def descramble(self) -> None:
"""Descramble the stream data and build Stream instances.
The initialization process takes advantage of Python's
"call-by-reference evaluation," which allows dictionary transforms to
be applied in-place, instead of holding references to mutations at each
interstitial step.
:rtype: None
"""
logger.info("init started")
self.vid_info = dict(parse_qsl(self.vid_info_raw))
if self.age_restricted:
self.player_config_args = self.vid_info
else:
assert self.watch_html is not None
self.player_config_args = get_ytplayer_config(self.watch_html)["args"]
# Fix for KeyError: 'title' issue #434
if "title" not in self.player_config_args: # type: ignore
i_start = self.watch_html.lower().index("<title>") + len("<title>")
i_end = self.watch_html.lower().index("</title>")
title = self.watch_html[i_start:i_end].strip()
index = title.lower().rfind(" - youtube")
title = title[:index] if index > 0 else title
self.player_config_args["title"] = unescape(title)
# https://github.com/nficano/pytube/issues/165
stream_maps = ["url_encoded_fmt_stream_map"]
if "adaptive_fmts" in self.player_config_args:
stream_maps.append("adaptive_fmts")
# unscramble the progressive and adaptive stream manifests.
for fmt in stream_maps:
if not self.age_restricted and fmt in self.vid_info:
apply_descrambler(self.vid_info, fmt)
apply_descrambler(self.player_config_args, fmt)
if not self.js:
if not self.embed_html:
self.embed_html = request.get(url=self.embed_url)
self.js_url = extract.js_url(self.embed_html)
self.js = request.get(self.js_url)
apply_signature(self.player_config_args, fmt, self.js)
# build instances of :class:`Stream <Stream>`
self.initialize_stream_objects(fmt)
# load the player_response object (contains subtitle information)
self.player_response = json.loads(self.player_config_args["player_response"])
del self.player_config_args["player_response"]
self.stream_monostate.title = self.title
self.stream_monostate.duration = self.length
logger.info("init finished successfully")
def descramble(self) -> None:
"""Descramble the stream data and build Stream instances.
The initialization process takes advantage of Python's
"call-by-reference evaluation," which allows dictionary transforms to
be applied in-place, instead of holding references to mutations at each
interstitial step.
:rtype: None
"""
self.vid_info = dict(parse_qsl(self.vid_info_raw))
self.player_config_args = self.vid_info
self.player_response = json.loads(self.vid_info['player_response'])
# On pre-signed videos, we need to use get_ytplayer_config to fix
# the player_response item
if 'streamingData' not in self.player_config_args['player_response']:
config_response = get_ytplayer_config(self.watch_html)
if 'args' in config_response:
self.player_config_args['player_response'] = config_response['args']['player_response'] # noqa: E501
else:
self.player_config_args['player_response'] = config_response
# https://github.com/nficano/pytube/issues/165
stream_maps = ["url_encoded_fmt_stream_map"]
if "adaptive_fmts" in self.player_config_args:
stream_maps.append("adaptive_fmts")
# unscramble the progressive and adaptive stream manifests.
for fmt in stream_maps:
if not self.age_restricted and fmt in self.vid_info:
apply_descrambler(self.vid_info, fmt)
apply_descrambler(self.player_config_args, fmt)
if not self.js:
if not self.embed_html:
self.embed_html = request.get(url=self.embed_url)
self.js_url = extract.js_url(self.embed_html)
self.js = request.get(self.js_url)
apply_signature(self.player_config_args, fmt, self.js)
# build instances of :class:`Stream <Stream>`
self.initialize_stream_objects(fmt)
# load the player_response object (contains subtitle information)
if isinstance(self.player_config_args["player_response"], str):
self.player_response = json.loads(
self.player_config_args["player_response"]
)
else:
self.player_response = self.player_config_args["player_response"]
del self.player_config_args["player_response"]
self.stream_monostate.title = self.title
self.stream_monostate.duration = self.length
def fmt_streams(self):
"""Returns a list of streams if they have been initialized.
If the streams have not been initialized, finds all relevant
streams and initializes them.
"""
self.check_availability()
if self._fmt_streams:
return self._fmt_streams
self._fmt_streams = []
# https://github.com/pytube/pytube/issues/165
stream_maps = ["url_encoded_fmt_stream_map"]
if "adaptive_fmts" in self.player_config_args:
stream_maps.append("adaptive_fmts")
# unscramble the progressive and adaptive stream manifests.
for fmt in stream_maps:
if not self.age_restricted and fmt in self.vid_info:
extract.apply_descrambler(self.vid_info, fmt)
extract.apply_descrambler(self.player_config_args, fmt)
# If the cached js doesn't work, try fetching a new js file
# https://github.com/pytube/pytube/issues/1054
try:
extract.apply_signature(self.player_config_args, fmt, self.js)
except exceptions.ExtractError:
# To force an update to the js file, we clear the cache and retry
self._js = None
self._js_url = None
pytube.__js__ = None
pytube.__js_url__ = None
extract.apply_signature(self.player_config_args, fmt, self.js)
# build instances of :class:`Stream <Stream>`
# Initialize stream objects
stream_manifest = self.player_config_args[fmt]
for stream in stream_manifest:
video = Stream(
stream=stream,
player_config_args=self.player_config_args,
monostate=self.stream_monostate,
)
self._fmt_streams.append(video)
self.stream_monostate.title = self.title
self.stream_monostate.duration = self.length
return self._fmt_streams
def _decipher(self, retry: bool = False):
if not pytube.__js__ or retry:
self._getJS()
try:
'''
These two are the main methods being used from PyTube.
Used to decipher the stream URLs using player JavaScript & the player_response passed from the getStream method of this derieved class.
These methods operate on the value of "player_response" key in dictionary of self._player_response & save _deciphered information in the "url_encoded_fmt_stream_map" key.
'''
apply_descrambler(self._player_response, "url_encoded_fmt_stream_map")
apply_signature(
self._player_response, "url_encoded_fmt_stream_map", pytube.__js__
)
except:
'''
Fetch updated player JavaScript to get new cipher algorithm.
'''
self._decipher(retry = False)
async def fmt_streams(self):
"""Returns a list of streams if they have been initialized.
If the streams have not been initialized, finds all relevant
streams and initializes them.
"""
await self.check_availability()
if self._fmt_streams:
return self._fmt_streams
self._fmt_streams = []
# https://github.com/pytube/pytube/issues/165
stream_maps = ["url_encoded_fmt_stream_map"]
if "adaptive_fmts" in (await self.player_config_args):
stream_maps.append("adaptive_fmts")
# unscramble the progressive and adaptive stream manifests.
for fmt in stream_maps:
if not (await self.age_restricted) and fmt in (await
self.vid_info):
apply_descrambler((await self.vid_info), fmt)
apply_descrambler((await self.player_config_args), fmt)
apply_signature((await self.player_config_args), fmt,
(await self.js))
# build instances of :class:`Stream <Stream>`
# Initialize stream objects
stream_manifest = (await self.player_config_args)[fmt]
for stream in stream_manifest:
video = Stream(
stream=stream,
player_config_args=(await self.player_config_args),
monostate=self.stream_monostate,
session=self._client_session,
)
self._fmt_streams.append(video)
self.stream_monostate.title = await self.title
self.stream_monostate.duration = await self.length
return self._fmt_streams
def fmt_streams(self):
"""Returns a list of streams if they have been initialized.
If the streams have not been initialized, finds all relevant
streams and initializes them.
"""
self.check_availability()
if self._fmt_streams:
return self._fmt_streams
self._fmt_streams = []
stream_manifest = extract.apply_descrambler(self.streaming_data)
# If the cached js doesn't work, try fetching a new js file
# https://github.com/pytube/pytube/issues/1054
try:
extract.apply_signature(stream_manifest, self.vid_info, self.js)
except exceptions.ExtractError:
# To force an update to the js file, we clear the cache and retry
self._js = None
self._js_url = None
pytube.__js__ = None
pytube.__js_url__ = None
extract.apply_signature(stream_manifest, self.vid_info, self.js)
# build instances of :class:`Stream <Stream>`
# Initialize stream objects
for stream in stream_manifest:
video = Stream(
stream=stream,
monostate=self.stream_monostate,
)
self._fmt_streams.append(video)
self.stream_monostate.title = self.title
self.stream_monostate.duration = self.length
return self._fmt_streams
def test_signature_cipher_does_not_error(stream_dict):
config_args = extract.get_ytplayer_config(stream_dict)['args']
extract.apply_descrambler(config_args, "url_encoded_fmt_stream_map")
assert "s" in config_args["url_encoded_fmt_stream_map"][0].keys()
def test_signature_cipher_does_not_error(stream_dict):
extract.apply_descrambler(stream_dict, "url_encoded_fmt_stream_map")
assert "s" in stream_dict["url_encoded_fmt_stream_map"][0].keys()
