]))
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
|
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
elif self.path.startswith('/incompleteread'):
|
|
|
|
|
payload = b'<html></html>'
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
|
|
|
|
self.send_header('Content-Length', '234234')
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
self.finish()
|
|
|
|
|
elif self.path.startswith('/timeout_'):
|
|
|
|
|
time.sleep(int(self.path[len('/timeout_'):]))
|
|
|
|
|
self._headers()
|
|
|
|
|
elif self.path == '/source_address':
|
|
|
|
|
payload = str(self.client_address[0]).encode()
|
|
|
|
|
self.send_response(200)
|
|
|
|
|
self.send_header('Content-Type', 'text/html; charset=utf-8')
|
|
|
|
|
self.send_header('Content-Length', str(len(payload)))
|
|
|
|
|
self.end_headers()
|
|
|
|
|
self.wfile.write(payload)
|
|
|
|
|
self.finish()
|
2015-01-30 01:57:37 +00:00
|
|
|
else:
|
2023-05-27 07:06:13 +00:00
|
|
|
self._status(404)
|
|
|
|
|
|
|
|
|
|
def send_header(self, keyword, value):
|
|
|
|
|
"""
|
|
|
|
|
Forcibly allow HTTP server to send non percent-encoded non-ASCII characters in headers.
|
|
|
|
|
This is against what is defined in RFC 3986, however we need to test we support this
|
|
|
|
|
since some sites incorrectly do this.
|
|
|
|
|
"""
|
|
|
|
|
if keyword.lower() == 'connection':
|
|
|
|
|
return super().send_header(keyword, value)
|
|
|
|
|
|
|
|
|
|
if not hasattr(self, '_headers_buffer'):
|
|
|
|
|
self._headers_buffer = []
|
|
|
|
|
|
|
|
|
|
self._headers_buffer.append(f'{keyword}: {value}\r\n'.encode())
|
2015-01-30 01:57:37 +00:00
|
|
|
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
class TestRequestHandlerBase:
|
|
|
|
|
@classmethod
|
|
|
|
|
def setup_class(cls):
|
|
|
|
|
cls.http_httpd = http.server.ThreadingHTTPServer(
|
2018-02-03 16:09:25 +00:00
|
|
|
('127.0.0.1', 0), HTTPTestRequestHandler)
|
2023-07-15 10:25:23 +00:00
|
|
|
cls.http_port = http_server_port(cls.http_httpd)
|
|
|
|
|
cls.http_server_thread = threading.Thread(target=cls.http_httpd.serve_forever)
|
2023-05-27 07:06:13 +00:00
|
|
|
# FIXME: we should probably stop the http server thread after each test
|
|
|
|
|
# See: https://github.com/yt-dlp/yt-dlp/pull/7094#discussion_r1199746041
|
2023-07-15 10:25:23 +00:00
|
|
|
cls.http_server_thread.daemon = True
|
|
|
|
|
cls.http_server_thread.start()
|
2023-05-27 07:06:13 +00:00
|
|
|
|
|
|
|
|
# HTTPS server
|
2015-01-30 01:57:37 +00:00
|
|
|
certfn = os.path.join(TEST_DIR, 'testcert.pem')
|
2023-07-15 10:25:23 +00:00
|
|
|
cls.https_httpd = http.server.ThreadingHTTPServer(
|
2018-02-03 16:09:25 +00:00
|
|
|
('127.0.0.1', 0), HTTPTestRequestHandler)
|
2022-04-17 20:42:48 +00:00
|
|
|
sslctx = ssl.SSLContext(ssl.PROTOCOL_TLS_SERVER)
|
|
|
|
|
sslctx.load_cert_chain(certfn, None)
|
2023-07-15 10:25:23 +00:00
|
|
|
cls.https_httpd.socket = sslctx.wrap_socket(cls.https_httpd.socket, server_side=True)
|
|
|
|
|
cls.https_port = http_server_port(cls.https_httpd)
|
|
|
|
|
cls.https_server_thread = threading.Thread(target=cls.https_httpd.serve_forever)
|
|
|
|
|
cls.https_server_thread.daemon = True
|
|
|
|
|
cls.https_server_thread.start()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestHTTPRequestHandler(TestRequestHandlerBase):
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_verify_cert(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
with pytest.raises(CertificateVerifyError):
|
|
|
|
|
validate_and_send(rh, Request(f'https://127.0.0.1:{self.https_port}/headers'))
|
|
|
|
|
|
|
|
|
|
with handler(verify=False) as rh:
|
|
|
|
|
r = validate_and_send(rh, Request(f'https://127.0.0.1:{self.https_port}/headers'))
|
|
|
|
|
assert r.status == 200
|
2023-05-27 07:06:13 +00:00
|
|
|
r.close()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_ssl_error(self, handler):
|
|
|
|
|
# HTTPS server with too old TLS version
|
|
|
|
|
# XXX: is there a better way to test this than to create a new server?
|
|
|
|
|
https_httpd = http.server.ThreadingHTTPServer(
|
|
|
|
|
('127.0.0.1', 0), HTTPTestRequestHandler)
|
|
|
|
|
sslctx = ssl.SSLContext(ssl.PROTOCOL_TLS_SERVER)
|
|
|
|
|
https_httpd.socket = sslctx.wrap_socket(https_httpd.socket, server_side=True)
|
|
|
|
|
https_port = http_server_port(https_httpd)
|
|
|
|
|
https_server_thread = threading.Thread(target=https_httpd.serve_forever)
|
|
|
|
|
https_server_thread.daemon = True
|
|
|
|
|
https_server_thread.start()
|
|
|
|
|
|
|
|
|
|
with handler(verify=False) as rh:
|
2024-03-17 03:52:38 +00:00
|
|
|
with pytest.raises(SSLError, match=r'(?i)ssl(?:v3|/tls).alert.handshake.failure') as exc_info:
|
2023-07-15 10:25:23 +00:00
|
|
|
validate_and_send(rh, Request(f'https://127.0.0.1:{https_port}/headers'))
|
|
|
|
|
assert not issubclass(exc_info.type, CertificateVerifyError)
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_percent_encode(self, handler):
|
|
|
|
|
with handler() as rh:
|
2023-05-27 07:06:13 +00:00
|
|
|
# Unicode characters should be encoded with uppercase percent-encoding
|
2023-07-15 10:25:23 +00:00
|
|
|
res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/中文.html'))
|
|
|
|
|
assert res.status == 200
|
2023-05-27 07:06:13 +00:00
|
|
|
res.close()
|
|
|
|
|
# don't normalize existing percent encodings
|
2023-07-15 10:25:23 +00:00
|
|
|
res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/%c7%9f'))
|
|
|
|
|
assert res.status == 200
|
2023-05-27 07:06:13 +00:00
|
|
|
res.close()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2024-01-20 21:03:33 +00:00
|
|
|
@pytest.mark.parametrize('path', [
|
|
|
|
|
'/a/b/./../../headers',
|
|
|
|
|
'/redirect_dotsegments',
|
|
|
|
|
# https://github.com/yt-dlp/yt-dlp/issues/9020
|
|
|
|
|
'/redirect_dotsegments_absolute',
|
|
|
|
|
])
|
|
|
|
|
def test_remove_dot_segments(self, handler, path):
|
|
|
|
|
with handler(verbose=True) as rh:
|
2023-07-28 22:40:20 +00:00
|
|
|
# This isn't a comprehensive test,
|
2024-01-20 21:03:33 +00:00
|
|
|
# but it should be enough to check whether the handler is removing dot segments in required scenarios
|
|
|
|
|
res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}{path}'))
|
2023-07-28 22:40:20 +00:00
|
|
|
assert res.status == 200
|
|
|
|
|
assert res.url == f'http://127.0.0.1:{self.http_port}/headers'
|
|
|
|
|
res.close()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
# Not supported by CurlCFFI (non-standard)
|
2023-10-13 23:33:00 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_unicode_path_redirection(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
r = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/302-non-ascii-redirect'))
|
|
|
|
|
assert r.url == f'http://127.0.0.1:{self.http_port}/%E4%B8%AD%E6%96%87.html'
|
2023-05-27 07:06:13 +00:00
|
|
|
r.close()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_raise_http_error(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
for bad_status in (400, 500, 599, 302):
|
|
|
|
|
with pytest.raises(HTTPError):
|
|
|
|
|
validate_and_send(rh, Request('http://127.0.0.1:%d/gen_%d' % (self.http_port, bad_status)))
|
|
|
|
|
|
|
|
|
|
# Should not raise an error
|
|
|
|
|
validate_and_send(rh, Request('http://127.0.0.1:%d/gen_200' % self.http_port)).close()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_response_url(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
# Response url should be that of the last url in redirect chain
|
|
|
|
|
res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_301'))
|
|
|
|
|
assert res.url == f'http://127.0.0.1:{self.http_port}/method'
|
|
|
|
|
res.close()
|
|
|
|
|
res2 = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/gen_200'))
|
|
|
|
|
assert res2.url == f'http://127.0.0.1:{self.http_port}/gen_200'
|
|
|
|
|
res2.close()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
# Covers some basic cases we expect some level of consistency between request handlers for
|
|
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
|
|
|
|
@pytest.mark.parametrize('redirect_status,method,expected', [
|
|
|
|
|
# A 303 must either use GET or HEAD for subsequent request
|
|
|
|
|
(303, 'POST', ('', 'GET', False)),
|
|
|
|
|
(303, 'HEAD', ('', 'HEAD', False)),
|
|
|
|
|
|
|
|
|
|
# 301 and 302 turn POST only into a GET
|
|
|
|
|
(301, 'POST', ('', 'GET', False)),
|
|
|
|
|
(301, 'HEAD', ('', 'HEAD', False)),
|
|
|
|
|
(302, 'POST', ('', 'GET', False)),
|
|
|
|
|
(302, 'HEAD', ('', 'HEAD', False)),
|
|
|
|
|
|
|
|
|
|
# 307 and 308 should not change method
|
|
|
|
|
(307, 'POST', ('testdata', 'POST', True)),
|
|
|
|
|
(308, 'POST', ('testdata', 'POST', True)),
|
|
|
|
|
(307, 'HEAD', ('', 'HEAD', False)),
|
|
|
|
|
(308, 'HEAD', ('', 'HEAD', False)),
|
|
|
|
|
])
|
|
|
|
|
def test_redirect(self, handler, redirect_status, method, expected):
|
2023-07-15 10:25:23 +00:00
|
|
|
with handler() as rh:
|
2024-03-17 03:52:38 +00:00
|
|
|
data = b'testdata' if method == 'POST' else None
|
|
|
|
|
headers = {}
|
|
|
|
|
if data is not None:
|
|
|
|
|
headers['Content-Type'] = 'application/test'
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_{redirect_status}', method=method, data=data,
|
|
|
|
|
headers=headers))
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
headers = b''
|
|
|
|
|
data_recv = b''
|
|
|
|
|
if data is not None:
|
|
|
|
|
data_recv += res.read(len(data))
|
|
|
|
|
if data_recv != data:
|
|
|
|
|
headers += data_recv
|
|
|
|
|
data_recv = b''
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
headers += res.read()
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
assert expected[0] == data_recv.decode()
|
|
|
|
|
assert expected[1] == res.headers.get('method')
|
|
|
|
|
assert expected[2] == ('content-length' in headers.decode().lower())
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_request_cookie_header(self, handler):
|
2023-06-06 08:44:51 +00:00
|
|
|
# We should accept a Cookie header being passed as in normal headers and handle it appropriately.
|
2023-07-15 10:25:23 +00:00
|
|
|
with handler() as rh:
|
2023-06-06 08:44:51 +00:00
|
|
|
# Specified Cookie header should be used
|
2023-07-15 10:25:23 +00:00
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
2023-05-27 07:06:13 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/headers',
|
2023-07-15 10:25:23 +00:00
|
|
|
headers={'Cookie': 'test=test'})).read().decode()
|
2024-03-17 03:52:38 +00:00
|
|
|
assert 'cookie: test=test' in res.lower()
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
# Specified Cookie header should be removed on any redirect
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
|
|
|
|
f'http://127.0.0.1:{self.http_port}/308-to-headers',
|
2024-03-17 03:52:38 +00:00
|
|
|
headers={'Cookie': 'test=test2'})).read().decode()
|
|
|
|
|
assert 'cookie: test=test2' not in res.lower()
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
# Specified Cookie header should override global cookiejar for that request
|
2024-03-17 03:52:38 +00:00
|
|
|
# Whether cookies from the cookiejar is applied on the redirect is considered undefined for now
|
2023-07-29 22:06:17 +00:00
|
|
|
cookiejar = YoutubeDLCookieJar()
|
2023-07-15 10:25:23 +00:00
|
|
|
cookiejar.set_cookie(http.cookiejar.Cookie(
|
|
|
|
|
version=0, name='test', value='ytdlp', port=None, port_specified=False,
|
|
|
|
|
domain='127.0.0.1', domain_specified=True, domain_initial_dot=False, path='/',
|
|
|
|
|
path_specified=True, secure=False, expires=None, discard=False, comment=None,
|
|
|
|
|
comment_url=None, rest={}))
|
|
|
|
|
|
|
|
|
|
with handler(cookiejar=cookiejar) as rh:
|
|
|
|
|
data = validate_and_send(
|
2024-03-17 03:52:38 +00:00
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'cookie': 'test=test3'})).read()
|
|
|
|
|
assert b'cookie: test=ytdlp' not in data.lower()
|
|
|
|
|
assert b'cookie: test=test3' in data.lower()
|
2023-07-15 10:25:23 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_redirect_loop(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
with pytest.raises(HTTPError, match='redirect loop'):
|
|
|
|
|
validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_loop'))
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_incompleteread(self, handler):
|
|
|
|
|
with handler(timeout=2) as rh:
|
2024-03-17 03:52:38 +00:00
|
|
|
with pytest.raises(IncompleteRead, match='13 bytes read, 234221 more expected'):
|
2023-07-15 10:25:23 +00:00
|
|
|
validate_and_send(rh, Request('http://127.0.0.1:%d/incompleteread' % self.http_port)).read()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_cookies(self, handler):
|
2023-07-29 22:06:17 +00:00
|
|
|
cookiejar = YoutubeDLCookieJar()
|
2023-07-15 10:25:23 +00:00
|
|
|
cookiejar.set_cookie(http.cookiejar.Cookie(
|
|
|
|
|
0, 'test', 'ytdlp', None, False, '127.0.0.1', True,
|
|
|
|
|
False, '/headers', True, False, None, False, None, None, {}))
|
|
|
|
|
|
|
|
|
|
with handler(cookiejar=cookiejar) as rh:
|
|
|
|
|
data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read()
|
2024-03-17 03:52:38 +00:00
|
|
|
assert b'cookie: test=ytdlp' in data.lower()
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
# Per request
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
data = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions={'cookiejar': cookiejar})).read()
|
2024-03-17 03:52:38 +00:00
|
|
|
assert b'cookie: test=ytdlp' in data.lower()
|
2023-07-15 10:25:23 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_headers(self, handler):
|
|
|
|
|
|
|
|
|
|
with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh:
|
|
|
|
|
# Global Headers
|
2024-03-17 03:52:38 +00:00
|
|
|
data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read().lower()
|
|
|
|
|
assert b'test1: test' in data
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
# Per request headers, merged with global
|
|
|
|
|
data = validate_and_send(rh, Request(
|
2024-03-17 03:52:38 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/headers', headers={'test2': 'changed', 'test3': 'test3'})).read().lower()
|
|
|
|
|
assert b'test1: test' in data
|
|
|
|
|
assert b'test2: changed' in data
|
|
|
|
|
assert b'test2: test2' not in data
|
|
|
|
|
assert b'test3: test3' in data
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
|
|
|
|
def test_read_timeout(self, handler):
|
2023-07-15 10:25:23 +00:00
|
|
|
with handler() as rh:
|
|
|
|
|
# Default timeout is 20 seconds, so this should go through
|
|
|
|
|
validate_and_send(
|
2024-03-17 03:52:38 +00:00
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1'))
|
2023-07-15 10:25:23 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
with handler(timeout=0.1) as rh:
|
2023-07-15 10:25:23 +00:00
|
|
|
with pytest.raises(TransportError):
|
|
|
|
|
validate_and_send(
|
2024-03-17 03:52:38 +00:00
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_5'))
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
# Per request timeout, should override handler timeout
|
|
|
|
|
validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1', extensions={'timeout': 4}))
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
|
|
|
|
def test_connect_timeout(self, handler):
|
|
|
|
|
# nothing should be listening on this port
|
|
|
|
|
connect_timeout_url = 'http://10.255.255.255'
|
|
|
|
|
with handler(timeout=0.01) as rh:
|
|
|
|
|
now = time.time()
|
|
|
|
|
with pytest.raises(TransportError):
|
|
|
|
|
validate_and_send(
|
|
|
|
|
rh, Request(connect_timeout_url))
|
|
|
|
|
assert 0.01 <= time.time() - now < 20
|
|
|
|
|
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
with pytest.raises(TransportError):
|
|
|
|
|
# Per request timeout, should override handler timeout
|
|
|
|
|
now = time.time()
|
|
|
|
|
validate_and_send(
|
|
|
|
|
rh, Request(connect_timeout_url, extensions={'timeout': 0.01}))
|
|
|
|
|
assert 0.01 <= time.time() - now < 20
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_source_address(self, handler):
|
|
|
|
|
source_address = f'127.0.0.{random.randint(5, 255)}'
|
2024-01-19 21:39:49 +00:00
|
|
|
# on some systems these loopback addresses we need for testing may not be available
|
|
|
|
|
# see: https://github.com/yt-dlp/yt-dlp/issues/8890
|
|
|
|
|
verify_address_availability(source_address)
|
2023-07-15 10:25:23 +00:00
|
|
|
with handler(source_address=source_address) as rh:
|
|
|
|
|
data = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/source_address')).read().decode()
|
|
|
|
|
assert source_address == data
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
# Not supported by CurlCFFI
|
2023-10-13 23:33:00 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_gzip_trailing_garbage(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
data = validate_and_send(rh, Request(f'http://localhost:{self.http_port}/trailing_garbage')).read().decode()
|
|
|
|
|
assert data == '<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
|
2023-10-13 23:33:00 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
@pytest.mark.skipif(not brotli, reason='brotli support is not installed')
|
|
|
|
|
def test_brotli(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
2023-05-27 10:40:05 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/content-encoding',
|
|
|
|
|
headers={'ytdl-encoding': 'br'}))
|
2023-07-15 10:25:23 +00:00
|
|
|
assert res.headers.get('Content-Encoding') == 'br'
|
|
|
|
|
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_deflate(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
2023-05-27 10:40:05 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/content-encoding',
|
|
|
|
|
headers={'ytdl-encoding': 'deflate'}))
|
2023-07-15 10:25:23 +00:00
|
|
|
assert res.headers.get('Content-Encoding') == 'deflate'
|
|
|
|
|
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_gzip(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
2023-05-27 10:40:05 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/content-encoding',
|
|
|
|
|
headers={'ytdl-encoding': 'gzip'}))
|
2023-07-15 10:25:23 +00:00
|
|
|
assert res.headers.get('Content-Encoding') == 'gzip'
|
|
|
|
|
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
|
2023-05-27 10:40:05 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_multiple_encodings(self, handler):
|
|
|
|
|
with handler() as rh:
|
2023-05-27 10:40:05 +00:00
|
|
|
for pair in ('gzip,deflate', 'deflate, gzip', 'gzip, gzip', 'deflate, deflate'):
|
2023-07-15 10:25:23 +00:00
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
2023-05-27 10:40:05 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/content-encoding',
|
|
|
|
|
headers={'ytdl-encoding': pair}))
|
2023-07-15 10:25:23 +00:00
|
|
|
assert res.headers.get('Content-Encoding') == pair
|
|
|
|
|
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
# Not supported by curl_cffi
|
2023-10-13 23:33:00 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_unsupported_encoding(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(
|
2023-05-27 10:40:05 +00:00
|
|
|
f'http://127.0.0.1:{self.http_port}/content-encoding',
|
2024-03-17 03:52:38 +00:00
|
|
|
headers={'ytdl-encoding': 'unsupported', 'Accept-Encoding': '*'}))
|
2023-07-15 10:25:23 +00:00
|
|
|
assert res.headers.get('Content-Encoding') == 'unsupported'
|
|
|
|
|
assert res.read() == b'raw'
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_read(self, handler):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers'))
|
|
|
|
|
assert res.readable()
|
|
|
|
|
assert res.read(1) == b'H'
|
|
|
|
|
assert res.read(3) == b'ost'
|
2024-03-17 03:52:38 +00:00
|
|
|
assert res.read().decode().endswith('\n\n')
|
|
|
|
|
assert res.read() == b''
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestHTTPProxy(TestRequestHandlerBase):
|
2024-03-17 03:52:38 +00:00
|
|
|
# Note: this only tests http urls over non-CONNECT proxy
|
2023-07-15 10:25:23 +00:00
|
|
|
@classmethod
|
|
|
|
|
def setup_class(cls):
|
|
|
|
|
super().setup_class()
|
|
|
|
|
# HTTP Proxy server
|
|
|
|
|
cls.proxy = http.server.ThreadingHTTPServer(
|
|
|
|
|
('127.0.0.1', 0), _build_proxy_handler('normal'))
|
|
|
|
|
cls.proxy_port = http_server_port(cls.proxy)
|
|
|
|
|
cls.proxy_thread = threading.Thread(target=cls.proxy.serve_forever)
|
|
|
|
|
cls.proxy_thread.daemon = True
|
|
|
|
|
cls.proxy_thread.start()
|
2023-05-27 10:40:05 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
# Geo proxy server
|
|
|
|
|
cls.geo_proxy = http.server.ThreadingHTTPServer(
|
|
|
|
|
('127.0.0.1', 0), _build_proxy_handler('geo'))
|
|
|
|
|
cls.geo_port = http_server_port(cls.geo_proxy)
|
|
|
|
|
cls.geo_proxy_thread = threading.Thread(target=cls.geo_proxy.serve_forever)
|
|
|
|
|
cls.geo_proxy_thread.daemon = True
|
|
|
|
|
cls.geo_proxy_thread.start()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_http_proxy(self, handler):
|
|
|
|
|
http_proxy = f'http://127.0.0.1:{self.proxy_port}'
|
|
|
|
|
geo_proxy = f'http://127.0.0.1:{self.geo_port}'
|
|
|
|
|
|
|
|
|
|
# Test global http proxy
|
|
|
|
|
# Test per request http proxy
|
|
|
|
|
# Test per request http proxy disables proxy
|
|
|
|
|
url = 'http://foo.com/bar'
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
# Global HTTP proxy
|
|
|
|
|
with handler(proxies={'http': http_proxy}) as rh:
|
|
|
|
|
res = validate_and_send(rh, Request(url)).read().decode()
|
|
|
|
|
assert res == f'normal: {url}'
|
|
|
|
|
|
|
|
|
|
# Per request proxy overrides global
|
|
|
|
|
res = validate_and_send(rh, Request(url, proxies={'http': geo_proxy})).read().decode()
|
|
|
|
|
assert res == f'geo: {url}'
|
|
|
|
|
|
|
|
|
|
# and setting to None disables all proxies for that request
|
|
|
|
|
real_url = f'http://127.0.0.1:{self.http_port}/headers'
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(real_url, proxies={'http': None})).read().decode()
|
|
|
|
|
assert res != f'normal: {real_url}'
|
|
|
|
|
assert 'Accept' in res
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_noproxy(self, handler):
|
|
|
|
|
with handler(proxies={'proxy': f'http://127.0.0.1:{self.proxy_port}'}) as rh:
|
|
|
|
|
# NO_PROXY
|
|
|
|
|
for no_proxy in (f'127.0.0.1:{self.http_port}', '127.0.0.1', 'localhost'):
|
|
|
|
|
nop_response = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', proxies={'no': no_proxy})).read().decode(
|
|
|
|
|
'utf-8')
|
|
|
|
|
assert 'Accept' in nop_response
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_allproxy(self, handler):
|
|
|
|
|
url = 'http://foo.com/bar'
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
response = validate_and_send(rh, Request(url, proxies={'all': f'http://127.0.0.1:{self.proxy_port}'})).read().decode(
|
|
|
|
|
'utf-8')
|
|
|
|
|
assert response == f'normal: {url}'
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_http_proxy_with_idn(self, handler):
|
|
|
|
|
with handler(proxies={
|
|
|
|
|
'http': f'http://127.0.0.1:{self.proxy_port}',
|
|
|
|
|
}) as rh:
|
|
|
|
|
url = 'http://中文.tw/'
|
|
|
|
|
response = rh.send(Request(url)).read().decode()
|
|
|
|
|
# b'xn--fiq228c' is '中文'.encode('idna')
|
|
|
|
|
assert response == 'normal: http://xn--fiq228c.tw/'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestClientCertificate:
|
|
|
|
|
@classmethod
|
|
|
|
|
def setup_class(cls):
|
2022-05-02 07:59:45 +00:00
|
|
|
certfn = os.path.join(TEST_DIR, 'testcert.pem')
|
2023-07-15 10:25:23 +00:00
|
|
|
cls.certdir = os.path.join(TEST_DIR, 'testdata', 'certificate')
|
|
|
|
|
cacertfn = os.path.join(cls.certdir, 'ca.crt')
|
|
|
|
|
cls.httpd = http.server.ThreadingHTTPServer(('127.0.0.1', 0), HTTPTestRequestHandler)
|
2022-05-02 07:59:45 +00:00
|
|
|
sslctx = ssl.SSLContext(ssl.PROTOCOL_TLS_SERVER)
|
|
|
|
|
sslctx.verify_mode = ssl.CERT_REQUIRED
|
|
|
|
|
sslctx.load_verify_locations(cafile=cacertfn)
|
|
|
|
|
sslctx.load_cert_chain(certfn, None)
|
2023-07-15 10:25:23 +00:00
|
|
|
cls.httpd.socket = sslctx.wrap_socket(cls.httpd.socket, server_side=True)
|
|
|
|
|
cls.port = http_server_port(cls.httpd)
|
|
|
|
|
cls.server_thread = threading.Thread(target=cls.httpd.serve_forever)
|
|
|
|
|
cls.server_thread.daemon = True
|
|
|
|
|
cls.server_thread.start()
|
|
|
|
|
|
|
|
|
|
def _run_test(self, handler, **handler_kwargs):
|
|
|
|
|
with handler(
|
2022-05-02 07:59:45 +00:00
|
|
|
# Disable client-side validation of unacceptable self-signed testcert.pem
|
|
|
|
|
# The test is of a check on the server side, so unaffected
|
2023-07-15 10:25:23 +00:00
|
|
|
verify=False,
|
|
|
|
|
**handler_kwargs,
|
|
|
|
|
) as rh:
|
|
|
|
|
validate_and_send(rh, Request(f'https://127.0.0.1:{self.port}/video.html')).read().decode()
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_certificate_combined_nopass(self, handler):
|
|
|
|
|
self._run_test(handler, client_cert={
|
|
|
|
|
'client_certificate': os.path.join(self.certdir, 'clientwithkey.crt'),
|
2022-05-02 07:59:45 +00:00
|
|
|
})
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_certificate_nocombined_nopass(self, handler):
|
|
|
|
|
self._run_test(handler, client_cert={
|
|
|
|
|
'client_certificate': os.path.join(self.certdir, 'client.crt'),
|
|
|
|
|
'client_certificate_key': os.path.join(self.certdir, 'client.key'),
|
|
|
|
|
})
|
2022-05-02 07:59:45 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_certificate_combined_pass(self, handler):
|
|
|
|
|
self._run_test(handler, client_cert={
|
|
|
|
|
'client_certificate': os.path.join(self.certdir, 'clientwithencryptedkey.crt'),
|
|
|
|
|
'client_certificate_password': 'foobar',
|
|
|
|
|
})
|
2022-05-02 07:59:45 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_certificate_nocombined_pass(self, handler):
|
|
|
|
|
self._run_test(handler, client_cert={
|
|
|
|
|
'client_certificate': os.path.join(self.certdir, 'client.crt'),
|
|
|
|
|
'client_certificate_key': os.path.join(self.certdir, 'clientencrypted.key'),
|
|
|
|
|
'client_certificate_password': 'foobar',
|
|
|
|
|
})
|
2022-05-02 07:59:45 +00:00
|
|
|
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True)
|
|
|
|
|
class TestHTTPImpersonateRequestHandler(TestRequestHandlerBase):
|
|
|
|
|
def test_supported_impersonate_targets(self, handler):
|
|
|
|
|
with handler(headers=std_headers) as rh:
|
|
|
|
|
# note: this assumes the impersonate request handler supports the impersonate extension
|
|
|
|
|
for target in rh.supported_targets:
|
|
|
|
|
res = validate_and_send(rh, Request(
|
|
|
|
|
f'http://127.0.0.1:{self.http_port}/headers', extensions={'impersonate': target}))
|
|
|
|
|
assert res.status == 200
|
|
|
|
|
assert std_headers['user-agent'].lower() not in res.read().decode().lower()
|
|
|
|
|
|
|
|
|
|
|
2024-02-17 22:32:34 +00:00
|
|
|
class TestRequestHandlerMisc:
|
|
|
|
|
"""Misc generic tests for request handlers, not related to request or validation testing"""
|
|
|
|
|
@pytest.mark.parametrize('handler,logger_name', [
|
|
|
|
|
('Requests', 'urllib3'),
|
|
|
|
|
('Websockets', 'websockets.client'),
|
|
|
|
|
('Websockets', 'websockets.server')
|
|
|
|
|
], indirect=['handler'])
|
|
|
|
|
def test_remove_logging_handler(self, handler, logger_name):
|
|
|
|
|
# Ensure any logging handlers, which may contain a YoutubeDL instance,
|
|
|
|
|
# are removed when we close the request handler
|
|
|
|
|
# See: https://github.com/yt-dlp/yt-dlp/issues/8922
|
|
|
|
|
logging_handlers = logging.getLogger(logger_name).handlers
|
|
|
|
|
before_count = len(logging_handlers)
|
|
|
|
|
rh = handler()
|
|
|
|
|
assert len(logging_handlers) == before_count + 1
|
|
|
|
|
rh.close()
|
|
|
|
|
assert len(logging_handlers) == before_count
|
|
|
|
|
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
class TestUrllibRequestHandler(TestRequestHandlerBase):
|
|
|
|
|
@pytest.mark.parametrize('handler', ['Urllib'], indirect=True)
|
|
|
|
|
def test_file_urls(self, handler):
|
|
|
|
|
# See https://github.com/ytdl-org/youtube-dl/issues/8227
|
|
|
|
|
tf = tempfile.NamedTemporaryFile(delete=False)
|
|
|
|
|
tf.write(b'foobar')
|
|
|
|
|
tf.close()
|
|
|
|
|
req = Request(pathlib.Path(tf.name).as_uri())
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
with pytest.raises(UnsupportedRequest):
|
|
|
|
|
rh.validate(req)
|
|
|
|
|
|
|
|
|
|
# Test that urllib never loaded FileHandler
|
|
|
|
|
with pytest.raises(TransportError):
|
|
|
|
|
rh.send(req)
|
|
|
|
|
|
|
|
|
|
with handler(enable_file_urls=True) as rh:
|
|
|
|
|
res = validate_and_send(rh, req)
|
|
|
|
|
assert res.read() == b'foobar'
|
|
|
|
|
res.close()
|
2022-05-02 07:59:45 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
os.unlink(tf.name)
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib'], indirect=True)
|
|
|
|
|
def test_http_error_returns_content(self, handler):
|
|
|
|
|
# urllib HTTPError will try close the underlying response if reference to the HTTPError object is lost
|
|
|
|
|
def get_response():
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
# headers url
|
|
|
|
|
try:
|
|
|
|
|
validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/gen_404'))
|
|
|
|
|
except HTTPError as e:
|
|
|
|
|
return e.response
|
|
|
|
|
|
|
|
|
|
assert get_response().read() == b'<html></html>'
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('handler', ['Urllib'], indirect=True)
|
|
|
|
|
def test_verify_cert_error_text(self, handler):
|
|
|
|
|
# Check the output of the error message
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
with pytest.raises(
|
|
|
|
|
CertificateVerifyError,
|
|
|
|
|
match=r'\[SSL: CERTIFICATE_VERIFY_FAILED\] certificate verify failed: self.signed certificate'
|
|
|
|
|
):
|
|
|
|
|
validate_and_send(rh, Request(f'https://127.0.0.1:{self.https_port}/headers'))
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('handler', ['Urllib'], indirect=True)
|
2023-07-24 19:18:52 +00:00
|
|
|
@pytest.mark.parametrize('req,match,version_check', [
|
|
|
|
|
# https://github.com/python/cpython/blob/987b712b4aeeece336eed24fcc87a950a756c3e2/Lib/http/client.py#L1256
|
|
|
|
|
# bpo-39603: Check implemented in 3.7.9+, 3.8.5+
|
|
|
|
|
(
|
|
|
|
|
Request('http://127.0.0.1', method='GET\n'),
|
|
|
|
|
'method can\'t contain control characters',
|
|
|
|
|
lambda v: v < (3, 7, 9) or (3, 8, 0) <= v < (3, 8, 5)
|
|
|
|
|
),
|
|
|
|
|
# https://github.com/python/cpython/blob/987b712b4aeeece336eed24fcc87a950a756c3e2/Lib/http/client.py#L1265
|
|
|
|
|
# bpo-38576: Check implemented in 3.7.8+, 3.8.3+
|
|
|
|
|
(
|
|
|
|
|
Request('http://127.0.0. 1', method='GET'),
|
|
|
|
|
'URL can\'t contain control characters',
|
|
|
|
|
lambda v: v < (3, 7, 8) or (3, 8, 0) <= v < (3, 8, 3)
|
|
|
|
|
),
|
|
|
|
|
# https://github.com/python/cpython/blob/987b712b4aeeece336eed24fcc87a950a756c3e2/Lib/http/client.py#L1288C31-L1288C50
|
|
|
|
|
(Request('http://127.0.0.1', headers={'foo\n': 'bar'}), 'Invalid header name', None),
|
|
|
|
|
])
|
|
|
|
|
def test_httplib_validation_errors(self, handler, req, match, version_check):
|
|
|
|
|
if version_check and version_check(sys.version_info):
|
|
|
|
|
pytest.skip(f'Python {sys.version} version does not have the required validation for this test.')
|
2023-07-15 10:25:23 +00:00
|
|
|
|
2023-07-24 19:18:52 +00:00
|
|
|
with handler() as rh:
|
|
|
|
|
with pytest.raises(RequestError, match=match) as exc_info:
|
|
|
|
|
validate_and_send(rh, req)
|
2023-07-15 10:25:23 +00:00
|
|
|
assert not isinstance(exc_info.value, TransportError)
|
|
|
|
|
|
|
|
|
|
|
2024-02-17 22:32:34 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Requests'], indirect=True)
|
2023-10-13 23:33:00 +00:00
|
|
|
class TestRequestsRequestHandler(TestRequestHandlerBase):
|
|
|
|
|
@pytest.mark.parametrize('raised,expected', [
|
|
|
|
|
(lambda: requests.exceptions.ConnectTimeout(), TransportError),
|
|
|
|
|
(lambda: requests.exceptions.ReadTimeout(), TransportError),
|
|
|
|
|
(lambda: requests.exceptions.Timeout(), TransportError),
|
|
|
|
|
(lambda: requests.exceptions.ConnectionError(), TransportError),
|
|
|
|
|
(lambda: requests.exceptions.ProxyError(), ProxyError),
|
|
|
|
|
(lambda: requests.exceptions.SSLError('12[CERTIFICATE_VERIFY_FAILED]34'), CertificateVerifyError),
|
|
|
|
|
(lambda: requests.exceptions.SSLError(), SSLError),
|
|
|
|
|
(lambda: requests.exceptions.InvalidURL(), RequestError),
|
|
|
|
|
(lambda: requests.exceptions.InvalidHeader(), RequestError),
|
|
|
|
|
# catch-all: https://github.com/psf/requests/blob/main/src/requests/adapters.py#L535
|
|
|
|
|
(lambda: urllib3.exceptions.HTTPError(), TransportError),
|
|
|
|
|
(lambda: requests.exceptions.RequestException(), RequestError)
|
|
|
|
|
# (lambda: requests.exceptions.TooManyRedirects(), HTTPError) - Needs a response object
|
|
|
|
|
])
|
|
|
|
|
def test_request_error_mapping(self, handler, monkeypatch, raised, expected):
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
def mock_get_instance(*args, **kwargs):
|
|
|
|
|
class MockSession:
|
|
|
|
|
def request(self, *args, **kwargs):
|
|
|
|
|
raise raised()
|
|
|
|
|
return MockSession()
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(rh, '_get_instance', mock_get_instance)
|
|
|
|
|
|
|
|
|
|
with pytest.raises(expected) as exc_info:
|
|
|
|
|
rh.send(Request('http://fake'))
|
|
|
|
|
|
|
|
|
|
assert exc_info.type is expected
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('raised,expected,match', [
|
|
|
|
|
(lambda: urllib3.exceptions.SSLError(), SSLError, None),
|
|
|
|
|
(lambda: urllib3.exceptions.TimeoutError(), TransportError, None),
|
|
|
|
|
(lambda: urllib3.exceptions.ReadTimeoutError(None, None, None), TransportError, None),
|
|
|
|
|
(lambda: urllib3.exceptions.ProtocolError(), TransportError, None),
|
|
|
|
|
(lambda: urllib3.exceptions.DecodeError(), TransportError, None),
|
|
|
|
|
(lambda: urllib3.exceptions.HTTPError(), TransportError, None), # catch-all
|
|
|
|
|
(
|
|
|
|
|
lambda: urllib3.exceptions.ProtocolError('error', http.client.IncompleteRead(partial=b'abc', expected=4)),
|
|
|
|
|
IncompleteRead,
|
|
|
|
|
'3 bytes read, 4 more expected'
|
|
|
|
|
),
|
|
|
|
|
(
|
2023-10-15 08:54:38 +00:00
|
|
|
lambda: urllib3.exceptions.ProtocolError('error', urllib3.exceptions.IncompleteRead(partial=3, expected=5)),
|
2023-10-13 23:33:00 +00:00
|
|
|
IncompleteRead,
|
|
|
|
|
'3 bytes read, 5 more expected'
|
|
|
|
|
),
|
|
|
|
|
])
|
|
|
|
|
def test_response_error_mapping(self, handler, monkeypatch, raised, expected, match):
|
|
|
|
|
from requests.models import Response as RequestsResponse
|
2023-11-20 08:04:04 +00:00
|
|
|
from urllib3.response import HTTPResponse as Urllib3Response
|
|
|
|
|
|
2023-10-13 23:33:00 +00:00
|
|
|
from yt_dlp.networking._requests import RequestsResponseAdapter
|
|
|
|
|
requests_res = RequestsResponse()
|
|
|
|
|
requests_res.raw = Urllib3Response(body=b'', status=200)
|
|
|
|
|
res = RequestsResponseAdapter(requests_res)
|
|
|
|
|
|
|
|
|
|
def mock_read(*args, **kwargs):
|
|
|
|
|
raise raised()
|
|
|
|
|
monkeypatch.setattr(res.fp, 'read', mock_read)
|
|
|
|
|
|
|
|
|
|
with pytest.raises(expected, match=match) as exc_info:
|
|
|
|
|
res.read()
|
|
|
|
|
|
|
|
|
|
assert exc_info.type is expected
|
|
|
|
|
|
2024-02-17 22:32:34 +00:00
|
|
|
def test_close(self, handler, monkeypatch):
|
|
|
|
|
rh = handler()
|
|
|
|
|
session = rh._get_instance(cookiejar=rh.cookiejar)
|
|
|
|
|
called = False
|
|
|
|
|
original_close = session.close
|
|
|
|
|
|
|
|
|
|
def mock_close(*args, **kwargs):
|
|
|
|
|
nonlocal called
|
|
|
|
|
called = True
|
|
|
|
|
return original_close(*args, **kwargs)
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(session, 'close', mock_close)
|
|
|
|
|
rh.close()
|
|
|
|
|
assert called
|
|
|
|
|
|
2023-10-13 23:33:00 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True)
|
|
|
|
|
class TestCurlCFFIRequestHandler(TestRequestHandlerBase):
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('params,extensions', [
|
|
|
|
|
({}, {'impersonate': ImpersonateTarget('chrome')}),
|
|
|
|
|
({'impersonate': ImpersonateTarget('chrome', '110')}, {}),
|
|
|
|
|
({'impersonate': ImpersonateTarget('chrome', '99')}, {'impersonate': ImpersonateTarget('chrome', '110')}),
|
|
|
|
|
])
|
|
|
|
|
def test_impersonate(self, handler, params, extensions):
|
|
|
|
|
with handler(headers=std_headers, **params) as rh:
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions=extensions)).read().decode()
|
|
|
|
|
assert 'sec-ch-ua: "Chromium";v="110"' in res
|
|
|
|
|
# Check that user agent is added over ours
|
|
|
|
|
assert 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36' in res
|
|
|
|
|
|
|
|
|
|
def test_headers(self, handler):
|
|
|
|
|
with handler(headers=std_headers) as rh:
|
|
|
|
|
# Ensure curl-impersonate overrides our standard headers (usually added
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', extensions={
|
|
|
|
|
'impersonate': ImpersonateTarget('safari')}, headers={'x-custom': 'test', 'sec-fetch-mode': 'custom'})).read().decode().lower()
|
|
|
|
|
|
|
|
|
|
assert std_headers['user-agent'].lower() not in res
|
|
|
|
|
assert std_headers['accept-language'].lower() not in res
|
|
|
|
|
assert std_headers['sec-fetch-mode'].lower() not in res
|
|
|
|
|
# other than UA, custom headers that differ from std_headers should be kept
|
|
|
|
|
assert 'sec-fetch-mode: custom' in res
|
|
|
|
|
assert 'x-custom: test' in res
|
|
|
|
|
# but when not impersonating don't remove std_headers
|
|
|
|
|
res = validate_and_send(
|
|
|
|
|
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'x-custom': 'test'})).read().decode().lower()
|
|
|
|
|
# std_headers should be present
|
|
|
|
|
for k, v in std_headers.items():
|
|
|
|
|
assert f'{k}: {v}'.lower() in res
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('raised,expected,match', [
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.PARTIAL_FILE), IncompleteRead, None),
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.OPERATION_TIMEDOUT), TransportError, None),
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.RECV_ERROR), TransportError, None),
|
|
|
|
|
])
|
|
|
|
|
def test_response_error_mapping(self, handler, monkeypatch, raised, expected, match):
|
|
|
|
|
import curl_cffi.requests
|
|
|
|
|
|
|
|
|
|
from yt_dlp.networking._curlcffi import CurlCFFIResponseAdapter
|
|
|
|
|
curl_res = curl_cffi.requests.Response()
|
|
|
|
|
res = CurlCFFIResponseAdapter(curl_res)
|
|
|
|
|
|
|
|
|
|
def mock_read(*args, **kwargs):
|
|
|
|
|
try:
|
|
|
|
|
raise raised()
|
|
|
|
|
except Exception as e:
|
|
|
|
|
e.response = curl_res
|
|
|
|
|
raise
|
|
|
|
|
monkeypatch.setattr(res.fp, 'read', mock_read)
|
|
|
|
|
|
|
|
|
|
with pytest.raises(expected, match=match) as exc_info:
|
|
|
|
|
res.read()
|
|
|
|
|
|
|
|
|
|
assert exc_info.type is expected
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('raised,expected,match', [
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.OPERATION_TIMEDOUT), TransportError, None),
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.PEER_FAILED_VERIFICATION), CertificateVerifyError, None),
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.SSL_CONNECT_ERROR), SSLError, None),
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.TOO_MANY_REDIRECTS), HTTPError, None),
|
|
|
|
|
(lambda: curl_cffi.requests.errors.RequestsError(
|
|
|
|
|
'', code=curl_cffi.const.CurlECode.PROXY), ProxyError, None),
|
|
|
|
|
])
|
|
|
|
|
def test_request_error_mapping(self, handler, monkeypatch, raised, expected, match):
|
|
|
|
|
import curl_cffi.requests
|
|
|
|
|
curl_res = curl_cffi.requests.Response()
|
|
|
|
|
curl_res.status_code = 301
|
|
|
|
|
|
|
|
|
|
with handler() as rh:
|
|
|
|
|
original_get_instance = rh._get_instance
|
|
|
|
|
|
|
|
|
|
def mock_get_instance(*args, **kwargs):
|
|
|
|
|
instance = original_get_instance(*args, **kwargs)
|
|
|
|
|
|
|
|
|
|
def request(*_, **__):
|
|
|
|
|
try:
|
|
|
|
|
raise raised()
|
|
|
|
|
except Exception as e:
|
|
|
|
|
e.response = curl_res
|
|
|
|
|
raise
|
|
|
|
|
monkeypatch.setattr(instance, 'request', request)
|
|
|
|
|
return instance
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(rh, '_get_instance', mock_get_instance)
|
|
|
|
|
|
|
|
|
|
with pytest.raises(expected) as exc_info:
|
|
|
|
|
rh.send(Request('http://fake'))
|
|
|
|
|
|
|
|
|
|
assert exc_info.type is expected
|
|
|
|
|
|
|
|
|
|
def test_response_reader(self, handler):
|
|
|
|
|
class FakeResponse:
|
|
|
|
|
def __init__(self, raise_error=False):
|
|
|
|
|
self.raise_error = raise_error
|
|
|
|
|
self.closed = False
|
|
|
|
|
|
|
|
|
|
def iter_content(self):
|
|
|
|
|
yield b'foo'
|
|
|
|
|
yield b'bar'
|
|
|
|
|
yield b'z'
|
|
|
|
|
if self.raise_error:
|
|
|
|
|
raise Exception('test')
|
|
|
|
|
|
|
|
|
|
def close(self):
|
|
|
|
|
self.closed = True
|
|
|
|
|
|
|
|
|
|
from yt_dlp.networking._curlcffi import CurlCFFIResponseReader
|
|
|
|
|
|
|
|
|
|
res = CurlCFFIResponseReader(FakeResponse())
|
|
|
|
|
assert res.readable
|
|
|
|
|
assert res.bytes_read == 0
|
|
|
|
|
assert res.read(1) == b'f'
|
|
|
|
|
assert res.bytes_read == 3
|
|
|
|
|
assert res._buffer == b'oo'
|
|
|
|
|
|
|
|
|
|
assert res.read(2) == b'oo'
|
|
|
|
|
assert res.bytes_read == 3
|
|
|
|
|
assert res._buffer == b''
|
|
|
|
|
|
|
|
|
|
assert res.read(2) == b'ba'
|
|
|
|
|
assert res.bytes_read == 6
|
|
|
|
|
assert res._buffer == b'r'
|
|
|
|
|
|
|
|
|
|
assert res.read(3) == b'rz'
|
|
|
|
|
assert res.bytes_read == 7
|
|
|
|
|
assert res._buffer == b''
|
|
|
|
|
assert res.closed
|
|
|
|
|
assert res._response.closed
|
|
|
|
|
|
|
|
|
|
# should handle no size param
|
|
|
|
|
res2 = CurlCFFIResponseReader(FakeResponse())
|
|
|
|
|
assert res2.read() == b'foobarz'
|
|
|
|
|
assert res2.bytes_read == 7
|
|
|
|
|
assert res2._buffer == b''
|
|
|
|
|
assert res2.closed
|
|
|
|
|
|
|
|
|
|
# should close on an exception
|
|
|
|
|
res3 = CurlCFFIResponseReader(FakeResponse(raise_error=True))
|
|
|
|
|
with pytest.raises(Exception, match='test'):
|
|
|
|
|
res3.read()
|
|
|
|
|
assert res3._buffer == b''
|
|
|
|
|
assert res3.bytes_read == 7
|
|
|
|
|
assert res3.closed
|
|
|
|
|
|
|
|
|
|
# buffer should be cleared on close
|
|
|
|
|
res4 = CurlCFFIResponseReader(FakeResponse())
|
|
|
|
|
res4.read(2)
|
|
|
|
|
assert res4._buffer == b'o'
|
|
|
|
|
res4.close()
|
|
|
|
|
assert res4.closed
|
|
|
|
|
assert res4._buffer == b''
|
|
|
|
|
|
|
|
|
|
|
2023-07-23 05:17:15 +00:00
|
|
|
def run_validation(handler, error, req, **handler_kwargs):
|
2023-07-15 10:25:23 +00:00
|
|
|
with handler(**handler_kwargs) as rh:
|
2023-07-23 05:17:15 +00:00
|
|
|
if error:
|
|
|
|
|
with pytest.raises(error):
|
2023-07-15 10:25:23 +00:00
|
|
|
rh.validate(req)
|
|
|
|
|
else:
|
|
|
|
|
rh.validate(req)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestRequestHandlerValidation:
|
|
|
|
|
|
|
|
|
|
class ValidationRH(RequestHandler):
|
|
|
|
|
def _send(self, request):
|
|
|
|
|
raise RequestError('test')
|
|
|
|
|
|
|
|
|
|
class NoCheckRH(ValidationRH):
|
|
|
|
|
_SUPPORTED_FEATURES = None
|
|
|
|
|
_SUPPORTED_PROXY_SCHEMES = None
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = None
|
|
|
|
|
|
2023-07-23 05:17:15 +00:00
|
|
|
def _check_extensions(self, extensions):
|
|
|
|
|
extensions.clear()
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
class HTTPSupportedRH(ValidationRH):
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ('http',)
|
|
|
|
|
|
|
|
|
|
URL_SCHEME_TESTS = [
|
|
|
|
|
# scheme, expected to fail, handler kwargs
|
|
|
|
|
('Urllib', [
|
|
|
|
|
('http', False, {}),
|
|
|
|
|
('https', False, {}),
|
|
|
|
|
('data', False, {}),
|
|
|
|
|
('ftp', False, {}),
|
2023-07-23 05:17:15 +00:00
|
|
|
('file', UnsupportedRequest, {}),
|
2023-07-15 10:25:23 +00:00
|
|
|
('file', False, {'enable_file_urls': True}),
|
|
|
|
|
]),
|
2023-10-13 23:33:00 +00:00
|
|
|
('Requests', [
|
|
|
|
|
('http', False, {}),
|
|
|
|
|
('https', False, {}),
|
|
|
|
|
]),
|
2023-11-20 08:04:04 +00:00
|
|
|
('Websockets', [
|
|
|
|
|
('ws', False, {}),
|
|
|
|
|
('wss', False, {}),
|
|
|
|
|
]),
|
2024-03-17 03:52:38 +00:00
|
|
|
('CurlCFFI', [
|
|
|
|
|
('http', False, {}),
|
|
|
|
|
('https', False, {}),
|
|
|
|
|
]),
|
2023-07-15 10:25:23 +00:00
|
|
|
(NoCheckRH, [('http', False, {})]),
|
2023-07-23 05:17:15 +00:00
|
|
|
(ValidationRH, [('http', UnsupportedRequest, {})])
|
2023-07-15 10:25:23 +00:00
|
|
|
]
|
|
|
|
|
|
|
|
|
|
PROXY_SCHEME_TESTS = [
|
|
|
|
|
# scheme, expected to fail
|
2023-11-20 08:04:04 +00:00
|
|
|
('Urllib', 'http', [
|
2023-07-15 10:25:23 +00:00
|
|
|
('http', False),
|
2023-07-23 05:17:15 +00:00
|
|
|
('https', UnsupportedRequest),
|
2023-07-15 10:25:23 +00:00
|
|
|
('socks4', False),
|
|
|
|
|
('socks4a', False),
|
|
|
|
|
('socks5', False),
|
|
|
|
|
('socks5h', False),
|
2023-07-23 05:17:15 +00:00
|
|
|
('socks', UnsupportedRequest),
|
2023-07-15 10:25:23 +00:00
|
|
|
]),
|
2023-11-20 08:04:04 +00:00
|
|
|
('Requests', 'http', [
|
2023-10-13 23:33:00 +00:00
|
|
|
('http', False),
|
|
|
|
|
('https', False),
|
|
|
|
|
('socks4', False),
|
|
|
|
|
('socks4a', False),
|
|
|
|
|
('socks5', False),
|
|
|
|
|
('socks5h', False),
|
|
|
|
|
]),
|
2024-03-17 03:52:38 +00:00
|
|
|
('CurlCFFI', 'http', [
|
|
|
|
|
('http', False),
|
|
|
|
|
('https', False),
|
|
|
|
|
('socks4', False),
|
|
|
|
|
('socks4a', False),
|
|
|
|
|
('socks5', False),
|
|
|
|
|
('socks5h', False),
|
|
|
|
|
]),
|
2023-11-20 08:04:04 +00:00
|
|
|
(NoCheckRH, 'http', [('http', False)]),
|
|
|
|
|
(HTTPSupportedRH, 'http', [('http', UnsupportedRequest)]),
|
|
|
|
|
('Websockets', 'ws', [('http', UnsupportedRequest)]),
|
|
|
|
|
(NoCheckRH, 'http', [('http', False)]),
|
|
|
|
|
(HTTPSupportedRH, 'http', [('http', UnsupportedRequest)]),
|
2023-07-15 10:25:23 +00:00
|
|
|
]
|
|
|
|
|
|
|
|
|
|
PROXY_KEY_TESTS = [
|
|
|
|
|
# key, expected to fail
|
|
|
|
|
('Urllib', [
|
|
|
|
|
('all', False),
|
|
|
|
|
('unrelated', False),
|
|
|
|
|
]),
|
2023-10-13 23:33:00 +00:00
|
|
|
('Requests', [
|
|
|
|
|
('all', False),
|
|
|
|
|
('unrelated', False),
|
|
|
|
|
]),
|
2024-03-17 03:52:38 +00:00
|
|
|
('CurlCFFI', [
|
|
|
|
|
('all', False),
|
|
|
|
|
('unrelated', False),
|
|
|
|
|
]),
|
2023-07-15 10:25:23 +00:00
|
|
|
(NoCheckRH, [('all', False)]),
|
2023-07-23 05:17:15 +00:00
|
|
|
(HTTPSupportedRH, [('all', UnsupportedRequest)]),
|
|
|
|
|
(HTTPSupportedRH, [('no', UnsupportedRequest)]),
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
EXTENSION_TESTS = [
|
2023-11-20 08:04:04 +00:00
|
|
|
('Urllib', 'http', [
|
2023-07-23 05:17:15 +00:00
|
|
|
({'cookiejar': 'notacookiejar'}, AssertionError),
|
2023-07-29 22:06:17 +00:00
|
|
|
({'cookiejar': YoutubeDLCookieJar()}, False),
|
|
|
|
|
({'cookiejar': CookieJar()}, AssertionError),
|
2023-07-23 05:17:15 +00:00
|
|
|
({'timeout': 1}, False),
|
|
|
|
|
({'timeout': 'notatimeout'}, AssertionError),
|
|
|
|
|
({'unsupported': 'value'}, UnsupportedRequest),
|
|
|
|
|
]),
|
2023-11-20 08:04:04 +00:00
|
|
|
('Requests', 'http', [
|
2023-10-13 23:33:00 +00:00
|
|
|
({'cookiejar': 'notacookiejar'}, AssertionError),
|
|
|
|
|
({'cookiejar': YoutubeDLCookieJar()}, False),
|
|
|
|
|
({'timeout': 1}, False),
|
|
|
|
|
({'timeout': 'notatimeout'}, AssertionError),
|
|
|
|
|
({'unsupported': 'value'}, UnsupportedRequest),
|
|
|
|
|
]),
|
2024-03-17 03:52:38 +00:00
|
|
|
('CurlCFFI', 'http', [
|
|
|
|
|
({'cookiejar': 'notacookiejar'}, AssertionError),
|
|
|
|
|
({'cookiejar': YoutubeDLCookieJar()}, False),
|
|
|
|
|
({'timeout': 1}, False),
|
|
|
|
|
({'timeout': 'notatimeout'}, AssertionError),
|
|
|
|
|
({'unsupported': 'value'}, UnsupportedRequest),
|
|
|
|
|
({'impersonate': ImpersonateTarget('badtarget', None, None, None)}, UnsupportedRequest),
|
|
|
|
|
({'impersonate': 123}, AssertionError),
|
|
|
|
|
({'impersonate': ImpersonateTarget('chrome', None, None, None)}, False),
|
|
|
|
|
({'impersonate': ImpersonateTarget(None, None, None, None)}, False),
|
|
|
|
|
({'impersonate': ImpersonateTarget()}, False),
|
|
|
|
|
({'impersonate': 'chrome'}, AssertionError)
|
|
|
|
|
]),
|
2023-11-20 08:04:04 +00:00
|
|
|
(NoCheckRH, 'http', [
|
2023-07-23 05:17:15 +00:00
|
|
|
({'cookiejar': 'notacookiejar'}, False),
|
|
|
|
|
({'somerandom': 'test'}, False), # but any extension is allowed through
|
|
|
|
|
]),
|
2023-11-20 08:04:04 +00:00
|
|
|
('Websockets', 'ws', [
|
|
|
|
|
({'cookiejar': YoutubeDLCookieJar()}, False),
|
|
|
|
|
({'timeout': 2}, False),
|
|
|
|
|
]),
|
2023-07-15 10:25:23 +00:00
|
|
|
]
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('handler,scheme,fail,handler_kwargs', [
|
|
|
|
|
(handler_tests[0], scheme, fail, handler_kwargs)
|
|
|
|
|
for handler_tests in URL_SCHEME_TESTS
|
|
|
|
|
for scheme, fail, handler_kwargs in handler_tests[1]
|
|
|
|
|
|
|
|
|
|
], indirect=['handler'])
|
|
|
|
|
def test_url_scheme(self, handler, scheme, fail, handler_kwargs):
|
|
|
|
|
run_validation(handler, fail, Request(f'{scheme}://'), **(handler_kwargs or {}))
|
|
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False), ('CurlCFFI', False)], indirect=['handler'])
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_no_proxy(self, handler, fail):
|
|
|
|
|
run_validation(handler, fail, Request('http://', proxies={'no': '127.0.0.1,github.com'}))
|
|
|
|
|
run_validation(handler, fail, Request('http://'), proxies={'no': '127.0.0.1,github.com'})
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('handler,proxy_key,fail', [
|
|
|
|
|
(handler_tests[0], proxy_key, fail)
|
|
|
|
|
for handler_tests in PROXY_KEY_TESTS
|
|
|
|
|
for proxy_key, fail in handler_tests[1]
|
|
|
|
|
], indirect=['handler'])
|
|
|
|
|
def test_proxy_key(self, handler, proxy_key, fail):
|
|
|
|
|
run_validation(handler, fail, Request('http://', proxies={proxy_key: 'http://example.com'}))
|
|
|
|
|
run_validation(handler, fail, Request('http://'), proxies={proxy_key: 'http://example.com'})
|
|
|
|
|
|
2023-11-20 08:04:04 +00:00
|
|
|
@pytest.mark.parametrize('handler,req_scheme,scheme,fail', [
|
|
|
|
|
(handler_tests[0], handler_tests[1], scheme, fail)
|
2023-07-15 10:25:23 +00:00
|
|
|
for handler_tests in PROXY_SCHEME_TESTS
|
2023-11-20 08:04:04 +00:00
|
|
|
for scheme, fail in handler_tests[2]
|
2023-07-15 10:25:23 +00:00
|
|
|
], indirect=['handler'])
|
2023-11-20 08:04:04 +00:00
|
|
|
def test_proxy_scheme(self, handler, req_scheme, scheme, fail):
|
|
|
|
|
run_validation(handler, fail, Request(f'{req_scheme}://', proxies={req_scheme: f'{scheme}://example.com'}))
|
|
|
|
|
run_validation(handler, fail, Request(f'{req_scheme}://'), proxies={req_scheme: f'{scheme}://example.com'})
|
2023-07-15 10:25:23 +00:00
|
|
|
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_empty_proxy(self, handler):
|
|
|
|
|
run_validation(handler, False, Request('http://', proxies={'http': None}))
|
|
|
|
|
run_validation(handler, False, Request('http://'), proxies={'http': None})
|
|
|
|
|
|
2023-07-27 14:56:02 +00:00
|
|
|
@pytest.mark.parametrize('proxy_url', ['//example.com', 'example.com', '127.0.0.1', '/a/b/c'])
|
2024-03-17 03:52:38 +00:00
|
|
|
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
|
2023-07-27 14:56:02 +00:00
|
|
|
def test_invalid_proxy_url(self, handler, proxy_url):
|
|
|
|
|
run_validation(handler, UnsupportedRequest, Request('http://', proxies={'http': proxy_url}))
|
2023-07-15 10:25:23 +00:00
|
|
|
|
2023-11-20 08:04:04 +00:00
|
|
|
@pytest.mark.parametrize('handler,scheme,extensions,fail', [
|
|
|
|
|
(handler_tests[0], handler_tests[1], extensions, fail)
|
2023-07-23 05:17:15 +00:00
|
|
|
for handler_tests in EXTENSION_TESTS
|
2023-11-20 08:04:04 +00:00
|
|
|
for extensions, fail in handler_tests[2]
|
2023-07-23 05:17:15 +00:00
|
|
|
], indirect=['handler'])
|
2023-11-20 08:04:04 +00:00
|
|
|
def test_extension(self, handler, scheme, extensions, fail):
|
2023-07-23 05:17:15 +00:00
|
|
|
run_validation(
|
2023-11-20 08:04:04 +00:00
|
|
|
handler, fail, Request(f'{scheme}://', extensions=extensions))
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
def test_invalid_request_type(self):
|
|
|
|
|
rh = self.ValidationRH(logger=FakeLogger())
|
|
|
|
|
for method in (rh.validate, rh.send):
|
|
|
|
|
with pytest.raises(TypeError, match='Expected an instance of Request'):
|
|
|
|
|
method('not a request')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class FakeResponse(Response):
|
|
|
|
|
def __init__(self, request):
|
|
|
|
|
# XXX: we could make request part of standard response interface
|
|
|
|
|
self.request = request
|
|
|
|
|
super().__init__(fp=io.BytesIO(b''), headers={}, url=request.url)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class FakeRH(RequestHandler):
|
|
|
|
|
|
2024-03-17 03:47:56 +00:00
|
|
|
def __init__(self, *args, **params):
|
|
|
|
|
self.params = params
|
|
|
|
|
super().__init__(*args, **params)
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def _validate(self, request):
|
|
|
|
|
return
|
|
|
|
|
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
if request.url.startswith('ssl://'):
|
|
|
|
|
raise SSLError(request.url[len('ssl://'):])
|
|
|
|
|
return FakeResponse(request)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class FakeRHYDL(FakeYDL):
|
|
|
|
|
def __init__(self, *args, **kwargs):
|
|
|
|
|
super().__init__(*args, **kwargs)
|
|
|
|
|
self._request_director = self.build_request_director([FakeRH])
|
|
|
|
|
|
|
|
|
|
|
2023-11-20 08:04:04 +00:00
|
|
|
class AllUnsupportedRHYDL(FakeYDL):
|
|
|
|
|
|
|
|
|
|
def __init__(self, *args, **kwargs):
|
|
|
|
|
|
|
|
|
|
class UnsupportedRH(RequestHandler):
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
|
_SUPPORTED_FEATURES = ()
|
|
|
|
|
_SUPPORTED_PROXY_SCHEMES = ()
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ()
|
|
|
|
|
|
|
|
|
|
super().__init__(*args, **kwargs)
|
|
|
|
|
self._request_director = self.build_request_director([UnsupportedRH])
|
|
|
|
|
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
class TestRequestDirector:
|
|
|
|
|
|
|
|
|
|
def test_handler_operations(self):
|
|
|
|
|
director = RequestDirector(logger=FakeLogger())
|
|
|
|
|
handler = FakeRH(logger=FakeLogger())
|
|
|
|
|
director.add_handler(handler)
|
|
|
|
|
assert director.handlers.get(FakeRH.RH_KEY) is handler
|
|
|
|
|
|
|
|
|
|
# Handler should overwrite
|
|
|
|
|
handler2 = FakeRH(logger=FakeLogger())
|
|
|
|
|
director.add_handler(handler2)
|
|
|
|
|
assert director.handlers.get(FakeRH.RH_KEY) is not handler
|
|
|
|
|
assert director.handlers.get(FakeRH.RH_KEY) is handler2
|
|
|
|
|
assert len(director.handlers) == 1
|
|
|
|
|
|
|
|
|
|
class AnotherFakeRH(FakeRH):
|
2015-03-20 13:59:38 +00:00
|
|
|
pass
|
2023-07-15 10:25:23 +00:00
|
|
|
director.add_handler(AnotherFakeRH(logger=FakeLogger()))
|
|
|
|
|
assert len(director.handlers) == 2
|
|
|
|
|
assert director.handlers.get(AnotherFakeRH.RH_KEY).RH_KEY == AnotherFakeRH.RH_KEY
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
director.handlers.pop(FakeRH.RH_KEY, None)
|
|
|
|
|
assert director.handlers.get(FakeRH.RH_KEY) is None
|
|
|
|
|
assert len(director.handlers) == 1
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
# RequestErrors should passthrough
|
|
|
|
|
with pytest.raises(SSLError):
|
|
|
|
|
director.send(Request('ssl://something'))
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_send(self):
|
|
|
|
|
director = RequestDirector(logger=FakeLogger())
|
|
|
|
|
with pytest.raises(RequestError):
|
|
|
|
|
director.send(Request('any://'))
|
|
|
|
|
director.add_handler(FakeRH(logger=FakeLogger()))
|
|
|
|
|
assert isinstance(director.send(Request('http://')), FakeResponse)
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_unsupported_handlers(self):
|
|
|
|
|
class SupportedRH(RequestHandler):
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ['http']
|
2015-03-20 13:59:38 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def _send(self, request: Request):
|
|
|
|
|
return Response(fp=io.BytesIO(b'supported'), headers={}, url=request.url)
|
2016-03-23 14:24:52 +00:00
|
|
|
|
2023-08-04 22:17:48 +00:00
|
|
|
director = RequestDirector(logger=FakeLogger())
|
2023-07-15 10:25:23 +00:00
|
|
|
director.add_handler(SupportedRH(logger=FakeLogger()))
|
2023-08-04 22:17:48 +00:00
|
|
|
director.add_handler(FakeRH(logger=FakeLogger()))
|
|
|
|
|
|
|
|
|
|
# First should take preference
|
2023-07-15 10:25:23 +00:00
|
|
|
assert director.send(Request('http://')).read() == b'supported'
|
|
|
|
|
assert director.send(Request('any://')).read() == b''
|
2016-11-17 11:42:56 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
director.handlers.pop(FakeRH.RH_KEY)
|
|
|
|
|
with pytest.raises(NoSupportingHandlers):
|
|
|
|
|
director.send(Request('any://'))
|
|
|
|
|
|
|
|
|
|
def test_unexpected_error(self):
|
|
|
|
|
director = RequestDirector(logger=FakeLogger())
|
|
|
|
|
|
|
|
|
|
class UnexpectedRH(FakeRH):
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
raise TypeError('something')
|
|
|
|
|
|
|
|
|
|
director.add_handler(UnexpectedRH(logger=FakeLogger))
|
|
|
|
|
with pytest.raises(NoSupportingHandlers, match=r'1 unexpected error'):
|
|
|
|
|
director.send(Request('any://'))
|
|
|
|
|
|
|
|
|
|
director.handlers.clear()
|
|
|
|
|
assert len(director.handlers) == 0
|
|
|
|
|
|
|
|
|
|
# Should not be fatal
|
|
|
|
|
director.add_handler(FakeRH(logger=FakeLogger()))
|
|
|
|
|
director.add_handler(UnexpectedRH(logger=FakeLogger))
|
|
|
|
|
assert director.send(Request('any://'))
|
|
|
|
|
|
2023-08-04 22:17:48 +00:00
|
|
|
def test_preference(self):
|
|
|
|
|
director = RequestDirector(logger=FakeLogger())
|
|
|
|
|
director.add_handler(FakeRH(logger=FakeLogger()))
|
|
|
|
|
|
|
|
|
|
class SomeRH(RequestHandler):
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ['http']
|
|
|
|
|
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
return Response(fp=io.BytesIO(b'supported'), headers={}, url=request.url)
|
|
|
|
|
|
|
|
|
|
def some_preference(rh, request):
|
|
|
|
|
return (0 if not isinstance(rh, SomeRH)
|
|
|
|
|
else 100 if 'prefer' in request.headers
|
|
|
|
|
else -1)
|
|
|
|
|
|
|
|
|
|
director.add_handler(SomeRH(logger=FakeLogger()))
|
|
|
|
|
director.preferences.add(some_preference)
|
|
|
|
|
|
|
|
|
|
assert director.send(Request('http://')).read() == b''
|
|
|
|
|
assert director.send(Request('http://', headers={'prefer': '1'})).read() == b'supported'
|
|
|
|
|
|
2024-02-17 22:32:34 +00:00
|
|
|
def test_close(self, monkeypatch):
|
|
|
|
|
director = RequestDirector(logger=FakeLogger())
|
|
|
|
|
director.add_handler(FakeRH(logger=FakeLogger()))
|
|
|
|
|
called = False
|
|
|
|
|
|
|
|
|
|
def mock_close(*args, **kwargs):
|
|
|
|
|
nonlocal called
|
|
|
|
|
called = True
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(director.handlers[FakeRH.RH_KEY], 'close', mock_close)
|
|
|
|
|
director.close()
|
|
|
|
|
assert called
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
# XXX: do we want to move this to test_YoutubeDL.py?
|
|
|
|
|
class TestYoutubeDLNetworking:
|
|
|
|
|
|
|
|
|
|
@staticmethod
|
|
|
|
|
def build_handler(ydl, handler: RequestHandler = FakeRH):
|
|
|
|
|
return ydl.build_request_director([handler]).handlers.get(handler.RH_KEY)
|
|
|
|
|
|
|
|
|
|
def test_compat_opener(self):
|
2023-05-27 07:06:13 +00:00
|
|
|
with FakeYDL() as ydl:
|
2023-07-15 10:25:23 +00:00
|
|
|
with warnings.catch_warnings():
|
|
|
|
|
warnings.simplefilter('ignore', category=DeprecationWarning)
|
|
|
|
|
assert isinstance(ydl._opener, urllib.request.OpenerDirector)
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('proxy,expected', [
|
|
|
|
|
('http://127.0.0.1:8080', {'all': 'http://127.0.0.1:8080'}),
|
|
|
|
|
('', {'all': '__noproxy__'}),
|
|
|
|
|
(None, {'http': 'http://127.0.0.1:8081', 'https': 'http://127.0.0.1:8081'}) # env, set https
|
|
|
|
|
])
|
2024-03-17 03:47:56 +00:00
|
|
|
def test_proxy(self, proxy, expected, monkeypatch):
|
|
|
|
|
monkeypatch.setenv('HTTP_PROXY', 'http://127.0.0.1:8081')
|
|
|
|
|
with FakeYDL({'proxy': proxy}) as ydl:
|
|
|
|
|
assert ydl.proxies == expected
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
def test_compat_request(self):
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
assert ydl.urlopen('test://')
|
|
|
|
|
urllib_req = urllib.request.Request('http://foo.bar', data=b'test', method='PUT', headers={'X-Test': '1'})
|
|
|
|
|
urllib_req.add_unredirected_header('Cookie', 'bob=bob')
|
|
|
|
|
urllib_req.timeout = 2
|
2023-07-09 07:53:02 +00:00
|
|
|
with warnings.catch_warnings():
|
|
|
|
|
warnings.simplefilter('ignore', category=DeprecationWarning)
|
|
|
|
|
req = ydl.urlopen(urllib_req).request
|
|
|
|
|
assert req.url == urllib_req.get_full_url()
|
|
|
|
|
assert req.data == urllib_req.data
|
|
|
|
|
assert req.method == urllib_req.get_method()
|
|
|
|
|
assert 'X-Test' in req.headers
|
|
|
|
|
assert 'Cookie' in req.headers
|
|
|
|
|
assert req.extensions.get('timeout') == 2
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
with pytest.raises(AssertionError):
|
|
|
|
|
ydl.urlopen(None)
|
|
|
|
|
|
|
|
|
|
def test_extract_basic_auth(self):
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
res = ydl.urlopen(Request('http://user:pass@foo.bar'))
|
|
|
|
|
assert res.request.headers['Authorization'] == 'Basic dXNlcjpwYXNz'
|
|
|
|
|
|
|
|
|
|
def test_sanitize_url(self):
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
res = ydl.urlopen(Request('httpss://foo.bar'))
|
|
|
|
|
assert res.request.url == 'https://foo.bar'
|
|
|
|
|
|
|
|
|
|
def test_file_urls_error(self):
|
|
|
|
|
# use urllib handler
|
|
|
|
|
with FakeYDL() as ydl:
|
|
|
|
|
with pytest.raises(RequestError, match=r'file:// URLs are disabled by default'):
|
|
|
|
|
ydl.urlopen('file://')
|
|
|
|
|
|
2023-11-20 08:04:04 +00:00
|
|
|
@pytest.mark.parametrize('scheme', (['ws', 'wss']))
|
|
|
|
|
def test_websocket_unavailable_error(self, scheme):
|
|
|
|
|
with AllUnsupportedRHYDL() as ydl:
|
|
|
|
|
with pytest.raises(RequestError, match=r'This request requires WebSocket support'):
|
|
|
|
|
ydl.urlopen(f'{scheme}://')
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_legacy_server_connect_error(self):
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
for error in ('UNSAFE_LEGACY_RENEGOTIATION_DISABLED', 'SSLV3_ALERT_HANDSHAKE_FAILURE'):
|
|
|
|
|
with pytest.raises(RequestError, match=r'Try using --legacy-server-connect'):
|
|
|
|
|
ydl.urlopen(f'ssl://{error}')
|
|
|
|
|
|
|
|
|
|
with pytest.raises(SSLError, match='testerror'):
|
|
|
|
|
ydl.urlopen('ssl://testerror')
|
|
|
|
|
|
2024-03-17 03:47:56 +00:00
|
|
|
def test_unsupported_impersonate_target(self):
|
|
|
|
|
class FakeImpersonationRHYDL(FakeYDL):
|
|
|
|
|
def __init__(self, *args, **kwargs):
|
|
|
|
|
class HTTPRH(RequestHandler):
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
pass
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ('http',)
|
|
|
|
|
_SUPPORTED_PROXY_SCHEMES = None
|
|
|
|
|
|
|
|
|
|
super().__init__(*args, **kwargs)
|
|
|
|
|
self._request_director = self.build_request_director([HTTPRH])
|
|
|
|
|
|
|
|
|
|
with FakeImpersonationRHYDL() as ydl:
|
|
|
|
|
with pytest.raises(
|
|
|
|
|
RequestError,
|
|
|
|
|
match=r'Impersonate target "test" is not available'
|
|
|
|
|
):
|
|
|
|
|
ydl.urlopen(Request('http://', extensions={'impersonate': ImpersonateTarget('test', None, None, None)}))
|
|
|
|
|
|
|
|
|
|
def test_unsupported_impersonate_extension(self):
|
|
|
|
|
class FakeHTTPRHYDL(FakeYDL):
|
|
|
|
|
def __init__(self, *args, **kwargs):
|
|
|
|
|
class IRH(ImpersonateRequestHandler):
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ('http',)
|
|
|
|
|
_SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget('abc',): 'test'}
|
|
|
|
|
_SUPPORTED_PROXY_SCHEMES = None
|
|
|
|
|
|
|
|
|
|
super().__init__(*args, **kwargs)
|
|
|
|
|
self._request_director = self.build_request_director([IRH])
|
|
|
|
|
|
|
|
|
|
with FakeHTTPRHYDL() as ydl:
|
|
|
|
|
with pytest.raises(
|
|
|
|
|
RequestError,
|
|
|
|
|
match=r'Impersonate target "test" is not available'
|
|
|
|
|
):
|
|
|
|
|
ydl.urlopen(Request('http://', extensions={'impersonate': ImpersonateTarget('test', None, None, None)}))
|
|
|
|
|
|
|
|
|
|
def test_raise_impersonate_error(self):
|
|
|
|
|
with pytest.raises(
|
|
|
|
|
YoutubeDLError,
|
|
|
|
|
match=r'Impersonate target "test" is not available'
|
|
|
|
|
):
|
|
|
|
|
FakeYDL({'impersonate': ImpersonateTarget('test', None, None, None)})
|
|
|
|
|
|
|
|
|
|
def test_pass_impersonate_param(self, monkeypatch):
|
|
|
|
|
|
|
|
|
|
class IRH(ImpersonateRequestHandler):
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ('http',)
|
|
|
|
|
_SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget('abc'): 'test'}
|
|
|
|
|
|
|
|
|
|
# Bypass the check on initialize
|
|
|
|
|
brh = FakeYDL.build_request_director
|
|
|
|
|
monkeypatch.setattr(FakeYDL, 'build_request_director', lambda cls, handlers, preferences=None: brh(cls, handlers=[IRH]))
|
|
|
|
|
|
|
|
|
|
with FakeYDL({
|
|
|
|
|
'impersonate': ImpersonateTarget('abc', None, None, None)
|
|
|
|
|
}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl, IRH)
|
|
|
|
|
assert rh.impersonate == ImpersonateTarget('abc', None, None, None)
|
|
|
|
|
|
|
|
|
|
def test_get_impersonate_targets(self):
|
|
|
|
|
handlers = []
|
|
|
|
|
for target_client in ('abc', 'xyz', 'asd'):
|
|
|
|
|
class TestRH(ImpersonateRequestHandler):
|
|
|
|
|
def _send(self, request: Request):
|
|
|
|
|
pass
|
|
|
|
|
_SUPPORTED_URL_SCHEMES = ('http',)
|
|
|
|
|
_SUPPORTED_IMPERSONATE_TARGET_MAP = {ImpersonateTarget(target_client,): 'test'}
|
|
|
|
|
RH_KEY = target_client
|
|
|
|
|
RH_NAME = target_client
|
|
|
|
|
handlers.append(TestRH)
|
|
|
|
|
|
|
|
|
|
with FakeYDL() as ydl:
|
|
|
|
|
ydl._request_director = ydl.build_request_director(handlers)
|
|
|
|
|
assert set(ydl._get_available_impersonate_targets()) == {
|
|
|
|
|
(ImpersonateTarget('xyz'), 'xyz'),
|
|
|
|
|
(ImpersonateTarget('abc'), 'abc'),
|
|
|
|
|
(ImpersonateTarget('asd'), 'asd')
|
|
|
|
|
}
|
|
|
|
|
assert ydl._impersonate_target_available(ImpersonateTarget('abc'))
|
|
|
|
|
assert ydl._impersonate_target_available(ImpersonateTarget())
|
|
|
|
|
assert not ydl._impersonate_target_available(ImpersonateTarget('zxy'))
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
@pytest.mark.parametrize('proxy_key,proxy_url,expected', [
|
|
|
|
|
('http', '__noproxy__', None),
|
|
|
|
|
('no', '127.0.0.1,foo.bar', '127.0.0.1,foo.bar'),
|
|
|
|
|
('https', 'example.com', 'http://example.com'),
|
2023-07-27 14:56:02 +00:00
|
|
|
('https', '//example.com', 'http://example.com'),
|
2023-07-15 10:25:23 +00:00
|
|
|
('https', 'socks5://example.com', 'socks5h://example.com'),
|
|
|
|
|
('http', 'socks://example.com', 'socks4://example.com'),
|
|
|
|
|
('http', 'socks4://example.com', 'socks4://example.com'),
|
2023-07-27 14:56:02 +00:00
|
|
|
('unrelated', '/bad/proxy', '/bad/proxy'), # clean_proxies should ignore bad proxies
|
2023-07-15 10:25:23 +00:00
|
|
|
])
|
2024-03-17 03:47:56 +00:00
|
|
|
def test_clean_proxy(self, proxy_key, proxy_url, expected, monkeypatch):
|
2023-07-15 10:25:23 +00:00
|
|
|
# proxies should be cleaned in urlopen()
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
req = ydl.urlopen(Request('test://', proxies={proxy_key: proxy_url})).request
|
|
|
|
|
assert req.proxies[proxy_key] == expected
|
|
|
|
|
|
|
|
|
|
# and should also be cleaned when building the handler
|
2024-03-17 03:47:56 +00:00
|
|
|
monkeypatch.setenv(f'{proxy_key.upper()}_PROXY', proxy_url)
|
|
|
|
|
with FakeYDL() as ydl:
|
|
|
|
|
rh = self.build_handler(ydl)
|
|
|
|
|
assert rh.proxies[proxy_key] == expected
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
def test_clean_proxy_header(self):
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
req = ydl.urlopen(Request('test://', headers={'ytdl-request-proxy': '//foo.bar'})).request
|
|
|
|
|
assert 'ytdl-request-proxy' not in req.headers
|
|
|
|
|
assert req.proxies == {'all': 'http://foo.bar'}
|
|
|
|
|
|
|
|
|
|
with FakeYDL({'http_headers': {'ytdl-request-proxy': '//foo.bar'}}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl)
|
|
|
|
|
assert 'ytdl-request-proxy' not in rh.headers
|
|
|
|
|
assert rh.proxies == {'all': 'http://foo.bar'}
|
|
|
|
|
|
|
|
|
|
def test_clean_header(self):
|
|
|
|
|
with FakeRHYDL() as ydl:
|
|
|
|
|
res = ydl.urlopen(Request('test://', headers={'Youtubedl-no-compression': True}))
|
|
|
|
|
assert 'Youtubedl-no-compression' not in res.request.headers
|
|
|
|
|
assert res.request.headers.get('Accept-Encoding') == 'identity'
|
|
|
|
|
|
|
|
|
|
with FakeYDL({'http_headers': {'Youtubedl-no-compression': True}}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl)
|
|
|
|
|
assert 'Youtubedl-no-compression' not in rh.headers
|
|
|
|
|
assert rh.headers.get('Accept-Encoding') == 'identity'
|
|
|
|
|
|
2023-08-16 23:42:48 +00:00
|
|
|
with FakeYDL({'http_headers': {'Ytdl-socks-proxy': 'socks://localhost:1080'}}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl)
|
|
|
|
|
assert 'Ytdl-socks-proxy' not in rh.headers
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_build_handler_params(self):
|
|
|
|
|
with FakeYDL({
|
|
|
|
|
'http_headers': {'test': 'testtest'},
|
|
|
|
|
'socket_timeout': 2,
|
|
|
|
|
'proxy': 'http://127.0.0.1:8080',
|
|
|
|
|
'source_address': '127.0.0.45',
|
|
|
|
|
'debug_printtraffic': True,
|
|
|
|
|
'compat_opts': ['no-certifi'],
|
|
|
|
|
'nocheckcertificate': True,
|
2023-07-20 13:31:17 +00:00
|
|
|
'legacyserverconnect': True,
|
2023-07-15 10:25:23 +00:00
|
|
|
}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl)
|
|
|
|
|
assert rh.headers.get('test') == 'testtest'
|
|
|
|
|
assert 'Accept' in rh.headers # ensure std_headers are still there
|
|
|
|
|
assert rh.timeout == 2
|
|
|
|
|
assert rh.proxies.get('all') == 'http://127.0.0.1:8080'
|
|
|
|
|
assert rh.source_address == '127.0.0.45'
|
|
|
|
|
assert rh.verbose is True
|
|
|
|
|
assert rh.prefer_system_certs is True
|
|
|
|
|
assert rh.verify is False
|
|
|
|
|
assert rh.legacy_ssl_support is True
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('ydl_params', [
|
|
|
|
|
{'client_certificate': 'fakecert.crt'},
|
|
|
|
|
{'client_certificate': 'fakecert.crt', 'client_certificate_key': 'fakekey.key'},
|
|
|
|
|
{'client_certificate': 'fakecert.crt', 'client_certificate_key': 'fakekey.key', 'client_certificate_password': 'foobar'},
|
|
|
|
|
{'client_certificate_key': 'fakekey.key', 'client_certificate_password': 'foobar'},
|
|
|
|
|
])
|
|
|
|
|
def test_client_certificate(self, ydl_params):
|
|
|
|
|
with FakeYDL(ydl_params) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl)
|
|
|
|
|
assert rh._client_cert == ydl_params # XXX: Too bound to implementation
|
|
|
|
|
|
|
|
|
|
def test_urllib_file_urls(self):
|
|
|
|
|
with FakeYDL({'enable_file_urls': False}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl, UrllibRH)
|
|
|
|
|
assert rh.enable_file_urls is False
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
with FakeYDL({'enable_file_urls': True}) as ydl:
|
|
|
|
|
rh = self.build_handler(ydl, UrllibRH)
|
|
|
|
|
assert rh.enable_file_urls is True
|
|
|
|
|
|
2023-10-13 23:33:00 +00:00
|
|
|
def test_compat_opt_prefer_urllib(self):
|
|
|
|
|
# This assumes urllib only has a preference when this compat opt is given
|
|
|
|
|
with FakeYDL({'compat_opts': ['prefer-legacy-http-handler']}) as ydl:
|
|
|
|
|
director = ydl.build_request_director([UrllibRH])
|
|
|
|
|
assert len(director.preferences) == 1
|
|
|
|
|
assert director.preferences.pop()(UrllibRH, None)
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
|
|
|
|
|
class TestRequest:
|
|
|
|
|
|
|
|
|
|
def test_query(self):
|
|
|
|
|
req = Request('http://example.com?q=something', query={'v': 'xyz'})
|
|
|
|
|
assert req.url == 'http://example.com?q=something&v=xyz'
|
|
|
|
|
|
|
|
|
|
req.update(query={'v': '123'})
|
|
|
|
|
assert req.url == 'http://example.com?q=something&v=123'
|
|
|
|
|
req.update(url='http://example.com', query={'v': 'xyz'})
|
|
|
|
|
assert req.url == 'http://example.com?v=xyz'
|
|
|
|
|
|
|
|
|
|
def test_method(self):
|
|
|
|
|
req = Request('http://example.com')
|
|
|
|
|
assert req.method == 'GET'
|
|
|
|
|
req.data = b'test'
|
|
|
|
|
assert req.method == 'POST'
|
|
|
|
|
req.data = None
|
|
|
|
|
assert req.method == 'GET'
|
|
|
|
|
req.data = b'test2'
|
|
|
|
|
req.method = 'PUT'
|
|
|
|
|
assert req.method == 'PUT'
|
|
|
|
|
req.data = None
|
|
|
|
|
assert req.method == 'PUT'
|
|
|
|
|
with pytest.raises(TypeError):
|
|
|
|
|
req.method = 1
|
|
|
|
|
|
|
|
|
|
def test_request_helpers(self):
|
|
|
|
|
assert HEADRequest('http://example.com').method == 'HEAD'
|
|
|
|
|
assert PUTRequest('http://example.com').method == 'PUT'
|
|
|
|
|
|
|
|
|
|
def test_headers(self):
|
|
|
|
|
req = Request('http://example.com', headers={'tesT': 'test'})
|
|
|
|
|
assert req.headers == HTTPHeaderDict({'test': 'test'})
|
|
|
|
|
req.update(headers={'teSt2': 'test2'})
|
|
|
|
|
assert req.headers == HTTPHeaderDict({'test': 'test', 'test2': 'test2'})
|
|
|
|
|
|
|
|
|
|
req.headers = new_headers = HTTPHeaderDict({'test': 'test'})
|
|
|
|
|
assert req.headers == HTTPHeaderDict({'test': 'test'})
|
|
|
|
|
assert req.headers is new_headers
|
|
|
|
|
|
|
|
|
|
# test converts dict to case insensitive dict
|
|
|
|
|
req.headers = new_headers = {'test2': 'test2'}
|
|
|
|
|
assert isinstance(req.headers, HTTPHeaderDict)
|
|
|
|
|
assert req.headers is not new_headers
|
|
|
|
|
|
|
|
|
|
with pytest.raises(TypeError):
|
|
|
|
|
req.headers = None
|
|
|
|
|
|
|
|
|
|
def test_data_type(self):
|
|
|
|
|
req = Request('http://example.com')
|
|
|
|
|
assert req.data is None
|
|
|
|
|
# test bytes is allowed
|
|
|
|
|
req.data = b'test'
|
|
|
|
|
assert req.data == b'test'
|
|
|
|
|
# test iterable of bytes is allowed
|
|
|
|
|
i = [b'test', b'test2']
|
|
|
|
|
req.data = i
|
|
|
|
|
assert req.data == i
|
|
|
|
|
|
|
|
|
|
# test file-like object is allowed
|
|
|
|
|
f = io.BytesIO(b'test')
|
|
|
|
|
req.data = f
|
|
|
|
|
assert req.data == f
|
|
|
|
|
|
|
|
|
|
# common mistake: test str not allowed
|
|
|
|
|
with pytest.raises(TypeError):
|
|
|
|
|
req.data = 'test'
|
|
|
|
|
assert req.data != 'test'
|
|
|
|
|
|
|
|
|
|
# common mistake: test dict is not allowed
|
|
|
|
|
with pytest.raises(TypeError):
|
|
|
|
|
req.data = {'test': 'test'}
|
|
|
|
|
assert req.data != {'test': 'test'}
|
|
|
|
|
|
|
|
|
|
def test_content_length_header(self):
|
|
|
|
|
req = Request('http://example.com', headers={'Content-Length': '0'}, data=b'')
|
|
|
|
|
assert req.headers.get('Content-Length') == '0'
|
|
|
|
|
|
|
|
|
|
req.data = b'test'
|
|
|
|
|
assert 'Content-Length' not in req.headers
|
|
|
|
|
|
|
|
|
|
req = Request('http://example.com', headers={'Content-Length': '10'})
|
|
|
|
|
assert 'Content-Length' not in req.headers
|
|
|
|
|
|
|
|
|
|
def test_content_type_header(self):
|
|
|
|
|
req = Request('http://example.com', headers={'Content-Type': 'test'}, data=b'test')
|
|
|
|
|
assert req.headers.get('Content-Type') == 'test'
|
|
|
|
|
req.data = b'test2'
|
|
|
|
|
assert req.headers.get('Content-Type') == 'test'
|
|
|
|
|
req.data = None
|
|
|
|
|
assert 'Content-Type' not in req.headers
|
|
|
|
|
req.data = b'test3'
|
|
|
|
|
assert req.headers.get('Content-Type') == 'application/x-www-form-urlencoded'
|
|
|
|
|
|
2023-07-20 13:23:30 +00:00
|
|
|
def test_update_req(self):
|
|
|
|
|
req = Request('http://example.com')
|
|
|
|
|
assert req.data is None
|
|
|
|
|
assert req.method == 'GET'
|
|
|
|
|
assert 'Content-Type' not in req.headers
|
|
|
|
|
# Test that zero-byte payloads will be sent
|
|
|
|
|
req.update(data=b'')
|
|
|
|
|
assert req.data == b''
|
|
|
|
|
assert req.method == 'POST'
|
|
|
|
|
assert req.headers.get('Content-Type') == 'application/x-www-form-urlencoded'
|
|
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
def test_proxies(self):
|
|
|
|
|
req = Request(url='http://example.com', proxies={'http': 'http://127.0.0.1:8080'})
|
|
|
|
|
assert req.proxies == {'http': 'http://127.0.0.1:8080'}
|
|
|
|
|
|
|
|
|
|
def test_extensions(self):
|
|
|
|
|
req = Request(url='http://example.com', extensions={'timeout': 2})
|
|
|
|
|
assert req.extensions == {'timeout': 2}
|
|
|
|
|
|
|
|
|
|
def test_copy(self):
|
|
|
|
|
req = Request(
|
|
|
|
|
url='http://example.com',
|
|
|
|
|
extensions={'cookiejar': CookieJar()},
|
|
|
|
|
headers={'Accept-Encoding': 'br'},
|
|
|
|
|
proxies={'http': 'http://127.0.0.1'},
|
|
|
|
|
data=[b'123']
|
|
|
|
|
)
|
|
|
|
|
req_copy = req.copy()
|
|
|
|
|
assert req_copy is not req
|
|
|
|
|
assert req_copy.url == req.url
|
|
|
|
|
assert req_copy.headers == req.headers
|
|
|
|
|
assert req_copy.headers is not req.headers
|
|
|
|
|
assert req_copy.proxies == req.proxies
|
|
|
|
|
assert req_copy.proxies is not req.proxies
|
|
|
|
|
|
|
|
|
|
# Data is not able to be copied
|
|
|
|
|
assert req_copy.data == req.data
|
|
|
|
|
assert req_copy.data is req.data
|
|
|
|
|
|
|
|
|
|
# Shallow copy extensions
|
|
|
|
|
assert req_copy.extensions is not req.extensions
|
|
|
|
|
assert req_copy.extensions['cookiejar'] == req.extensions['cookiejar']
|
|
|
|
|
|
|
|
|
|
# Subclasses are copied by default
|
|
|
|
|
class AnotherRequest(Request):
|
|
|
|
|
pass
|
2023-05-27 07:06:13 +00:00
|
|
|
|
2023-07-15 10:25:23 +00:00
|
|
|
req = AnotherRequest(url='http://127.0.0.1')
|
|
|
|
|
assert isinstance(req.copy(), AnotherRequest)
|
|
|
|
|
|
|
|
|
|
def test_url(self):
|
|
|
|
|
req = Request(url='https://фtest.example.com/ some spaceв?ä=c',)
|
|
|
|
|
assert req.url == 'https://xn--test-z6d.example.com/%20some%20space%D0%B2?%C3%A4=c'
|
|
|
|
|
|
|
|
|
|
assert Request(url='//example.com').url == 'http://example.com'
|
|
|
|
|
|
|
|
|
|
with pytest.raises(TypeError):
|
|
|
|
|
Request(url='https://').url = None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestResponse:
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('reason,status,expected', [
|
|
|
|
|
('custom', 200, 'custom'),
|
|
|
|
|
(None, 404, 'Not Found'), # fallback status
|
|
|
|
|
('', 403, 'Forbidden'),
|
|
|
|
|
(None, 999, None)
|
|
|
|
|
])
|
|
|
|
|
def test_reason(self, reason, status, expected):
|
|
|
|
|
res = Response(io.BytesIO(b''), url='test://', headers={}, status=status, reason=reason)
|
|
|
|
|
assert res.reason == expected
|
|
|
|
|
|
|
|
|
|
def test_headers(self):
|
|
|
|
|
headers = Message()
|
|
|
|
|
headers.add_header('Test', 'test')
|
|
|
|
|
headers.add_header('Test', 'test2')
|
|
|
|
|
headers.add_header('content-encoding', 'br')
|
|
|
|
|
res = Response(io.BytesIO(b''), headers=headers, url='test://')
|
|
|
|
|
assert res.headers.get_all('test') == ['test', 'test2']
|
|
|
|
|
assert 'Content-Encoding' in res.headers
|
|
|
|
|
|
|
|
|
|
def test_get_header(self):
|
|
|
|
|
headers = Message()
|
|
|
|
|
headers.add_header('Set-Cookie', 'cookie1')
|
|
|
|
|
headers.add_header('Set-cookie', 'cookie2')
|
|
|
|
|
headers.add_header('Test', 'test')
|
|
|
|
|
headers.add_header('Test', 'test2')
|
|
|
|
|
res = Response(io.BytesIO(b''), headers=headers, url='test://')
|
|
|
|
|
assert res.get_header('test') == 'test, test2'
|
|
|
|
|
assert res.get_header('set-Cookie') == 'cookie1'
|
|
|
|
|
assert res.get_header('notexist', 'default') == 'default'
|
|
|
|
|
|
|
|
|
|
def test_compat(self):
|
|
|
|
|
res = Response(io.BytesIO(b''), url='test://', status=404, headers={'test': 'test'})
|
2023-07-09 07:53:02 +00:00
|
|
|
with warnings.catch_warnings():
|
|
|
|
|
warnings.simplefilter('ignore', category=DeprecationWarning)
|
|
|
|
|
assert res.code == res.getcode() == res.status
|
|
|
|
|
assert res.geturl() == res.url
|
|
|
|
|
assert res.info() is res.headers
|
|
|
|
|
assert res.getheader('test') == res.get_header('test')
|
2024-03-17 03:47:56 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestImpersonateTarget:
|
|
|
|
|
@pytest.mark.parametrize('target_str,expected', [
|
|
|
|
|
('abc', ImpersonateTarget('abc', None, None, None)),
|
|
|
|
|
('abc-120_esr', ImpersonateTarget('abc', '120_esr', None, None)),
|
|
|
|
|
('abc-120:xyz', ImpersonateTarget('abc', '120', 'xyz', None)),
|
|
|
|
|
('abc-120:xyz-5.6', ImpersonateTarget('abc', '120', 'xyz', '5.6')),
|
|
|
|
|
('abc:xyz', ImpersonateTarget('abc', None, 'xyz', None)),
|
|
|
|
|
('abc:', ImpersonateTarget('abc', None, None, None)),
|
|
|
|
|
('abc-120:', ImpersonateTarget('abc', '120', None, None)),
|
|
|
|
|
(':xyz', ImpersonateTarget(None, None, 'xyz', None)),
|
|
|
|
|
(':xyz-6.5', ImpersonateTarget(None, None, 'xyz', '6.5')),
|
|
|
|
|
(':', ImpersonateTarget(None, None, None, None)),
|
|
|
|
|
('', ImpersonateTarget(None, None, None, None)),
|
|
|
|
|
])
|
|
|
|
|
def test_target_from_str(self, target_str, expected):
|
|
|
|
|
assert ImpersonateTarget.from_str(target_str) == expected
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('target_str', [
|
|
|
|
|
'-120', ':-12.0', '-12:-12', '-:-',
|
|
|
|
|
'::', 'a-c-d:', 'a-c-d:e-f-g', 'a:b:'
|
|
|
|
|
])
|
|
|
|
|
def test_target_from_invalid_str(self, target_str):
|
|
|
|
|
with pytest.raises(ValueError):
|
|
|
|
|
ImpersonateTarget.from_str(target_str)
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('target,expected', [
|
|
|
|
|
(ImpersonateTarget('abc', None, None, None), 'abc'),
|
|
|
|
|
(ImpersonateTarget('abc', '120', None, None), 'abc-120'),
|
|
|
|
|
(ImpersonateTarget('abc', '120', 'xyz', None), 'abc-120:xyz'),
|
|
|
|
|
(ImpersonateTarget('abc', '120', 'xyz', '5'), 'abc-120:xyz-5'),
|
|
|
|
|
(ImpersonateTarget('abc', None, 'xyz', None), 'abc:xyz'),
|
|
|
|
|
(ImpersonateTarget('abc', '120', None, None), 'abc-120'),
|
|
|
|
|
(ImpersonateTarget('abc', '120', 'xyz', None), 'abc-120:xyz'),
|
|
|
|
|
(ImpersonateTarget('abc', None, 'xyz'), 'abc:xyz'),
|
|
|
|
|
(ImpersonateTarget(None, None, 'xyz', '6.5'), ':xyz-6.5'),
|
|
|
|
|
(ImpersonateTarget('abc', ), 'abc'),
|
|
|
|
|
(ImpersonateTarget(None, None, None, None), ''),
|
|
|
|
|
])
|
|
|
|
|
def test_str(self, target, expected):
|
|
|
|
|
assert str(target) == expected
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('args', [
|
|
|
|
|
('abc', None, None, '5'),
|
|
|
|
|
('abc', '120', None, '5'),
|
|
|
|
|
(None, '120', None, None),
|
|
|
|
|
(None, '120', None, '5'),
|
|
|
|
|
(None, None, None, '5'),
|
|
|
|
|
(None, '120', 'xyz', '5'),
|
|
|
|
|
])
|
|
|
|
|
def test_invalid_impersonate_target(self, args):
|
|
|
|
|
with pytest.raises(ValueError):
|
|
|
|
|
ImpersonateTarget(*args)
|
|
|
|
|
|
|
|
|
|
@pytest.mark.parametrize('target1,target2,is_in,is_eq', [
|
|
|
|
|
(ImpersonateTarget('abc', None, None, None), ImpersonateTarget('abc', None, None, None), True, True),
|
|
|
|
|
(ImpersonateTarget('abc', None, None, None), ImpersonateTarget('abc', '120', None, None), True, False),
|
|
|
|
|
(ImpersonateTarget('abc', None, 'xyz', 'test'), ImpersonateTarget('abc', '120', 'xyz', None), True, False),
|
|
|
|
|
(ImpersonateTarget('abc', '121', 'xyz', 'test'), ImpersonateTarget('abc', '120', 'xyz', 'test'), False, False),
|
|
|
|
|
(ImpersonateTarget('abc'), ImpersonateTarget('abc', '120', 'xyz', 'test'), True, False),
|
|
|
|
|
(ImpersonateTarget('abc', '120', 'xyz', 'test'), ImpersonateTarget('abc'), True, False),
|
|
|
|
|
(ImpersonateTarget(), ImpersonateTarget('abc', '120', 'xyz'), True, False),
|
|
|
|
|
(ImpersonateTarget(), ImpersonateTarget(), True, True),
|
|
|
|
|
])
|
|
|
|
|
def test_impersonate_target_in(self, target1, target2, is_in, is_eq):
|
|
|
|
|
assert (target1 in target2) is is_in
|
|
|
|
|
assert (target1 == target2) is is_eq
|
