mirrors/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-16 21:33:19 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'yt-dlp:master' into rls/arm-ubuntu-bump

Browse source
This commit is contained in:
bashonly 2024-03-17 01:56:07 -05:00
parent 8afe9d1f10 86d2f4d248
commit bd43228f06
No known key found for this signature in database
GPG key ID: 783F096F253D15B0
20 changed files with 1094 additions and 184 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

22
.github/workflows/build.yml vendored
View file

@ -249,9 +249,25 @@ jobs:
run: |
brew install coreutils
python3 devscripts/install_deps.py --user -o --include build
python3 devscripts/install_deps.py --print --include pyinstaller > requirements.txt
python3 devscripts/install_deps.py --print --include pyinstaller_macos > requirements.txt
# We need to ignore wheels otherwise we break universal2 builds
python3 -m pip install -U --user --no-binary :all: -r requirements.txt
# We need to fuse our own universal2 wheels for curl_cffi
python3 -m pip install -U --user delocate
mkdir curl_cffi_whls curl_cffi_universal2
python3 devscripts/install_deps.py --print -o --include curl_cffi > requirements.txt
for platform in "macosx_11_0_arm64" "macosx_11_0_x86_64"; do
python3 -m pip download \
--only-binary=:all: \
--platform "${platform}" \
--pre -d curl_cffi_whls \
-r requirements.txt
done
python3 -m delocate.cmd.delocate_fuse curl_cffi_whls/curl_cffi*.whl -w curl_cffi_universal2
python3 -m delocate.cmd.delocate_fuse curl_cffi_whls/cffi*.whl -w curl_cffi_universal2
cd curl_cffi_universal2
for wheel in *cffi*.whl; do mv -n -- "${wheel}" "${wheel/x86_64/universal2}"; done
python3 -m pip install -U --user *cffi*.whl
- name: Prepare
run: |
@ -305,7 +321,7 @@ jobs:
run: |
brew install coreutils
python3 devscripts/install_deps.py --user -o --include build
python3 devscripts/install_deps.py --user --include pyinstaller
python3 devscripts/install_deps.py --user --include pyinstaller_macos --include curl_cffi
- name: Prepare
run: |
@ -347,7 +363,7 @@ jobs:
- name: Install Requirements
run: | # Custom pyinstaller built with https://github.com/yt-dlp/pyinstaller-builds
python devscripts/install_deps.py -o --include build
python devscripts/install_deps.py --include py2exe
python devscripts/install_deps.py --include py2exe --include curl_cffi
python -m pip install -U "https://yt-dlp.github.io/Pyinstaller-Builds/x86_64/pyinstaller-5.8.0-py3-none-any.whl"
- name: Prepare

2
.github/workflows/core.yml vendored
View file

@ -53,7 +53,7 @@ jobs:
with:
python-version: ${{ matrix.python-version }}
- name: Install test requirements
run: python3 ./devscripts/install_deps.py --include dev
run: python3 ./devscripts/install_deps.py --include dev --include curl_cffi
- name: Run tests
continue-on-error: False
run: |

13
README.md
View file

@ -196,6 +196,15 @@ ### Networking
* [**websockets**](https://github.com/aaugustin/websockets)\* - For downloading over websocket. Licensed under [BSD-3-Clause](https://github.com/aaugustin/websockets/blob/main/LICENSE)
* [**requests**](https://github.com/psf/requests)\* - HTTP library. For HTTPS proxy and persistent connections support. Licensed under [Apache-2.0](https://github.com/psf/requests/blob/main/LICENSE)
#### Impersonation
The following provide support for impersonating browser requests. This may be required for some sites that employ TLS fingerprinting.
* [**curl_cffi**](https://github.com/yifeikong/curl_cffi) (recommended) - Python binding for [curl-impersonate](https://github.com/lwthiker/curl-impersonate). Provides impersonation targets for Chrome, Edge and Safari. Licensed under [MIT](https://github.com/yifeikong/curl_cffi/blob/main/LICENSE)
* Can be installed with the `curl_cffi` group, e.g. `pip install yt-dlp[default,curl_cffi]`
* Only included in `yt-dlp.exe`, `yt-dlp_macos` and `yt-dlp_macos_legacy` builds
### Metadata
* [**mutagen**](https://github.com/quodlibet/mutagen)\* - For `--embed-thumbnail` in certain formats. Licensed under [GPLv2+](https://github.com/quodlibet/mutagen/blob/master/COPYING)
@ -389,6 +398,10 @@ ## Network Options:
direct connection
--socket-timeout SECONDS Time to wait before giving up, in seconds
--source-address IP Client-side IP address to bind to
--impersonate CLIENT[:OS] Client to impersonate for requests. E.g.
chrome, chrome-110, chrome:windows-10. Pass
--impersonate="" to impersonate any client.
--list-impersonate-targets List available clients to impersonate.
-4, --force-ipv4 Make all connections via IPv4
-6, --force-ipv6 Make all connections via IPv6
--enable-file-urls Enable file:// URLs. This is disabled by

2
pyproject.toml
View file

@ -53,6 +53,7 @@ dependencies = [
[project.optional-dependencies]
default = []
curl_cffi = ["curl-cffi==0.5.10; implementation_name=='cpython'"]
secretstorage = [
"cffi",
"secretstorage",
@ -69,6 +70,7 @@ dev = [
"pytest",
]
pyinstaller = ["pyinstaller>=6.3"]
pyinstaller_macos = ["pyinstaller==5.13.2"] # needed for curl_cffi builds
py2exe = ["py2exe>=0.12"]
[project.urls]

630
test/test_networking.py
View file

@ -27,9 +27,10 @@
from email.message import Message
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,
@ -50,10 +51,13 @@
TransportError,
UnsupportedRequest,
)
from yt_dlp.networking.impersonate import (
ImpersonateRequestHandler,
ImpersonateTarget,
)
from yt_dlp.utils import YoutubeDLError
from yt_dlp.utils._utils import _YDLLogger as FakeLogger
from yt_dlp.utils.networking import HTTPHeaderDict
from test.conftest import validate_and_send
from yt_dlp.utils.networking import HTTPHeaderDict, std_headers
TEST_DIR = os.path.dirname(os.path.abspath(__file__))
@ -75,6 +79,7 @@ def do_GET(self):
class HTTPTestRequestHandler(http.server.BaseHTTPRequestHandler):
protocol_version = 'HTTP/1.1'
default_request_version = 'HTTP/1.1'
def log_message(self, format, *args):
pass
@ -112,6 +117,8 @@ def _status(self, status):
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()
@ -195,7 +202,8 @@ def do_GET(self):
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':
@ -310,7 +318,7 @@ def setup_class(cls):
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):
@ -321,7 +329,7 @@ def test_verify_cert(self, handler):
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?
@ -335,11 +343,11 @@ def test_ssl_error(self, handler):
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
@ -351,7 +359,7 @@ def test_percent_encode(self, handler):
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',
@ -367,6 +375,7 @@ def test_remove_dot_segments(self, handler, path):
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:
@ -374,7 +383,7 @@ def test_unicode_path_redirection(self, handler):
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):
@ -384,7 +393,7 @@ def test_raise_http_error(self, handler):
# 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
@ -395,62 +404,50 @@ def test_response_url(self, handler):
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:
@ -459,16 +456,17 @@ def test_request_cookie_header(self, handler):
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,
@ -478,23 +476,23 @@ def test_request_cookie_header(self, handler):
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(
@ -503,47 +501,66 @@ def test_cookies(self, handler):
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
@ -554,6 +571,7 @@ def test_source_address(self, handler):
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:
@ -571,7 +589,7 @@ def test_brotli(self, handler):
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(
@ -581,7 +599,7 @@ def test_deflate(self, handler):
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(
@ -591,7 +609,7 @@ def test_gzip(self, handler):
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'):
@ -602,17 +620,18 @@ def test_multiple_encodings(self, handler):
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(
@ -620,9 +639,12 @@ def test_read(self, handler):
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()
@ -642,7 +664,7 @@ def setup_class(cls):
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}'
@ -668,7 +690,7 @@ def test_http_proxy(self, handler):
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
@ -678,7 +700,7 @@ def test_noproxy(self, handler):
'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:
@ -686,7 +708,7 @@ def test_allproxy(self, handler):
'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}',
@ -698,7 +720,6 @@ def test_http_proxy_with_idn(self, handler):
class TestClientCertificate:
@classmethod
def setup_class(cls):
certfn = os.path.join(TEST_DIR, 'testcert.pem')
@ -724,27 +745,27 @@ def _run_test(self, handler, **handler_kwargs):
) 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'),
@ -753,6 +774,18 @@ def test_certificate_nocombined_pass(self, handler):
})
@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', [
@ -931,6 +964,172 @@ def mock_close(*args, **kwargs):
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:
@ -975,6 +1174,10 @@ class HTTPSupportedRH(ValidationRH):
('ws', False, {}),
('wss', False, {}),
]),
('CurlCFFI', [
('http', False, {}),
('https', False, {}),
]),
(NoCheckRH, [('http', False, {})]),
(ValidationRH, [('http', UnsupportedRequest, {})])
]
@ -998,6 +1201,14 @@ class HTTPSupportedRH(ValidationRH):
('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)]),
@ -1015,6 +1226,10 @@ class HTTPSupportedRH(ValidationRH):
('all', False),
('unrelated', False),
]),
('CurlCFFI', [
('all', False),
('unrelated', False),
]),
(NoCheckRH, [('all', False)]),
(HTTPSupportedRH, [('all', UnsupportedRequest)]),
(HTTPSupportedRH, [('no', UnsupportedRequest)]),
@ -1036,6 +1251,19 @@ class HTTPSupportedRH(ValidationRH):
({'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
@ -1055,7 +1283,7 @@ class HTTPSupportedRH(ValidationRH):
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'})
@ -1078,13 +1306,13 @@ 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'})
@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}))
@ -1113,6 +1341,10 @@ def __init__(self, request):
class FakeRH(RequestHandler):
def __init__(self, *args, **params):
self.params = params
super().__init__(*args, **params)
def _validate(self, request):
return
@ -1271,15 +1503,10 @@ def test_compat_opener(self):
('', {'all': '__noproxy__'}),
(None, {'http': 'http://127.0.0.1:8081', 'https': 'http://127.0.0.1:8081'}) # env, set https
])
def test_proxy(self, proxy, expected):
old_http_proxy = os.environ.get('HTTP_PROXY')
try:
os.environ['HTTP_PROXY'] = 'http://127.0.0.1:8081' # ensure that provided proxies override env
with FakeYDL({'proxy': proxy}) as ydl:
assert ydl.proxies == expected
finally:
if old_http_proxy:
os.environ['HTTP_PROXY'] = old_http_proxy
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
def test_compat_request(self):
with FakeRHYDL() as ydl:
@ -1331,6 +1558,95 @@ def test_legacy_server_connect_error(self):
with pytest.raises(SSLError, match='testerror'):
ydl.urlopen('ssl://testerror')
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'))
@pytest.mark.parametrize('proxy_key,proxy_url,expected', [
('http', '__noproxy__', None),
('no', '127.0.0.1,foo.bar', '127.0.0.1,foo.bar'),
@ -1341,23 +1657,17 @@ def test_legacy_server_connect_error(self):
('http', 'socks4://example.com', 'socks4://example.com'),
('unrelated', '/bad/proxy', '/bad/proxy'), # clean_proxies should ignore bad proxies
])
def test_clean_proxy(self, proxy_key, proxy_url, expected):
def test_clean_proxy(self, proxy_key, proxy_url, expected, monkeypatch):
# 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
env_key = f'{proxy_key.upper()}_PROXY'
old_env_proxy = os.environ.get(env_key)
try:
os.environ[env_key] = proxy_url # ensure that provided proxies override env
with FakeYDL() as ydl:
rh = self.build_handler(ydl)
assert rh.proxies[proxy_key] == expected
finally:
if old_env_proxy:
os.environ[env_key] = old_env_proxy
monkeypatch.setenv(f'{proxy_key.upper()}_PROXY', proxy_url)
with FakeYDL() as ydl:
rh = self.build_handler(ydl)
assert rh.proxies[proxy_key] == expected
def test_clean_proxy_header(self):
with FakeRHYDL() as ydl:
@ -1629,3 +1939,71 @@ def test_compat(self):
assert res.geturl() == res.url
assert res.info() is res.headers
assert res.getheader('test') == res.get_header('test')
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

33
test/test_socks.py
View file

@ -286,8 +286,14 @@ def ctx(request):
return CTX_MAP[request.param]()
@pytest.mark.parametrize(
'handler,ctx', [
('Urllib', 'http'),
('Requests', 'http'),
('Websockets', 'ws'),
('CurlCFFI', 'http')
], indirect=True)
class TestSocks4Proxy:
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks4_no_auth(self, handler, ctx):
with handler() as rh:
with ctx.socks_server(Socks4ProxyHandler) as server_address:
@ -295,7 +301,6 @@ def test_socks4_no_auth(self, handler, ctx):
rh, proxies={'all': f'socks4://{server_address}'})
assert response['version'] == 4
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks4_auth(self, handler, ctx):
with handler() as rh:
with ctx.socks_server(Socks4ProxyHandler, user_id='user') as server_address:
@ -305,7 +310,6 @@ def test_socks4_auth(self, handler, ctx):
rh, proxies={'all': f'socks4://user:@{server_address}'})
assert response['version'] == 4
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks4a_ipv4_target(self, handler, ctx):
with ctx.socks_server(Socks4ProxyHandler) as server_address:
with handler(proxies={'all': f'socks4a://{server_address}'}) as rh:
@ -313,7 +317,6 @@ def test_socks4a_ipv4_target(self, handler, ctx):
assert response['version'] == 4
assert (response['ipv4_address'] == '127.0.0.1') != (response['domain_address'] == '127.0.0.1')
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks4a_domain_target(self, handler, ctx):
with ctx.socks_server(Socks4ProxyHandler) as server_address:
with handler(proxies={'all': f'socks4a://{server_address}'}) as rh:
@ -322,7 +325,6 @@ def test_socks4a_domain_target(self, handler, ctx):
assert response['ipv4_address'] is None
assert response['domain_address'] == 'localhost'
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_ipv4_client_source_address(self, handler, ctx):
with ctx.socks_server(Socks4ProxyHandler) as server_address:
source_address = f'127.0.0.{random.randint(5, 255)}'
@ -333,7 +335,6 @@ def test_ipv4_client_source_address(self, handler, ctx):
assert response['client_address'][0] == source_address
assert response['version'] == 4
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
@pytest.mark.parametrize('reply_code', [
Socks4CD.REQUEST_REJECTED_OR_FAILED,
Socks4CD.REQUEST_REJECTED_CANNOT_CONNECT_TO_IDENTD,
@ -345,7 +346,6 @@ def test_socks4_errors(self, handler, ctx, reply_code):
with pytest.raises(ProxyError):
ctx.socks_info_request(rh)
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_ipv6_socks4_proxy(self, handler, ctx):
with ctx.socks_server(Socks4ProxyHandler, bind_ip='::1') as server_address:
with handler(proxies={'all': f'socks4://{server_address}'}) as rh:
@ -354,7 +354,6 @@ def test_ipv6_socks4_proxy(self, handler, ctx):
assert response['ipv4_address'] == '127.0.0.1'
assert response['version'] == 4
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_timeout(self, handler, ctx):
with ctx.socks_server(Socks4ProxyHandler, sleep=2) as server_address:
with handler(proxies={'all': f'socks4://{server_address}'}, timeout=0.5) as rh:
@ -362,9 +361,15 @@ def test_timeout(self, handler, ctx):
ctx.socks_info_request(rh)
@pytest.mark.parametrize(
'handler,ctx', [
('Urllib', 'http'),
('Requests', 'http'),
('Websockets', 'ws'),
('CurlCFFI', 'http')
], indirect=True)
class TestSocks5Proxy:
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5_no_auth(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@ -372,7 +377,6 @@ def test_socks5_no_auth(self, handler, ctx):
assert response['auth_methods'] == [0x0]
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5_user_pass(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler, auth=('test', 'testpass')) as server_address:
with handler() as rh:
@ -385,7 +389,6 @@ def test_socks5_user_pass(self, handler, ctx):
assert response['auth_methods'] == [Socks5Auth.AUTH_NONE, Socks5Auth.AUTH_USER_PASS]
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5_ipv4_target(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@ -393,7 +396,6 @@ def test_socks5_ipv4_target(self, handler, ctx):
assert response['ipv4_address'] == '127.0.0.1'
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5_domain_target(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@ -401,7 +403,6 @@ def test_socks5_domain_target(self, handler, ctx):
assert (response['ipv4_address'] == '127.0.0.1') != (response['ipv6_address'] == '::1')
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5h_domain_target(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
with handler(proxies={'all': f'socks5h://{server_address}'}) as rh:
@ -410,7 +411,6 @@ def test_socks5h_domain_target(self, handler, ctx):
assert response['domain_address'] == 'localhost'
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5h_ip_target(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
with handler(proxies={'all': f'socks5h://{server_address}'}) as rh:
@ -419,7 +419,6 @@ def test_socks5h_ip_target(self, handler, ctx):
assert response['domain_address'] is None
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_socks5_ipv6_destination(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@ -427,7 +426,6 @@ def test_socks5_ipv6_destination(self, handler, ctx):
assert response['ipv6_address'] == '::1'
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_ipv6_socks5_proxy(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler, bind_ip='::1') as server_address:
with handler(proxies={'all': f'socks5://{server_address}'}) as rh:
@ -438,7 +436,6 @@ def test_ipv6_socks5_proxy(self, handler, ctx):
# XXX: is there any feasible way of testing IPv6 source addresses?
# Same would go for non-proxy source_address test...
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
def test_ipv4_client_source_address(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler) as server_address:
source_address = f'127.0.0.{random.randint(5, 255)}'
@ -448,7 +445,6 @@ def test_ipv4_client_source_address(self, handler, ctx):
assert response['client_address'][0] == source_address
assert response['version'] == 5
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Requests', 'http'), ('Websockets', 'ws')], indirect=True)
@pytest.mark.parametrize('reply_code', [
Socks5Reply.GENERAL_FAILURE,
Socks5Reply.CONNECTION_NOT_ALLOWED,
@ -465,7 +461,6 @@ def test_socks5_errors(self, handler, ctx, reply_code):
with pytest.raises(ProxyError):
ctx.socks_info_request(rh)
@pytest.mark.parametrize('handler,ctx', [('Urllib', 'http'), ('Websockets', 'ws')], indirect=True)
def test_timeout(self, handler, ctx):
with ctx.socks_server(Socks5ProxyHandler, sleep=2) as server_address:
with handler(proxies={'all': f'socks5://{server_address}'}, timeout=1) as rh:

53
test/test_websockets.py
View file

@ -32,8 +32,6 @@
)
from yt_dlp.utils.networking import HTTPHeaderDict
from test.conftest import validate_and_send
TEST_DIR = os.path.dirname(os.path.abspath(__file__))
@ -66,7 +64,9 @@ def process_request(self, request):
def create_websocket_server(**ws_kwargs):
import websockets.sync.server
wsd = websockets.sync.server.serve(websocket_handler, '127.0.0.1', 0, process_request=process_request, **ws_kwargs)
wsd = websockets.sync.server.serve(
websocket_handler, '127.0.0.1', 0,
process_request=process_request, open_timeout=2, **ws_kwargs)
ws_port = wsd.socket.getsockname()[1]
ws_server_thread = threading.Thread(target=wsd.serve_forever)
ws_server_thread.daemon = True
@ -100,6 +100,19 @@ def create_mtls_wss_websocket_server():
return create_websocket_server(ssl_context=sslctx)
def ws_validate_and_send(rh, req):
rh.validate(req)
max_tries = 3
for i in range(max_tries):
try:
return rh.send(req)
except TransportError as e:
if i < (max_tries - 1) and 'connection closed during handshake' in str(e):
# websockets server sometimes hangs on new connections
continue
raise
@pytest.mark.skipif(not websockets, reason='websockets must be installed to test websocket request handlers')
class TestWebsSocketRequestHandlerConformance:
@classmethod
@ -119,7 +132,7 @@ def setup_class(cls):
@pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
def test_basic_websockets(self, handler):
with handler() as rh:
ws = validate_and_send(rh, Request(self.ws_base_url))
ws = ws_validate_and_send(rh, Request(self.ws_base_url))
assert 'upgrade' in ws.headers
assert ws.status == 101
ws.send('foo')
@ -131,7 +144,7 @@ def test_basic_websockets(self, handler):
@pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
def test_send_types(self, handler, msg, opcode):
with handler() as rh:
ws = validate_and_send(rh, Request(self.ws_base_url))
ws = ws_validate_and_send(rh, Request(self.ws_base_url))
ws.send(msg)
assert int(ws.recv()) == opcode
ws.close()
@ -140,10 +153,10 @@ def test_send_types(self, handler, msg, opcode):
def test_verify_cert(self, handler):
with handler() as rh:
with pytest.raises(CertificateVerifyError):
validate_and_send(rh, Request(self.wss_base_url))
ws_validate_and_send(rh, Request(self.wss_base_url))
with handler(verify=False) as rh:
ws = validate_and_send(rh, Request(self.wss_base_url))
ws = ws_validate_and_send(rh, Request(self.wss_base_url))
assert ws.status == 101
ws.close()
@ -151,7 +164,7 @@ def test_verify_cert(self, handler):
def test_ssl_error(self, handler):
with handler(verify=False) as rh:
with pytest.raises(SSLError, match=r'ssl(?:v3|/tls) alert handshake failure') as exc_info:
validate_and_send(rh, Request(self.bad_wss_host))
ws_validate_and_send(rh, Request(self.bad_wss_host))
assert not issubclass(exc_info.type, CertificateVerifyError)
@pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
@ -163,7 +176,7 @@ def test_ssl_error(self, handler):
])
def test_percent_encode(self, handler, path, expected):
with handler() as rh:
ws = validate_and_send(rh, Request(f'{self.ws_base_url}{path}'))
ws = ws_validate_and_send(rh, Request(f'{self.ws_base_url}{path}'))
ws.send('path')
assert ws.recv() == expected
assert ws.status == 101
@ -174,7 +187,7 @@ def test_remove_dot_segments(self, handler):
with handler() as rh:
# This isn't a comprehensive test,
# but it should be enough to check whether the handler is removing dot segments
ws = validate_and_send(rh, Request(f'{self.ws_base_url}/a/b/./../../test'))
ws = ws_validate_and_send(rh, Request(f'{self.ws_base_url}/a/b/./../../test'))
assert ws.status == 101
ws.send('path')
assert ws.recv() == '/test'
@ -187,7 +200,7 @@ def test_remove_dot_segments(self, handler):
def test_raise_http_error(self, handler, status):
with handler() as rh:
with pytest.raises(HTTPError) as exc_info:
validate_and_send(rh, Request(f'{self.ws_base_url}/gen_{status}'))
ws_validate_and_send(rh, Request(f'{self.ws_base_url}/gen_{status}'))
assert exc_info.value.status == status
@pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
@ -198,7 +211,7 @@ def test_raise_http_error(self, handler, status):
def test_timeout(self, handler, params, extensions):
with handler(**params) as rh:
with pytest.raises(TransportError):
validate_and_send(rh, Request(self.ws_base_url, extensions=extensions))
ws_validate_and_send(rh, Request(self.ws_base_url, extensions=extensions))
@pytest.mark.parametrize('handler', ['Websockets'], indirect=True)
def test_cookies(self, handler):
@ -210,18 +223,18 @@ def test_cookies(self, handler):
comment_url=None, rest={}))
with handler(cookiejar=cookiejar) as rh:
ws = validate_and_send(rh, Request(self.ws_base_url))
ws = ws_validate_and_send(rh, Request(self.ws_base_url))
ws.send('headers')
assert json.loads(ws.recv())['cookie'] == 'test=ytdlp'
ws.close()
with handler() as rh:
ws = validate_and_send(rh, Request(self.ws_base_url))
ws = ws_validate_and_send(rh, Request(self.ws_base_url))
ws.send('headers')
assert 'cookie' not in json.loads(ws.recv())
ws.close()
ws = validate_and_send(rh, Request(self.ws_base_url, extensions={'cookiejar': cookiejar}))
ws = ws_validate_and_send(rh, Request(self.ws_base_url, extensions={'cookiejar': cookiejar}))
ws.send('headers')
assert json.loads(ws.recv())['cookie'] == 'test=ytdlp'
ws.close()
@ -231,7 +244,7 @@ def test_source_address(self, handler):
source_address = f'127.0.0.{random.randint(5, 255)}'
verify_address_availability(source_address)
with handler(source_address=source_address) as rh:
ws = validate_and_send(rh, Request(self.ws_base_url))
ws = ws_validate_and_send(rh, Request(self.ws_base_url))
ws.send('source_address')
assert source_address == ws.recv()
ws.close()
@ -240,7 +253,7 @@ def test_source_address(self, handler):
def test_response_url(self, handler):
with handler() as rh:
url = f'{self.ws_base_url}/something'
ws = validate_and_send(rh, Request(url))
ws = ws_validate_and_send(rh, Request(url))
assert ws.url == url
ws.close()
@ -248,14 +261,14 @@ def test_response_url(self, handler):
def test_request_headers(self, handler):
with handler(headers=HTTPHeaderDict({'test1': 'test', 'test2': 'test2'})) as rh:
# Global Headers
ws = validate_and_send(rh, Request(self.ws_base_url))
ws = ws_validate_and_send(rh, Request(self.ws_base_url))
ws.send('headers')
headers = HTTPHeaderDict(json.loads(ws.recv()))
assert headers['test1'] == 'test'
ws.close()
# Per request headers, merged with global
ws = validate_and_send(rh, Request(
ws = ws_validate_and_send(rh, Request(
self.ws_base_url, headers={'test2': 'changed', 'test3': 'test3'}))
ws.send('headers')
headers = HTTPHeaderDict(json.loads(ws.recv()))
@ -288,7 +301,7 @@ def test_mtls(self, handler, client_cert):
verify=False,
client_cert=client_cert
) as rh:
validate_and_send(rh, Request(self.mtls_wss_base_url)).close()
ws_validate_and_send(rh, Request(self.mtls_wss_base_url)).close()
def create_fake_ws_connection(raised):

43
yt_dlp/YoutubeDL.py
View file

@ -42,6 +42,7 @@
SSLError,
network_exceptions,
)
from .networking.impersonate import ImpersonateRequestHandler
from .plugins import directories as plugin_directories
from .postprocessor import _PLUGIN_CLASSES as plugin_pps
from .postprocessor import (
@ -99,6 +100,7 @@
SameFileError,
UnavailableVideoError,
UserNotLive,
YoutubeDLError,
age_restricted,
args_to_str,
bug_reports_message,
@ -402,6 +404,8 @@ class YoutubeDL:
- "detect_or_warn": check whether we can do anything
about it, warn otherwise (default)
source_address: Client-side IP address to bind to.
impersonate: Client to impersonate for requests.
An ImpersonateTarget (from yt_dlp.networking.impersonate)
sleep_interval_requests: Number of seconds to sleep between requests
during extraction
sleep_interval: Number of seconds to sleep before each download when
@ -713,6 +717,13 @@ def check_deprecated(param, option, suggestion):
for msg in self.params.get('_deprecation_warnings', []):
self.deprecated_feature(msg)
if impersonate_target := self.params.get('impersonate'):
if not self._impersonate_target_available(impersonate_target):
raise YoutubeDLError(
f'Impersonate target "{impersonate_target}" is not available. '
f'Use --list-impersonate-targets to see available targets. '
f'You may be missing dependencies required to support this target.')
if 'list-formats' in self.params['compat_opts']:
self.params['listformats_table'] = False
@ -4077,6 +4088,22 @@ def _opener(self):
handler = self._request_director.handlers['Urllib']
return handler._get_instance(cookiejar=self.cookiejar, proxies=self.proxies)
def _get_available_impersonate_targets(self):
# todo(future): make available as public API
return [
(target, rh.RH_NAME)
for rh in self._request_director.handlers.values()
if isinstance(rh, ImpersonateRequestHandler)
for target in rh.supported_targets
]
def _impersonate_target_available(self, target):
# todo(future): make available as public API
return any(
rh.is_supported_target(target)
for rh in self._request_director.handlers.values()
if isinstance(rh, ImpersonateRequestHandler))
def urlopen(self, req):
""" Start an HTTP download """
if isinstance(req, str):
@ -4108,9 +4135,13 @@ def urlopen(self, req):
raise RequestError(
'file:// URLs are disabled by default in yt-dlp for security reasons. '
'Use --enable-file-urls to enable at your own risk.', cause=ue) from ue
if 'unsupported proxy type: "https"' in ue.msg.lower():
if (
'unsupported proxy type: "https"' in ue.msg.lower()
and 'requests' not in self._request_director.handlers
and 'curl_cffi' not in self._request_director.handlers
):
raise RequestError(
'To use an HTTPS proxy for this request, one of the following dependencies needs to be installed: requests')
'To use an HTTPS proxy for this request, one of the following dependencies needs to be installed: requests, curl_cffi')
elif (
re.match(r'unsupported url scheme: "wss?"', ue.msg.lower())
@ -4120,6 +4151,13 @@ def urlopen(self, req):
'This request requires WebSocket support. '
'Ensure one of the following dependencies are installed: websockets',
cause=ue) from ue
elif re.match(r'unsupported (?:extensions: impersonate|impersonate target)', ue.msg.lower()):
raise RequestError(
f'Impersonate target "{req.extensions["impersonate"]}" is not available.'
f' See --list-impersonate-targets for available targets.'
f' This request requires browser impersonation, however you may be missing dependencies'
f' required to support this target.')
raise
except SSLError as e:
if 'UNSAFE_LEGACY_RENEGOTIATION_DISABLED' in str(e):
@ -4152,6 +4190,7 @@ def build_request_director(self, handlers, preferences=None):
'timeout': 'socket_timeout',
'legacy_ssl_support': 'legacyserverconnect',
'enable_file_urls': 'enable_file_urls',
'impersonate': 'impersonate',
'client_cert': {
'client_certificate': 'client_certificate',
'client_certificate_key': 'client_certificate_key',

41
yt_dlp/__init__.py
View file

@ -19,6 +19,7 @@
from .downloader.external import get_external_downloader
from .extractor import list_extractor_classes
from .extractor.adobepass import MSO_INFO
from .networking.impersonate import ImpersonateTarget
from .options import parseOpts
from .postprocessor import (
FFmpegExtractAudioPP,
@ -48,6 +49,7 @@
float_or_none,
format_field,
int_or_none,
join_nonempty,
match_filter_func,
parse_bytes,
parse_duration,
@ -388,6 +390,9 @@ def parse_chapters(name, value, advanced=False):
f'Supported keyrings are: {", ".join(sorted(SUPPORTED_KEYRINGS))}')
opts.cookiesfrombrowser = (browser_name, profile, keyring, container)
if opts.impersonate is not None:
opts.impersonate = ImpersonateTarget.from_str(opts.impersonate.lower())
# MetadataParser
def metadataparser_actions(f):
if isinstance(f, str):
@ -911,6 +916,7 @@ def parse_options(argv=None):
'postprocessors': postprocessors,
'fixup': opts.fixup,
'source_address': opts.source_address,
'impersonate': opts.impersonate,
'call_home': opts.call_home,
'sleep_interval_requests': opts.sleep_interval_requests,
'sleep_interval': opts.sleep_interval,
@ -980,6 +986,41 @@ def _real_main(argv=None):
traceback.print_exc()
ydl._download_retcode = 100
if opts.list_impersonate_targets:
known_targets = [
# List of simplified targets we know are supported,
# to help users know what dependencies may be required.
(ImpersonateTarget('chrome'), 'curl_cffi'),
(ImpersonateTarget('edge'), 'curl_cffi'),
(ImpersonateTarget('safari'), 'curl_cffi'),
]
available_targets = ydl._get_available_impersonate_targets()
def make_row(target, handler):
return [
join_nonempty(target.client.title(), target.version, delim='-') or '-',
join_nonempty((target.os or "").title(), target.os_version, delim='-') or '-',
handler,
]
rows = [make_row(target, handler) for target, handler in available_targets]
for known_target, known_handler in known_targets:
if not any(
known_target in target and handler == known_handler
for target, handler in available_targets
):
rows.append([
ydl._format_out(text, ydl.Styles.SUPPRESS)
for text in make_row(known_target, f'{known_handler} (not available)')
])
ydl.to_screen('[info] Available impersonate targets')
ydl.to_stdout(render_table(['Client', 'OS', 'Source'], rows, extra_gap=2, delim='-'))
return
if not actual_use:
if pre_process:
return ydl._download_retcode

6
yt_dlp/__pyinstaller/hook-yt_dlp.py
View file

@ -1,6 +1,6 @@
import sys
from PyInstaller.utils.hooks import collect_submodules
from PyInstaller.utils.hooks import collect_submodules, collect_data_files
def pycryptodome_module():
@ -25,10 +25,12 @@ def get_hidden_imports():
for module in ('websockets', 'requests', 'urllib3'):
yield from collect_submodules(module)
# These are auto-detected, but explicitly add them just in case
yield from ('mutagen', 'brotli', 'certifi', 'secretstorage')
yield from ('mutagen', 'brotli', 'certifi', 'secretstorage', 'curl_cffi')
hiddenimports = list(get_hidden_imports())
print(f'Adding imports: {hiddenimports}')
excludedimports = ['youtube_dl', 'youtube_dlc', 'test', 'ytdlp_plugins', 'devscripts', 'bundle']
datas = collect_data_files('curl_cffi', includes=['cacert.pem'])

4
yt_dlp/dependencies/__init__.py
View file

@ -74,6 +74,10 @@
if hasattr(xattr, 'set'): # pyxattr
xattr._yt_dlp__identifier = 'pyxattr'
try:
import curl_cffi
except ImportError:
curl_cffi = None
from . import Cryptodome

18
yt_dlp/extractor/imgur.py
View file

@ -76,6 +76,23 @@ class ImgurIE(ImgurBaseIE):
'thumbnail': 'https://i.imgur.com/jxBXAMCh.jpg',
'dislike_count': int,
},
}, {
# needs Accept header, ref: https://github.com/yt-dlp/yt-dlp/issues/9458
'url': 'https://imgur.com/zV03bd5',
'md5': '59df97884e8ba76143ff6b640a0e2904',
'info_dict': {
'id': 'zV03bd5',
'ext': 'mp4',
'title': 'Ive - Liz',
'timestamp': 1710491255,
'upload_date': '20240315',
'like_count': int,
'dislike_count': int,
'duration': 56.92,
'comment_count': int,
'release_timestamp': 1710491255,
'release_date': '20240315',
},
}]
def _real_extract(self, url):
@ -192,6 +209,7 @@ def og_get_size(media_type):
'id': video_id,
'formats': formats,
'thumbnail': url_or_none(search('thumbnailUrl')),
'http_headers': {'Accept': '*/*'},
}

7
yt_dlp/networking/__init__.py
View file

@ -28,3 +28,10 @@
pass
except Exception as e:
warnings.warn(f'Failed to import "websockets" request handler: {e}' + bug_reports_message())
try:
from . import _curlcffi # noqa: F401
except ImportError:
pass
except Exception as e:
warnings.warn(f'Failed to import "curl_cffi" request handler: {e}' + bug_reports_message())

221
yt_dlp/networking/_curlcffi.py Normal file
View file

@ -0,0 +1,221 @@
from __future__ import annotations
import io
import math
import urllib.parse
from ._helper import InstanceStoreMixin, select_proxy
from .common import (
Features,
Request,
Response,
register_preference,
register_rh,
)
from .exceptions import (
CertificateVerifyError,
HTTPError,
IncompleteRead,
ProxyError,
SSLError,
TransportError,
)
from .impersonate import ImpersonateRequestHandler, ImpersonateTarget
from ..dependencies import curl_cffi
from ..utils import int_or_none
if curl_cffi is None:
raise ImportError('curl_cffi is not installed')
curl_cffi_version = tuple(int_or_none(x, default=0) for x in curl_cffi.__version__.split('.'))
if curl_cffi_version != (0, 5, 10):
curl_cffi._yt_dlp__version = f'{curl_cffi.__version__} (unsupported)'
raise ImportError('Only curl_cffi 0.5.10 is supported')
import curl_cffi.requests
from curl_cffi.const import CurlECode, CurlOpt
class CurlCFFIResponseReader(io.IOBase):
def __init__(self, response: curl_cffi.requests.Response):
self._response = response
self._iterator = response.iter_content()
self._buffer = b''
self.bytes_read = 0
def readable(self):
return True
def read(self, size=None):
exception_raised = True
try:
while self._iterator and (size is None or len(self._buffer) < size):
chunk = next(self._iterator, None)
if chunk is None:
self._iterator = None
break
self._buffer += chunk
self.bytes_read += len(chunk)
if size is None:
size = len(self._buffer)
data = self._buffer[:size]
self._buffer = self._buffer[size:]
# "free" the curl instance if the response is fully read.
# curl_cffi doesn't do this automatically and only allows one open response per thread
if not self._iterator and not self._buffer:
self.close()
exception_raised = False
return data
finally:
if exception_raised:
self.close()
def close(self):
if not self.closed:
self._response.close()
self._buffer = b''
super().close()
class CurlCFFIResponseAdapter(Response):
fp: CurlCFFIResponseReader
def __init__(self, response: curl_cffi.requests.Response):
super().__init__(
fp=CurlCFFIResponseReader(response),
headers=response.headers,
url=response.url,
status=response.status_code)
def read(self, amt=None):
try:
return self.fp.read(amt)
except curl_cffi.requests.errors.RequestsError as e:
if e.code == CurlECode.PARTIAL_FILE:
content_length = int_or_none(e.response.headers.get('Content-Length'))
raise IncompleteRead(
partial=self.fp.bytes_read,
expected=content_length - self.fp.bytes_read if content_length is not None else None,
cause=e) from e
raise TransportError(cause=e) from e
@register_rh
class CurlCFFIRH(ImpersonateRequestHandler, InstanceStoreMixin):
RH_NAME = 'curl_cffi'
_SUPPORTED_URL_SCHEMES = ('http', 'https')
_SUPPORTED_FEATURES = (Features.NO_PROXY, Features.ALL_PROXY)
_SUPPORTED_PROXY_SCHEMES = ('http', 'https', 'socks4', 'socks4a', 'socks5', 'socks5h')
_SUPPORTED_IMPERSONATE_TARGET_MAP = {
ImpersonateTarget('chrome', '110', 'windows', '10'): curl_cffi.requests.BrowserType.chrome110,
ImpersonateTarget('chrome', '107', 'windows', '10'): curl_cffi.requests.BrowserType.chrome107,
ImpersonateTarget('chrome', '104', 'windows', '10'): curl_cffi.requests.BrowserType.chrome104,
ImpersonateTarget('chrome', '101', 'windows', '10'): curl_cffi.requests.BrowserType.chrome101,
ImpersonateTarget('chrome', '100', 'windows', '10'): curl_cffi.requests.BrowserType.chrome100,
ImpersonateTarget('chrome', '99', 'windows', '10'): curl_cffi.requests.BrowserType.chrome99,
ImpersonateTarget('edge', '101', 'windows', '10'): curl_cffi.requests.BrowserType.edge101,
ImpersonateTarget('edge', '99', 'windows', '10'): curl_cffi.requests.BrowserType.edge99,
ImpersonateTarget('safari', '15.5', 'macos', '12'): curl_cffi.requests.BrowserType.safari15_5,
ImpersonateTarget('safari', '15.3', 'macos', '11'): curl_cffi.requests.BrowserType.safari15_3,
ImpersonateTarget('chrome', '99', 'android', '12'): curl_cffi.requests.BrowserType.chrome99_android,
}
def _create_instance(self, cookiejar=None):
return curl_cffi.requests.Session(cookies=cookiejar)
def _check_extensions(self, extensions):
super()._check_extensions(extensions)
extensions.pop('impersonate', None)
extensions.pop('cookiejar', None)
extensions.pop('timeout', None)
def _send(self, request: Request):
max_redirects_exceeded = False
session: curl_cffi.requests.Session = self._get_instance(
cookiejar=self._get_cookiejar(request) if 'cookie' not in request.headers else None)
if self.verbose:
session.curl.setopt(CurlOpt.VERBOSE, 1)
proxies = self._get_proxies(request)
if 'no' in proxies:
session.curl.setopt(CurlOpt.NOPROXY, proxies['no'])
proxies.pop('no', None)
# curl doesn't support per protocol proxies, so we select the one that matches the request protocol
proxy = select_proxy(request.url, proxies=proxies)
if proxy:
session.curl.setopt(CurlOpt.PROXY, proxy)
scheme = urllib.parse.urlparse(request.url).scheme.lower()
if scheme != 'http':
# Enable HTTP CONNECT for HTTPS urls.
# Don't use CONNECT for http for compatibility with urllib behaviour.
# See: https://curl.se/libcurl/c/CURLOPT_HTTPPROXYTUNNEL.html
session.curl.setopt(CurlOpt.HTTPPROXYTUNNEL, 1)
headers = self._get_impersonate_headers(request)
if self._client_cert:
session.curl.setopt(CurlOpt.SSLCERT, self._client_cert['client_certificate'])
client_certificate_key = self._client_cert.get('client_certificate_key')
client_certificate_password = self._client_cert.get('client_certificate_password')
if client_certificate_key:
session.curl.setopt(CurlOpt.SSLKEY, client_certificate_key)
if client_certificate_password:
session.curl.setopt(CurlOpt.KEYPASSWD, client_certificate_password)
timeout = self._calculate_timeout(request)
# set CURLOPT_LOW_SPEED_LIMIT and CURLOPT_LOW_SPEED_TIME to act as a read timeout. [1]
# curl_cffi does not currently do this. [2]
# Note: CURLOPT_LOW_SPEED_TIME is in seconds, so we need to round up to the nearest second. [3]
# [1] https://unix.stackexchange.com/a/305311
# [2] https://github.com/yifeikong/curl_cffi/issues/156
# [3] https://curl.se/libcurl/c/CURLOPT_LOW_SPEED_TIME.html
session.curl.setopt(CurlOpt.LOW_SPEED_LIMIT, 1) # 1 byte per second
session.curl.setopt(CurlOpt.LOW_SPEED_TIME, math.ceil(timeout))
try:
curl_response = session.request(
method=request.method,
url=request.url,
headers=headers,
data=request.data,
verify=self.verify,
max_redirects=5,
timeout=timeout,
impersonate=self._SUPPORTED_IMPERSONATE_TARGET_MAP.get(
self._get_request_target(request)),
interface=self.source_address,
stream=True
)
except curl_cffi.requests.errors.RequestsError as e:
if e.code == CurlECode.PEER_FAILED_VERIFICATION:
raise CertificateVerifyError(cause=e) from e
elif e.code == CurlECode.SSL_CONNECT_ERROR:
raise SSLError(cause=e) from e
elif e.code == CurlECode.TOO_MANY_REDIRECTS:
max_redirects_exceeded = True
curl_response = e.response
elif e.code == CurlECode.PROXY:
raise ProxyError(cause=e) from e
else:
raise TransportError(cause=e) from e
response = CurlCFFIResponseAdapter(curl_response)
if not 200 <= response.status < 300:
raise HTTPError(response, redirect_loop=max_redirects_exceeded)
return response
@register_preference(CurlCFFIRH)
def curl_cffi_preference(rh, request):
return -100

7
yt_dlp/networking/_requests.py
View file

@ -307,8 +307,7 @@ def _send(self, request):
max_redirects_exceeded = False
session = self._get_instance(
cookiejar=request.extensions.get('cookiejar') or self.cookiejar)
session = self._get_instance(cookiejar=self._get_cookiejar(request))
try:
requests_res = session.request(
@ -316,8 +315,8 @@ def _send(self, request):
url=request.url,
data=request.data,
headers=headers,
timeout=float(request.extensions.get('timeout') or self.timeout),
proxies=request.proxies or self.proxies,
timeout=self._calculate_timeout(request),
proxies=self._get_proxies(request),
allow_redirects=True,
stream=True
)

6
yt_dlp/networking/_urllib.py
View file

@ -389,11 +389,11 @@ def _send(self, request):
)
opener = self._get_instance(
proxies=request.proxies or self.proxies,
cookiejar=request.extensions.get('cookiejar') or self.cookiejar
proxies=self._get_proxies(request),
cookiejar=self._get_cookiejar(request)
)
try:
res = opener.open(urllib_req, timeout=float(request.extensions.get('timeout') or self.timeout))
res = opener.open(urllib_req, timeout=self._calculate_timeout(request))
except urllib.error.HTTPError as e:
if isinstance(e.fp, (http.client.HTTPResponse, urllib.response.addinfourl)):
# Prevent file object from being closed when urllib.error.HTTPError is destroyed.

6
yt_dlp/networking/_websockets.py
View file

@ -112,10 +112,10 @@ def close(self):
logging.getLogger(name).removeHandler(handler)
def _send(self, request):
timeout = float(request.extensions.get('timeout') or self.timeout)
timeout = self._calculate_timeout(request)
headers = self._merge_headers(request.headers)
if 'cookie' not in headers:
cookiejar = request.extensions.get('cookiejar') or self.cookiejar
cookiejar = self._get_cookiejar(request)
cookie_header = cookiejar.get_cookie_header(request.url)
if cookie_header:
headers['cookie'] = cookie_header
@ -125,7 +125,7 @@ def _send(self, request):
'source_address': (self.source_address, 0) if self.source_address else None,
'timeout': timeout
}
proxy = select_proxy(request.url, request.proxies or self.proxies or {})
proxy = select_proxy(request.url, self._get_proxies(request))
try:
if proxy:
socks_proxy_options = make_socks_proxy_opts(proxy)

11
yt_dlp/networking/common.py
View file

@ -256,6 +256,15 @@ def _make_sslcontext(self):
def _merge_headers(self, request_headers):
return HTTPHeaderDict(self.headers, request_headers)
def _calculate_timeout(self, request):
return float(request.extensions.get('timeout') or self.timeout)
def _get_cookiejar(self, request):
return request.extensions.get('cookiejar') or self.cookiejar
def _get_proxies(self, request):
return (request.proxies or self.proxies).copy()
def _check_url_scheme(self, request: Request):
scheme = urllib.parse.urlparse(request.url).scheme.lower()
if self._SUPPORTED_URL_SCHEMES is not None and scheme not in self._SUPPORTED_URL_SCHEMES:
@ -491,7 +500,7 @@ class Response(io.IOBase):
def __init__(
self,
fp: typing.IO,
fp: io.IOBase,
url: str,
headers: Mapping[str, str],
status: int = 200,

141
yt_dlp/networking/impersonate.py Normal file
View file

@ -0,0 +1,141 @@
from __future__ import annotations
import re
from abc import ABC
from dataclasses import dataclass
from typing import Any
from .common import RequestHandler, register_preference
from .exceptions import UnsupportedRequest
from ..compat.types import NoneType
from ..utils import classproperty, join_nonempty
from ..utils.networking import std_headers
@dataclass(order=True, frozen=True)
class ImpersonateTarget:
"""
A target for browser impersonation.
Parameters:
@param client: the client to impersonate
@param version: the client version to impersonate
@param os: the client OS to impersonate
@param os_version: the client OS version to impersonate
Note: None is used to indicate to match any.
"""
client: str | None = None
version: str | None = None
os: str | None = None
os_version: str | None = None
def __post_init__(self):
if self.version and not self.client:
raise ValueError('client is required if version is set')
if self.os_version and not self.os:
raise ValueError('os is required if os_version is set')
def __contains__(self, target: ImpersonateTarget):
if not isinstance(target, ImpersonateTarget):
return False
return (
(self.client is None or target.client is None or self.client == target.client)
and (self.version is None or target.version is None or self.version == target.version)
and (self.os is None or target.os is None or self.os == target.os)
and (self.os_version is None or target.os_version is None or self.os_version == target.os_version)
)
def __str__(self):
return f'{join_nonempty(self.client, self.version)}:{join_nonempty(self.os, self.os_version)}'.rstrip(':')
@classmethod
def from_str(cls, target: str):
mobj = re.fullmatch(r'(?:(?P<client>[^:-]+)(?:-(?P<version>[^:-]+))?)?(?::(?:(?P<os>[^:-]+)(?:-(?P<os_version>[^:-]+))?)?)?', target)
if not mobj:
raise ValueError(f'Invalid impersonate target "{target}"')
return cls(**mobj.groupdict())
class ImpersonateRequestHandler(RequestHandler, ABC):
"""
Base class for request handlers that support browser impersonation.
This provides a method for checking the validity of the impersonate extension,
which can be used in _check_extensions.
Impersonate targets consist of a client, version, os and os_ver.
See the ImpersonateTarget class for more details.
The following may be defined:
- `_SUPPORTED_IMPERSONATE_TARGET_MAP`: a dict mapping supported targets to custom object.
Any Request with an impersonate target not in this list will raise an UnsupportedRequest.
Set to None to disable this check.
Note: Entries are in order of preference
Parameters:
@param impersonate: the default impersonate target to use for requests.
Set to None to disable impersonation.
"""
_SUPPORTED_IMPERSONATE_TARGET_MAP: dict[ImpersonateTarget, Any] = {}
def __init__(self, *, impersonate: ImpersonateTarget = None, **kwargs):
super().__init__(**kwargs)
self.impersonate = impersonate
def _check_impersonate_target(self, target: ImpersonateTarget):
assert isinstance(target, (ImpersonateTarget, NoneType))
if target is None or not self.supported_targets:
return
if not self.is_supported_target(target):
raise UnsupportedRequest(f'Unsupported impersonate target: {target}')
def _check_extensions(self, extensions):
super()._check_extensions(extensions)
if 'impersonate' in extensions:
self._check_impersonate_target(extensions.get('impersonate'))
def _validate(self, request):
super()._validate(request)
self._check_impersonate_target(self.impersonate)
def _resolve_target(self, target: ImpersonateTarget | None):
"""Resolve a target to a supported target."""
if target is None:
return
for supported_target in self.supported_targets:
if target in supported_target:
if self.verbose:
self._logger.stdout(
f'{self.RH_NAME}: resolved impersonate target {target} to {supported_target}')
return supported_target
@classproperty
def supported_targets(self) -> tuple[ImpersonateTarget, ...]:
return tuple(self._SUPPORTED_IMPERSONATE_TARGET_MAP.keys())
def is_supported_target(self, target: ImpersonateTarget):
assert isinstance(target, ImpersonateTarget)
return self._resolve_target(target) is not None
def _get_request_target(self, request):
"""Get the requested target for the request"""
return self._resolve_target(request.extensions.get('impersonate') or self.impersonate)
def _get_impersonate_headers(self, request):
headers = self._merge_headers(request.headers)
if self._get_request_target(request) is not None:
# remove all headers present in std_headers
# todo: change this to not depend on std_headers
for k, v in std_headers.items():
if headers.get(k) == v:
headers.pop(k)
return headers
@register_preference(ImpersonateRequestHandler)
def impersonate_preference(rh, request):
if request.extensions.get('impersonate') or rh.impersonate:
return 1000
return 0

12
yt_dlp/options.py
View file

@ -515,6 +515,18 @@ def _alias_callback(option, opt_str, value, parser, opts, nargs):
metavar='IP', dest='source_address', default=None,
help='Client-side IP address to bind to',
)
network.add_option(
'--impersonate',
metavar='CLIENT[:OS]', dest='impersonate', default=None,
help=(
'Client to impersonate for requests. E.g. chrome, chrome-110, chrome:windows-10. '
'Pass --impersonate="" to impersonate any client.'),
)
network.add_option(
'--list-impersonate-targets',
dest='list_impersonate_targets', default=False, action='store_true',
help='List available clients to impersonate.',
)
network.add_option(
'-4', '--force-ipv4',
action='store_const', const='0.0.0.0', dest='source_address',