1
0
mirror of https://github.com/yt-dlp/yt-dlp synced 2025-07-15 18:17:46 -05:00

[rh:curlcffi] Add support for curl_cffi

Authored by: coletdjnz, Grub4K, pukkandan, bashonly

Co-authored-by: Simon Sawicki <contact@grub4k.xyz>
Co-authored-by: pukkandan <pukkandan.ytdlp@gmail.com>
Co-authored-by: bashonly <bashonly@protonmail.com>
This commit is contained in:
coletdjnz
2024-03-16 22:52:38 -05:00
committed by bashonly
parent 0b81d4d252
commit 52f5be1f1e
14 changed files with 628 additions and 140 deletions

View File

@ -30,7 +30,7 @@ from http.cookiejar import CookieJar
from test.conftest import validate_and_send
from test.helper import FakeYDL, http_server_port, verify_address_availability
from yt_dlp.cookies import YoutubeDLCookieJar
from yt_dlp.dependencies import brotli, requests, urllib3
from yt_dlp.dependencies import brotli, curl_cffi, requests, urllib3
from yt_dlp.networking import (
HEADRequest,
PUTRequest,
@ -57,7 +57,7 @@ from yt_dlp.networking.impersonate import (
)
from yt_dlp.utils import YoutubeDLError
from yt_dlp.utils._utils import _YDLLogger as FakeLogger
from yt_dlp.utils.networking import HTTPHeaderDict
from yt_dlp.utils.networking import HTTPHeaderDict, std_headers
TEST_DIR = os.path.dirname(os.path.abspath(__file__))
@ -79,6 +79,7 @@ def _build_proxy_handler(name):
class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
protocol_version = 'HTTP/1.1'
default_request_version = 'HTTP/1.1'
def log_message(self, format, *args):
pass
@ -116,6 +117,8 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
def _read_data(self):
if 'Content-Length' in self.headers:
return self.rfile.read(int(self.headers['Content-Length']))
else:
return b''
def do_POST(self):
data = self._read_data() + str(self.headers).encode()
@ -199,7 +202,8 @@ class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
self._headers()
elif self.path.startswith('/308-to-headers'):
self.send_response(308)
self.send_header('Location', '/headers')
# redirect to "localhost" for testing cookie redirection handling
self.send_header('Location', f'http://localhost:{self.connection.getsockname()[1]}/headers')
self.send_header('Content-Length', '0')
self.end_headers()
elif self.path == '/trailing_garbage':
@ -314,7 +318,7 @@ class TestRequestHandlerBase:
class TestHTTPRequestHandler(TestRequestHandlerBase):
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_verify_cert(self, handler):
with handler() as rh:
with pytest.raises(CertificateVerifyError):
@ -325,7 +329,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert r.status == 200
r.close()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
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?
@ -339,11 +343,11 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
https_server_thread.start()
with handler(verify=False) as rh:
with pytest.raises(SSLError, match=r'ssl(?:v3|/tls) alert handshake failure') as exc_info:
with pytest.raises(SSLError, match=r'(?i)ssl(?:v3|/tls).alert.handshake.failure') as exc_info:
validate_and_send(rh, Request(f'https://127.0.0.1:{https_port}/headers'))
assert not issubclass(exc_info.type, CertificateVerifyError)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_percent_encode(self, handler):
with handler() as rh:
# Unicode characters should be encoded with uppercase percent-encoding
@ -355,7 +359,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.status == 200
res.close()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
@pytest.mark.parametrize('path', [
'/a/b/./../../headers',
'/redirect_dotsegments',
@ -371,6 +375,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.url == f'http://127.0.0.1:{self.http_port}/headers'
res.close()
# Not supported by CurlCFFI (non-standard)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
def test_unicode_path_redirection(self, handler):
with handler() as rh:
@ -378,7 +383,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert r.url == f'http://127.0.0.1:{self.http_port}/%E4%B8%AD%E6%96%87.html'
r.close()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_raise_http_error(self, handler):
with handler() as rh:
for bad_status in (400, 500, 599, 302):
@ -388,7 +393,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
# Should not raise an error
validate_and_send(rh, Request('http://127.0.0.1:%d/gen_200' % self.http_port)).close()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_response_url(self, handler):
with handler() as rh:
# Response url should be that of the last url in redirect chain
@ -399,62 +404,50 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res2.url == f'http://127.0.0.1:{self.http_port}/gen_200'
res2.close()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
def test_redirect(self, handler):
# 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):
with handler() as rh:
def do_req(redirect_status, method, assert_no_content=False):
data = b'testdata' if method in ('POST', 'PUT') else None
res = validate_and_send(
rh, Request(f'http://127.0.0.1:{self.http_port}/redirect_{redirect_status}', method=method, data=data))
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))
headers = b''
data_sent = b''
if data is not None:
data_sent += res.read(len(data))
if data_sent != data:
headers += data_sent
data_sent = b''
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''
headers += res.read()
headers += res.read()
if assert_no_content or data is None:
assert b'Content-Type' not in headers
assert b'Content-Length' not in headers
else:
assert b'Content-Type' in headers
assert b'Content-Length' in headers
assert expected[0] == data_recv.decode()
assert expected[1] == res.headers.get('method')
assert expected[2] == ('content-length' in headers.decode().lower())
return data_sent.decode(), res.headers.get('method', '')
# A 303 must either use GET or HEAD for subsequent request
assert do_req(303, 'POST', True) == ('', 'GET')
assert do_req(303, 'HEAD') == ('', 'HEAD')
assert do_req(303, 'PUT', True) == ('', 'GET')
# 301 and 302 turn POST only into a GET
assert do_req(301, 'POST', True) == ('', 'GET')
assert do_req(301, 'HEAD') == ('', 'HEAD')
assert do_req(302, 'POST', True) == ('', 'GET')
assert do_req(302, 'HEAD') == ('', 'HEAD')
assert do_req(301, 'PUT') == ('testdata', 'PUT')
assert do_req(302, 'PUT') == ('testdata', 'PUT')
# 307 and 308 should not change method
for m in ('POST', 'PUT'):
assert do_req(307, m) == ('testdata', m)
assert do_req(308, m) == ('testdata', m)
assert do_req(307, 'HEAD') == ('', 'HEAD')
assert do_req(308, 'HEAD') == ('', 'HEAD')
# These should not redirect and instead raise an HTTPError
for code in (300, 304, 305, 306):
with pytest.raises(HTTPError):
do_req(code, 'GET')
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_request_cookie_header(self, handler):
# We should accept a Cookie header being passed as in normal headers and handle it appropriately.
with handler() as rh:
@ -463,16 +456,17 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
rh, Request(
f'http://127.0.0.1:{self.http_port}/headers',
headers={'Cookie': 'test=test'})).read().decode()
assert 'Cookie: test=test' in res
assert 'cookie: test=test' in res.lower()
# 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',
headers={'Cookie': 'test=test'})).read().decode()
assert 'Cookie: test=test' not in res
headers={'Cookie': 'test=test2'})).read().decode()
assert 'cookie: test=test2' not in res.lower()
# Specified Cookie header should override global cookiejar for that request
# Whether cookies from the cookiejar is applied on the redirect is considered undefined for now
cookiejar = YoutubeDLCookieJar()
cookiejar.set_cookie(http.cookiejar.Cookie(
version=0, name='test', value='ytdlp', port=None, port_specified=False,
@ -482,23 +476,23 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
with handler(cookiejar=cookiejar) as rh:
data = validate_and_send(
rh, Request(f'http://127.0.0.1:{self.http_port}/headers', headers={'cookie': 'test=test'})).read()
assert b'Cookie: test=ytdlp' not in data
assert b'Cookie: test=test' in data
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()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
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'))
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_incompleteread(self, handler):
with handler(timeout=2) as rh:
with pytest.raises(IncompleteRead):
with pytest.raises(IncompleteRead, match='13 bytes read, 234221 more expected'):
validate_and_send(rh, Request('http://127.0.0.1:%d/incompleteread' % self.http_port)).read()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_cookies(self, handler):
cookiejar = YoutubeDLCookieJar()
cookiejar.set_cookie(http.cookiejar.Cookie(
@ -507,47 +501,66 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
with handler(cookiejar=cookiejar) as rh:
data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read()
assert b'Cookie: test=ytdlp' in data
assert b'cookie: test=ytdlp' in data.lower()
# 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()
assert b'Cookie: test=ytdlp' in data
assert b'cookie: test=ytdlp' in data.lower()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_headers(self, handler):
with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh:
# Global Headers
data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read()
assert b'Test1: test' in data
data = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/headers')).read().lower()
assert b'test1: test' in data
# Per request headers, merged with global
data = validate_and_send(rh, Request(
f'http://127.0.0.1:{self.http_port}/headers', headers={'test2': 'changed', 'test3': 'test3'})).read()
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
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'], indirect=True)
def test_timeout(self, handler):
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_read_timeout(self, handler):
with handler() as rh:
# Default timeout is 20 seconds, so this should go through
validate_and_send(
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_3'))
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1'))
with handler(timeout=0.5) as rh:
with handler(timeout=0.1) as rh:
with pytest.raises(TransportError):
validate_and_send(
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_1'))
rh, Request(f'http://127.0.0.1:{self.http_port}/timeout_5'))
# 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}))
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@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)
def test_source_address(self, handler):
source_address = f'127.0.0.{random.randint(5, 255)}'
# on some systems these loopback addresses we need for testing may not be available
@ -558,6 +571,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
rh, Request(f'http://127.0.0.1:{self.http_port}/source_address')).read().decode()
assert source_address == data
# Not supported by CurlCFFI
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
def test_gzip_trailing_garbage(self, handler):
with handler() as rh:
@ -575,7 +589,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.headers.get('Content-Encoding') == 'br'
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_deflate(self, handler):
with handler() as rh:
res = validate_and_send(
@ -585,7 +599,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.headers.get('Content-Encoding') == 'deflate'
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_gzip(self, handler):
with handler() as rh:
res = validate_and_send(
@ -595,7 +609,7 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.headers.get('Content-Encoding') == 'gzip'
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_multiple_encodings(self, handler):
with handler() as rh:
for pair in ('gzip,deflate', 'deflate, gzip', 'gzip, gzip', 'deflate, deflate'):
@ -606,17 +620,18 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.headers.get('Content-Encoding') == pair
assert res.read() == b'<html><video src="/vid.mp4" /></html>'
# Not supported by curl_cffi
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
def test_unsupported_encoding(self, handler):
with handler() as rh:
res = validate_and_send(
rh, Request(
f'http://127.0.0.1:{self.http_port}/content-encoding',
headers={'ytdl-encoding': 'unsupported'}))
headers={'ytdl-encoding': 'unsupported', 'Accept-Encoding': '*'}))
assert res.headers.get('Content-Encoding') == 'unsupported'
assert res.read() == b'raw'
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_read(self, handler):
with handler() as rh:
res = validate_and_send(
@ -624,9 +639,12 @@ class TestHTTPRequestHandler(TestRequestHandlerBase):
assert res.readable()
assert res.read(1) == b'H'
assert res.read(3) == b'ost'
assert res.read().decode().endswith('\n\n')
assert res.read() == b''
class TestHTTPProxy(TestRequestHandlerBase):
# Note: this only tests http urls over non-CONNECT proxy
@classmethod
def setup_class(cls):
super().setup_class()
@ -646,7 +664,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
cls.geo_proxy_thread.daemon = True
cls.geo_proxy_thread.start()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
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}'
@ -672,7 +690,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
assert res != f'normal: {real_url}'
assert 'Accept' in res
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_noproxy(self, handler):
with handler(proxies={'proxy': f'http://127.0.0.1:{self.proxy_port}'}) as rh:
# NO_PROXY
@ -682,7 +700,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
'utf-8')
assert 'Accept' in nop_response
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_allproxy(self, handler):
url = 'http://foo.com/bar'
with handler() as rh:
@ -690,7 +708,7 @@ class TestHTTPProxy(TestRequestHandlerBase):
'utf-8')
assert response == f'normal: {url}'
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_http_proxy_with_idn(self, handler):
with handler(proxies={
'http': f'http://127.0.0.1:{self.proxy_port}',
@ -702,7 +720,6 @@ class TestHTTPProxy(TestRequestHandlerBase):
class TestClientCertificate:
@classmethod
def setup_class(cls):
certfn = os.path.join(TEST_DIR, 'testcert.pem')
@ -728,27 +745,27 @@ class TestClientCertificate:
) as rh:
validate_and_send(rh, Request(f'https://127.0.0.1:{self.port}/video.html')).read().decode()
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_certificate_combined_nopass(self, handler):
self._run_test(handler, client_cert={
'client_certificate': os.path.join(self.certdir, 'clientwithkey.crt'),
})
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
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'),
})
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
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',
})
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_certificate_nocombined_pass(self, handler):
self._run_test(handler, client_cert={
'client_certificate': os.path.join(self.certdir, 'client.crt'),
@ -757,6 +774,18 @@ class TestClientCertificate:
})
@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()
class TestRequestHandlerMisc:
"""Misc generic tests for request handlers, not related to request or validation testing"""
@pytest.mark.parametrize('handler,logger_name', [
@ -935,6 +964,172 @@ class TestRequestsRequestHandler(TestRequestHandlerBase):
assert called
@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''
def run_validation(handler, error, req, **handler_kwargs):
with handler(**handler_kwargs) as rh:
if error:
@ -979,6 +1174,10 @@ class TestRequestHandlerValidation:
('ws', False, {}),
('wss', False, {}),
]),
('CurlCFFI', [
('http', False, {}),
('https', False, {}),
]),
(NoCheckRH, [('http', False, {})]),
(ValidationRH, [('http', UnsupportedRequest, {})])
]
@ -1002,6 +1201,14 @@ class TestRequestHandlerValidation:
('socks5', False),
('socks5h', False),
]),
('CurlCFFI', 'http', [
('http', False),
('https', False),
('socks4', False),
('socks4a', False),
('socks5', False),
('socks5h', False),
]),
(NoCheckRH, 'http', [('http', False)]),
(HTTPSupportedRH, 'http', [('http', UnsupportedRequest)]),
('Websockets', 'ws', [('http', UnsupportedRequest)]),
@ -1019,6 +1226,10 @@ class TestRequestHandlerValidation:
('all', False),
('unrelated', False),
]),
('CurlCFFI', [
('all', False),
('unrelated', False),
]),
(NoCheckRH, [('all', False)]),
(HTTPSupportedRH, [('all', UnsupportedRequest)]),
(HTTPSupportedRH, [('no', UnsupportedRequest)]),
@ -1040,6 +1251,19 @@ class TestRequestHandlerValidation:
({'timeout': 'notatimeout'}, AssertionError),
({'unsupported': 'value'}, UnsupportedRequest),
]),
('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)
]),
(NoCheckRH, 'http', [
({'cookiejar': 'notacookiejar'}, False),
({'somerandom': 'test'}, False), # but any extension is allowed through
@ -1059,7 +1283,7 @@ class TestRequestHandlerValidation:
def test_url_scheme(self, handler, scheme, fail, handler_kwargs):
run_validation(handler, fail, Request(f'{scheme}://'), **(handler_kwargs or {}))
@pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False)], indirect=['handler'])
@pytest.mark.parametrize('handler,fail', [('Urllib', False), ('Requests', False), ('CurlCFFI', False)], indirect=['handler'])
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'})
@ -1082,13 +1306,13 @@ class TestRequestHandlerValidation:
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'})
@pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', HTTPSupportedRH, 'Requests', 'CurlCFFI'], indirect=True)
def test_empty_proxy(self, handler):
run_validation(handler, False, Request('http://', proxies={'http': None}))
run_validation(handler, False, Request('http://'), proxies={'http': None})
@pytest.mark.parametrize('proxy_url', ['//example.com', 'example.com', '127.0.0.1', '/a/b/c'])
@pytest.mark.parametrize('handler', ['Urllib', 'Requests'], indirect=True)
@pytest.mark.parametrize('handler', ['Urllib', 'Requests', 'CurlCFFI'], indirect=True)
def test_invalid_proxy_url(self, handler, proxy_url):
run_validation(handler, UnsupportedRequest, Request('http://', proxies={'http': proxy_url}))