mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2025-12-29 11:01:31 +00:00
[ie/youtube] Implement external n/sig solver (#14157)
Closes #14404, Closes #14431, Closes #14680, Closes #14707 Authored by: bashonly, coletdjnz, seproDev, Grub4K Co-authored-by: coletdjnz <coletdjnz@protonmail.com> Co-authored-by: bashonly <bashonly@protonmail.com> Co-authored-by: sepro <sepro@sepr0.com>
This commit is contained in:
60
test/test_jsc/conftest.py
Normal file
60
test/test_jsc/conftest.py
Normal file
@@ -0,0 +1,60 @@
|
||||
import re
|
||||
import pathlib
|
||||
|
||||
import pytest
|
||||
|
||||
import yt_dlp.globals
|
||||
from yt_dlp import YoutubeDL
|
||||
from yt_dlp.extractor.common import InfoExtractor
|
||||
|
||||
|
||||
_TESTDATA_PATH = pathlib.Path(__file__).parent.parent / 'testdata/sigs'
|
||||
_player_re = re.compile(r'^.+/player/(?P<id>[a-zA-Z0-9_/.-]+)\.js$')
|
||||
_player_id_trans = str.maketrans(dict.fromkeys('/.-', '_'))
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def ie() -> InfoExtractor:
|
||||
runtime_names = yt_dlp.globals.supported_js_runtimes.value
|
||||
ydl = YoutubeDL({'js_runtimes': {key: {} for key in runtime_names}})
|
||||
ie = ydl.get_info_extractor('Youtube')
|
||||
|
||||
def _load_player(video_id, player_url, fatal=True):
|
||||
match = _player_re.match(player_url)
|
||||
test_id = match.group('id').translate(_player_id_trans)
|
||||
cached_file = _TESTDATA_PATH / f'player-{test_id}.js'
|
||||
|
||||
if cached_file.exists():
|
||||
return cached_file.read_text()
|
||||
|
||||
if code := ie._download_webpage(player_url, video_id, fatal=fatal):
|
||||
_TESTDATA_PATH.mkdir(exist_ok=True, parents=True)
|
||||
cached_file.write_text(code)
|
||||
return code
|
||||
|
||||
return None
|
||||
|
||||
ie._load_player = _load_player
|
||||
return ie
|
||||
|
||||
|
||||
class MockLogger:
|
||||
def trace(self, message: str):
|
||||
print(f'trace: {message}')
|
||||
|
||||
def debug(self, message: str, *, once=False):
|
||||
print(f'debug: {message}')
|
||||
|
||||
def info(self, message: str):
|
||||
print(f'info: {message}')
|
||||
|
||||
def warning(self, message: str, *, once=False):
|
||||
print(f'warning: {message}')
|
||||
|
||||
def error(self, message: str):
|
||||
print(f'error: {message}')
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def logger():
|
||||
return MockLogger()
|
||||
128
test/test_jsc/test_ejs_integration.py
Normal file
128
test/test_jsc/test_ejs_integration.py
Normal file
@@ -0,0 +1,128 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import dataclasses
|
||||
import enum
|
||||
import importlib.util
|
||||
import json
|
||||
|
||||
import pytest
|
||||
|
||||
from yt_dlp.extractor.youtube.jsc.provider import (
|
||||
JsChallengeRequest,
|
||||
JsChallengeType,
|
||||
JsChallengeProviderResponse,
|
||||
JsChallengeResponse,
|
||||
NChallengeInput,
|
||||
NChallengeOutput,
|
||||
SigChallengeInput,
|
||||
SigChallengeOutput,
|
||||
)
|
||||
from yt_dlp.extractor.youtube.jsc._builtin.bun import BunJCP
|
||||
from yt_dlp.extractor.youtube.jsc._builtin.deno import DenoJCP
|
||||
from yt_dlp.extractor.youtube.jsc._builtin.node import NodeJCP
|
||||
from yt_dlp.extractor.youtube.jsc._builtin.quickjs import QuickJSJCP
|
||||
|
||||
|
||||
_has_ejs = bool(importlib.util.find_spec('yt_dlp_ejs'))
|
||||
pytestmark = pytest.mark.skipif(not _has_ejs, reason='yt-dlp-ejs not available')
|
||||
|
||||
|
||||
class Variant(enum.Enum):
|
||||
main = 'player_ias.vflset/en_US/base.js'
|
||||
tcc = 'player_ias_tcc.vflset/en_US/base.js'
|
||||
tce = 'player_ias_tce.vflset/en_US/base.js'
|
||||
es5 = 'player_es5.vflset/en_US/base.js'
|
||||
es6 = 'player_es6.vflset/en_US/base.js'
|
||||
tv = 'tv-player-ias.vflset/tv-player-ias.js'
|
||||
tv_es6 = 'tv-player-es6.vflset/tv-player-es6.js'
|
||||
phone = 'player-plasma-ias-phone-en_US.vflset/base.js'
|
||||
tablet = 'player-plasma-ias-tablet-en_US.vflset/base.js'
|
||||
|
||||
|
||||
@dataclasses.dataclass
|
||||
class Challenge:
|
||||
player: str
|
||||
variant: Variant
|
||||
type: JsChallengeType
|
||||
values: dict[str, str] = dataclasses.field(default_factory=dict)
|
||||
|
||||
def url(self, /):
|
||||
return f'https://www.youtube.com/s/player/{self.player}/{self.variant.value}'
|
||||
|
||||
|
||||
CHALLENGES: list[Challenge] = [
|
||||
Challenge('3d3ba064', Variant.tce, JsChallengeType.N, {
|
||||
'ZdZIqFPQK-Ty8wId': 'qmtUsIz04xxiNW',
|
||||
'4GMrWHyKI5cEvhDO': 'N9gmEX7YhKTSmw',
|
||||
}),
|
||||
Challenge('3d3ba064', Variant.tce, JsChallengeType.SIG, {
|
||||
'gN7a-hudCuAuPH6fByOk1_GNXN0yNMHShjZXS2VOgsEItAJz0tipeavEOmNdYN-wUtcEqD3bCXjc0iyKfAyZxCBGgIARwsSdQfJ2CJtt':
|
||||
'ttJC2JfQdSswRAIgGBCxZyAfKyi0cjXCb3gqEctUw-NYdNmOEvaepit0zJAtIEsgOV2SXZjhSHMNy0NXNG_1kNyBf6HPuAuCduh-a7O',
|
||||
}),
|
||||
Challenge('5ec65609', Variant.tce, JsChallengeType.N, {
|
||||
'0eRGgQWJGfT5rFHFj': '4SvMpDQH-vBJCw',
|
||||
}),
|
||||
Challenge('5ec65609', Variant.tce, JsChallengeType.SIG, {
|
||||
'AAJAJfQdSswRQIhAMG5SN7-cAFChdrE7tLA6grH0rTMICA1mmDc0HoXgW3CAiAQQ4=CspfaF_vt82XH5yewvqcuEkvzeTsbRuHssRMyJQ=I':
|
||||
'AJfQdSswRQIhAMG5SN7-cAFChdrE7tLA6grI0rTMICA1mmDc0HoXgW3CAiAQQ4HCspfaF_vt82XH5yewvqcuEkvzeTsbRuHssRMyJQ==',
|
||||
}),
|
||||
Challenge('6742b2b9', Variant.tce, JsChallengeType.N, {
|
||||
'_HPB-7GFg1VTkn9u': 'qUAsPryAO_ByYg',
|
||||
'K1t_fcB6phzuq2SF': 'Y7PcOt3VE62mog',
|
||||
}),
|
||||
Challenge('6742b2b9', Variant.tce, JsChallengeType.SIG, {
|
||||
'MMGZJMUucirzS_SnrSPYsc85CJNnTUi6GgR5NKn-znQEICACojE8MHS6S7uYq4TGjQX_D4aPk99hNU6wbTvorvVVMgIARwsSdQfJAA':
|
||||
'AJfQdSswRAIgMVVvrovTbw6UNh99kPa4D_XQjGT4qYu7S6SHM8EjoCACIEQnz-nKN5RgG6iUTnNJC58csYPSrnS_SzricuUMJZGM',
|
||||
}),
|
||||
Challenge('2b83d2e0', Variant.main, JsChallengeType.N, {
|
||||
'0eRGgQWJGfT5rFHFj': 'euHbygrCMLksxd',
|
||||
}),
|
||||
Challenge('2b83d2e0', Variant.main, JsChallengeType.SIG, {
|
||||
'MMGZJMUucirzS_SnrSPYsc85CJNnTUi6GgR5NKn-znQEICACojE8MHS6S7uYq4TGjQX_D4aPk99hNU6wbTvorvVVMgIARwsSdQfJA':
|
||||
'-MGZJMUucirzS_SnrSPYsc85CJNnTUi6GgR5NKnMznQEICACojE8MHS6S7uYq4TGjQX_D4aPk99hNU6wbTvorvVVMgIARwsSdQfJ',
|
||||
}),
|
||||
Challenge('638ec5c6', Variant.main, JsChallengeType.N, {
|
||||
'ZdZIqFPQK-Ty8wId': '1qov8-KM-yH',
|
||||
}),
|
||||
Challenge('638ec5c6', Variant.main, JsChallengeType.SIG, {
|
||||
'gN7a-hudCuAuPH6fByOk1_GNXN0yNMHShjZXS2VOgsEItAJz0tipeavEOmNdYN-wUtcEqD3bCXjc0iyKfAyZxCBGgIARwsSdQfJ2CJtt':
|
||||
'MhudCuAuP-6fByOk1_GNXN7gNHHShjyXS2VOgsEItAJz0tipeav0OmNdYN-wUtcEqD3bCXjc0iyKfAyZxCBGgIARwsSdQfJ2CJtt',
|
||||
}),
|
||||
]
|
||||
|
||||
requests: list[JsChallengeRequest] = []
|
||||
responses: list[JsChallengeProviderResponse] = []
|
||||
for test in CHALLENGES:
|
||||
input_type, output_type = {
|
||||
JsChallengeType.N: (NChallengeInput, NChallengeOutput),
|
||||
JsChallengeType.SIG: (SigChallengeInput, SigChallengeOutput),
|
||||
}[test.type]
|
||||
|
||||
request = JsChallengeRequest(test.type, input_type(test.url(), list(test.values.keys())), test.player)
|
||||
requests.append(request)
|
||||
responses.append(JsChallengeProviderResponse(request, JsChallengeResponse(test.type, output_type(test.values))))
|
||||
|
||||
|
||||
@pytest.fixture(params=[BunJCP, DenoJCP, NodeJCP, QuickJSJCP])
|
||||
def jcp(request, ie, logger):
|
||||
obj = request.param(ie, logger, None)
|
||||
if not obj.is_available():
|
||||
pytest.skip(f'{obj.PROVIDER_NAME} is not available')
|
||||
obj.is_dev = True
|
||||
return obj
|
||||
|
||||
|
||||
@pytest.mark.download
|
||||
def test_bulk_requests(jcp):
|
||||
assert list(jcp.bulk_solve(requests)) == responses
|
||||
|
||||
|
||||
@pytest.mark.download
|
||||
def test_using_cached_player(jcp):
|
||||
first_player_requests = requests[:3]
|
||||
player = jcp._get_player(first_player_requests[0].video_id, first_player_requests[0].input.player_url)
|
||||
initial = json.loads(jcp._run_js_runtime(jcp._construct_stdin(player, False, first_player_requests)))
|
||||
preprocessed = initial.pop('preprocessed_player')
|
||||
result = json.loads(jcp._run_js_runtime(jcp._construct_stdin(preprocessed, True, first_player_requests)))
|
||||
|
||||
assert initial == result
|
||||
194
test/test_jsc/test_provider.py
Normal file
194
test/test_jsc/test_provider.py
Normal file
@@ -0,0 +1,194 @@
|
||||
|
||||
import pytest
|
||||
|
||||
from yt_dlp.extractor.youtube.jsc.provider import (
|
||||
JsChallengeProvider,
|
||||
JsChallengeRequest,
|
||||
JsChallengeProviderResponse,
|
||||
JsChallengeProviderRejectedRequest,
|
||||
JsChallengeType,
|
||||
JsChallengeResponse,
|
||||
NChallengeOutput,
|
||||
NChallengeInput,
|
||||
JsChallengeProviderError,
|
||||
register_provider,
|
||||
register_preference,
|
||||
)
|
||||
from yt_dlp.extractor.youtube.pot._provider import IEContentProvider
|
||||
from yt_dlp.utils import ExtractorError
|
||||
from yt_dlp.extractor.youtube.jsc._registry import _jsc_preferences, _jsc_providers
|
||||
|
||||
|
||||
class ExampleJCP(JsChallengeProvider):
|
||||
PROVIDER_NAME = 'example-provider'
|
||||
PROVIDER_VERSION = '0.0.1'
|
||||
BUG_REPORT_LOCATION = 'https://example.com/issues'
|
||||
|
||||
_SUPPORTED_TYPES = [JsChallengeType.N]
|
||||
|
||||
def is_available(self) -> bool:
|
||||
return True
|
||||
|
||||
def _real_bulk_solve(self, requests):
|
||||
for request in requests:
|
||||
results = dict.fromkeys(request.input.challenges, 'example-solution')
|
||||
response = JsChallengeResponse(
|
||||
type=request.type,
|
||||
output=NChallengeOutput(results=results))
|
||||
yield JsChallengeProviderResponse(request=request, response=response)
|
||||
|
||||
|
||||
PLAYER_URL = 'https://example.com/player.js'
|
||||
|
||||
|
||||
class TestJsChallengeProvider:
|
||||
# note: some test covered in TestPoTokenProvider which shares the same base class
|
||||
def test_base_type(self):
|
||||
assert issubclass(JsChallengeProvider, IEContentProvider)
|
||||
|
||||
def test_create_provider_missing_bulk_solve_method(self, ie, logger):
|
||||
class MissingMethodsJCP(JsChallengeProvider):
|
||||
def is_available(self) -> bool:
|
||||
return True
|
||||
|
||||
with pytest.raises(TypeError, match='bulk_solve'):
|
||||
MissingMethodsJCP(ie=ie, logger=logger, settings={})
|
||||
|
||||
def test_create_provider_missing_available_method(self, ie, logger):
|
||||
class MissingMethodsJCP(JsChallengeProvider):
|
||||
def _real_bulk_solve(self, requests):
|
||||
raise JsChallengeProviderRejectedRequest('Not implemented')
|
||||
|
||||
with pytest.raises(TypeError, match='is_available'):
|
||||
MissingMethodsJCP(ie=ie, logger=logger, settings={})
|
||||
|
||||
def test_barebones_provider(self, ie, logger):
|
||||
class BarebonesProviderJCP(JsChallengeProvider):
|
||||
def is_available(self) -> bool:
|
||||
return True
|
||||
|
||||
def _real_bulk_solve(self, requests):
|
||||
raise JsChallengeProviderRejectedRequest('Not implemented')
|
||||
|
||||
provider = BarebonesProviderJCP(ie=ie, logger=logger, settings={})
|
||||
assert provider.PROVIDER_NAME == 'BarebonesProvider'
|
||||
assert provider.PROVIDER_KEY == 'BarebonesProvider'
|
||||
assert provider.PROVIDER_VERSION == '0.0.0'
|
||||
assert provider.BUG_REPORT_MESSAGE == 'please report this issue to the provider developer at (developer has not provided a bug report location) .'
|
||||
|
||||
def test_example_provider_success(self, ie, logger):
|
||||
provider = ExampleJCP(ie=ie, logger=logger, settings={})
|
||||
|
||||
request = JsChallengeRequest(
|
||||
type=JsChallengeType.N,
|
||||
input=NChallengeInput(player_url=PLAYER_URL, challenges=['example-challenge']))
|
||||
|
||||
request_two = JsChallengeRequest(
|
||||
type=JsChallengeType.N,
|
||||
input=NChallengeInput(player_url=PLAYER_URL, challenges=['example-challenge-2']))
|
||||
|
||||
responses = list(provider.bulk_solve([request, request_two]))
|
||||
assert len(responses) == 2
|
||||
assert all(isinstance(r, JsChallengeProviderResponse) for r in responses)
|
||||
assert responses == [
|
||||
JsChallengeProviderResponse(
|
||||
request=request,
|
||||
response=JsChallengeResponse(
|
||||
type=JsChallengeType.N,
|
||||
output=NChallengeOutput(results={'example-challenge': 'example-solution'}),
|
||||
),
|
||||
),
|
||||
JsChallengeProviderResponse(
|
||||
request=request_two,
|
||||
response=JsChallengeResponse(
|
||||
type=JsChallengeType.N,
|
||||
output=NChallengeOutput(results={'example-challenge-2': 'example-solution'}),
|
||||
),
|
||||
),
|
||||
]
|
||||
|
||||
def test_provider_unsupported_challenge_type(self, ie, logger):
|
||||
provider = ExampleJCP(ie=ie, logger=logger, settings={})
|
||||
request_supported = JsChallengeRequest(
|
||||
type=JsChallengeType.N,
|
||||
input=NChallengeInput(player_url=PLAYER_URL, challenges=['example-challenge']))
|
||||
request_unsupported = JsChallengeRequest(
|
||||
type=JsChallengeType.SIG,
|
||||
input=NChallengeInput(player_url=PLAYER_URL, challenges=['example-challenge']))
|
||||
responses = list(provider.bulk_solve([request_supported, request_unsupported, request_supported]))
|
||||
assert len(responses) == 3
|
||||
# Requests are validated first before continuing to _real_bulk_solve
|
||||
assert isinstance(responses[0], JsChallengeProviderResponse)
|
||||
assert isinstance(responses[0].error, JsChallengeProviderRejectedRequest)
|
||||
assert responses[0].request is request_unsupported
|
||||
assert str(responses[0].error) == 'JS Challenge type "JsChallengeType.SIG" is not supported by example-provider'
|
||||
|
||||
assert responses[1:] == [
|
||||
JsChallengeProviderResponse(
|
||||
request=request_supported,
|
||||
response=JsChallengeResponse(
|
||||
type=JsChallengeType.N,
|
||||
output=NChallengeOutput(results={'example-challenge': 'example-solution'}),
|
||||
),
|
||||
),
|
||||
JsChallengeProviderResponse(
|
||||
request=request_supported,
|
||||
response=JsChallengeResponse(
|
||||
type=JsChallengeType.N,
|
||||
output=NChallengeOutput(results={'example-challenge': 'example-solution'}),
|
||||
),
|
||||
),
|
||||
]
|
||||
|
||||
def test_provider_get_player(self, ie, logger):
|
||||
ie._load_player = lambda video_id, player_url, fatal: (video_id, player_url, fatal)
|
||||
provider = ExampleJCP(ie=ie, logger=logger, settings={})
|
||||
assert provider._get_player('video123', PLAYER_URL) == ('video123', PLAYER_URL, True)
|
||||
|
||||
def test_provider_get_player_error(self, ie, logger):
|
||||
def raise_error(video_id, player_url, fatal):
|
||||
raise ExtractorError('Failed to load player')
|
||||
|
||||
ie._load_player = raise_error
|
||||
provider = ExampleJCP(ie=ie, logger=logger, settings={})
|
||||
with pytest.raises(JsChallengeProviderError, match='Failed to load player for JS challenge'):
|
||||
provider._get_player('video123', PLAYER_URL)
|
||||
|
||||
def test_require_class_end_with_suffix(self, ie, logger):
|
||||
class InvalidSuffix(JsChallengeProvider):
|
||||
PROVIDER_NAME = 'invalid-suffix'
|
||||
|
||||
def _real_bulk_solve(self, requests):
|
||||
raise JsChallengeProviderRejectedRequest('Not implemented')
|
||||
|
||||
def is_available(self) -> bool:
|
||||
return True
|
||||
|
||||
provider = InvalidSuffix(ie=ie, logger=logger, settings={})
|
||||
|
||||
with pytest.raises(AssertionError):
|
||||
provider.PROVIDER_KEY # noqa: B018
|
||||
|
||||
|
||||
def test_register_provider(ie):
|
||||
|
||||
@register_provider
|
||||
class UnavailableProviderJCP(JsChallengeProvider):
|
||||
def is_available(self) -> bool:
|
||||
return False
|
||||
|
||||
def _real_bulk_solve(self, requests):
|
||||
raise JsChallengeProviderRejectedRequest('Not implemented')
|
||||
|
||||
assert _jsc_providers.value.get('UnavailableProvider') == UnavailableProviderJCP
|
||||
_jsc_providers.value.pop('UnavailableProvider')
|
||||
|
||||
|
||||
def test_register_preference(ie):
|
||||
before = len(_jsc_preferences.value)
|
||||
|
||||
@register_preference(ExampleJCP)
|
||||
def unavailable_preference(*args, **kwargs):
|
||||
return 1
|
||||
|
||||
assert len(_jsc_preferences.value) == before + 1
|
||||
Reference in New Issue
Block a user