You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

3856 lines
183 KiB

import base64
import collections
import getpass
import hashlib
import http.client
import http.cookiejar
import http.cookies
import itertools
import json
import math
import netrc
import os
import random
import sys
import time
import urllib.parse
import urllib.request
import xml.etree.ElementTree
from ..compat import functools, re # isort: split
from ..compat import compat_etree_fromstring, compat_expanduser, compat_os_name
from ..downloader import FileDownloader
from ..downloader.f4m import get_base_url, remove_encrypted_media
from ..utils import (
class InfoExtractor:
"""Information Extractor class.
Information extractors are the classes that, given a URL, extract
information about the video (or videos) the URL refers to. This
information includes the real video URL, the video title, author and
others. The information is stored in a dictionary which is then
passed to the YoutubeDL. The YoutubeDL processes this
information possibly downloading the video to the file system, among
other possible outcomes.
The type field determines the type of the result.
By far the most common value (and the default if _type is missing) is
"video", which indicates a single video.
For a video, the dictionaries must include the following fields:
id: Video identifier.
title: Video title, unescaped. Set to an empty string if video has
no title as opposed to "None" which signifies that the
extractor failed to obtain a title
Additionally, it must contain either a formats entry or a url one:
formats: A list of dictionaries for each format available, ordered
from worst to best quality.
Potential fields:
* url The mandatory URL representing the media:
for plain file media - HTTP URL of this file,
for HLS - URL of the M3U8 media playlist,
for HDS - URL of the F4M manifest,
for DASH
- HTTP URL to plain file media (in case of
unfragmented media)
- URL of the MPD manifest or base URL
representing the media if MPD manifest
is parsed from a string (in case of
fragmented media)
for MSS - URL of the ISM manifest.
* manifest_url
The URL of the manifest file in case of
fragmented media:
for HLS - URL of the M3U8 master playlist,
for HDS - URL of the F4M manifest,
for DASH - URL of the MPD manifest,
for MSS - URL of the ISM manifest.
* manifest_stream_number (For internal use only)
The index of the stream in the manifest file
* ext Will be calculated from URL if missing
* format A human-readable description of the format
("mp4 container with h264/opus").
Calculated from the format_id, width, height.
and format_note fields if missing.
* format_id A short description of the format
("mp4_h264_opus" or "19").
Technically optional, but strongly recommended.
* format_note Additional info about the format
("3D" or "DASH video")
* width Width of the video, if known
* height Height of the video, if known
* resolution Textual description of width and height
* dynamic_range The dynamic range of the video. One of:
"SDR" (None), "HDR10", "HDR10+, "HDR12", "HLG, "DV"
* tbr Average bitrate of audio and video in KBit/s
* abr Average audio bitrate in KBit/s
* acodec Name of the audio codec in use
* asr Audio sampling rate in Hertz
* vbr Average video bitrate in KBit/s
* fps Frame rate
* vcodec Name of the video codec in use
* container Name of the container format
* filesize The number of bytes, if known in advance
* filesize_approx An estimate for the number of bytes
* player_url SWF Player URL (used for rtmpdump).
* protocol The protocol that will be used for the actual
download, lower-case. One of "http", "https" or
one of the protocols defined in downloader.PROTOCOL_MAP
* fragment_base_url
Base URL for fragments. Each fragment's path
value (if present) will be relative to
this URL.
* fragments A list of fragments of a fragmented media.
Each fragment entry must contain either an url
or a path. If an url is present it should be
considered by a client. Otherwise both path and
fragment_base_url must be present. Here is
the list of all potential fields:
* "url" - fragment's URL
* "path" - fragment's path relative to
* "duration" (optional, int or float)
* "filesize" (optional, int)
* is_from_start Is a live format that can be downloaded
from the start. Boolean
* preference Order number of this format. If this field is
present and not None, the formats get sorted
by this field, regardless of all other values.
-1 for default (order by other properties),
-2 or smaller for less than default.
< -1000 to hide the format (if there is
another one which is strictly better)
* language Language code, e.g. "de" or "en-US".
* language_preference Is this in the language mentioned in
the URL?
10 if it's what the URL is about,
-1 for default (don't know),
-10 otherwise, other values reserved for now.
* quality Order number of the video quality of this
format, irrespective of the file format.
-1 for default (order by other properties),
-2 or smaller for less than default.
* source_preference Order number for this video source
(quality takes higher priority)
-1 for default (order by other properties),
-2 or smaller for less than default.
* http_headers A dictionary of additional HTTP headers
to add to the request.
* stretched_ratio If given and not 1, indicates that the
video's pixels are not square.
width : height ratio as float.
* no_resume The server does not support resuming the
(HTTP or RTMP) download. Boolean.
* has_drm The format has DRM and cannot be downloaded. Boolean
* downloader_options A dictionary of downloader options
(For internal use only)
* http_chunk_size Chunk size for HTTP downloads
* ffmpeg_args Extra arguments for ffmpeg downloader
RTMP formats can also have the additional fields: page_url,
app, play_path, tc_url, flash_version, rtmp_live, rtmp_conn,
rtmp_protocol, rtmp_real_time
url: Final video URL.
ext: Video filename extension.
format: The video format, defaults to ext (used for --get-format)
player_url: SWF Player URL (used for rtmpdump).
The following fields are optional:
direct: True if a direct video file was given (must only be set by GenericIE)
alt_title: A secondary title of the video.
display_id An alternative identifier for the video, not necessarily
unique, but available before title. Typically, id is
something like "4234987", title "Dancing naked mole rats",
and display_id "dancing-naked-mole-rats"
thumbnails: A list of dictionaries, with the following entries:
* "id" (optional, string) - Thumbnail format ID
* "url"
* "preference" (optional, int) - quality of the image
* "width" (optional, int)
* "height" (optional, int)
* "resolution" (optional, string "{width}x{height}",
* "filesize" (optional, int)
* "http_headers" (dict) - HTTP headers for the request
thumbnail: Full URL to a video thumbnail image.
description: Full video description.
uploader: Full name of the video uploader.
license: License name the video is licensed under.
creator: The creator of the video.
timestamp: UNIX timestamp of the moment the video was uploaded
upload_date: Video upload date in UTC (YYYYMMDD).
If not explicitly set, calculated from timestamp
release_timestamp: UNIX timestamp of the moment the video was released.
If it is not clear whether to use timestamp or this, use the former
release_date: The date (YYYYMMDD) when the video was released in UTC.
If not explicitly set, calculated from release_timestamp
modified_timestamp: UNIX timestamp of the moment the video was last modified.
modified_date: The date (YYYYMMDD) when the video was last modified in UTC.
If not explicitly set, calculated from modified_timestamp
uploader_id: Nickname or id of the video uploader.
uploader_url: Full URL to a personal webpage of the video uploader.
channel: Full name of the channel the video is uploaded on.
Note that channel fields may or may not repeat uploader
fields. This depends on a particular extractor.
channel_id: Id of the channel.
channel_url: Full URL to a channel webpage.
channel_follower_count: Number of followers of the channel.
location: Physical location where the video was filmed.
subtitles: The available subtitles as a dictionary in the format
{tag: subformats}. "tag" is usually a language code, and
"subformats" is a list sorted from lower to higher
preference, each element is a dictionary with the "ext"
entry and one of:
* "data": The subtitles file contents
* "url": A URL pointing to the subtitles file
It can optionally also have:
* "name": Name or description of the subtitles
* "http_headers": A dictionary of additional HTTP headers
to add to the request.
"ext" will be calculated from URL if missing
automatic_captions: Like 'subtitles'; contains automatically generated
captions instead of normal subtitles
duration: Length of the video in seconds, as an integer or float.
view_count: How many users have watched the video on the platform.
like_count: Number of positive ratings of the video
dislike_count: Number of negative ratings of the video
repost_count: Number of reposts of the video
average_rating: Average rating give by users, the scale used depends on the webpage
comment_count: Number of comments on the video
comments: A list of comments, each with one or more of the following
properties (all but one of text or html optional):
* "author" - human-readable name of the comment author
* "author_id" - user ID of the comment author
* "author_thumbnail" - The thumbnail of the comment author
* "id" - Comment ID
* "html" - Comment as HTML
* "text" - Plain text of the comment
* "timestamp" - UNIX timestamp of comment
* "parent" - ID of the comment this one is replying to.
Set to "root" to indicate that this is a
comment to the original video.
* "like_count" - Number of positive ratings of the comment
* "dislike_count" - Number of negative ratings of the comment
* "is_favorited" - Whether the comment is marked as
favorite by the video uploader
* "author_is_uploader" - Whether the comment is made by
the video uploader
age_limit: Age restriction for the video, as an integer (years)
webpage_url: The URL to the video webpage, if given to yt-dlp it
should allow to get the same result again. (It will be set
by YoutubeDL if it's missing)
categories: A list of categories that the video falls in, for example
["Sports", "Berlin"]
tags: A list of tags assigned to the video, e.g. ["sweden", "pop music"]
cast: A list of the video cast
is_live: True, False, or None (=unknown). Whether this video is a
live stream that goes on instead of a fixed-length video.
was_live: True, False, or None (=unknown). Whether this video was
originally a live stream.
live_status: 'is_live', 'is_upcoming', 'was_live', 'not_live' or None (=unknown)
If absent, automatically set from is_live, was_live
start_time: Time in seconds where the reproduction should start, as
specified in the URL.
end_time: Time in seconds where the reproduction should end, as
specified in the URL.
chapters: A list of dictionaries, with the following entries:
* "start_time" - The start time of the chapter in seconds
* "end_time" - The end time of the chapter in seconds
* "title" (optional, string)
playable_in_embed: Whether this video is allowed to play in embedded
players on other sites. Can be True (=always allowed),
False (=never allowed), None (=unknown), or a string
specifying the criteria for embedability (Eg: 'whitelist')
availability: Under what condition the video is available. One of
'private', 'premium_only', 'subscriber_only', 'needs_auth',
'unlisted' or 'public'. Use 'InfoExtractor._availability'
to set it
__post_extractor: A function to be called just before the metadata is
written to either disk, logger or console. The function
must return a dict which will be added to the info_dict.
This is usefull for additional information that is
time-consuming to extract. Note that the fields thus
extracted will not be available to output template and
match_filter. So, only "comments" and "comment_count" are
currently allowed to be extracted via this method.
The following fields should only be used when the video belongs to some logical
chapter or section:
chapter: Name or title of the chapter the video belongs to.
chapter_number: Number of the chapter the video belongs to, as an integer.
chapter_id: Id of the chapter the video belongs to, as a unicode string.
The following fields should only be used when the video is an episode of some
series, programme or podcast:
series: Title of the series or programme the video episode belongs to.
series_id: Id of the series or programme the video episode belongs to, as a unicode string.
season: Title of the season the video episode belongs to.
season_number: Number of the season the video episode belongs to, as an integer.
season_id: Id of the season the video episode belongs to, as a unicode string.
episode: Title of the video episode. Unlike mandatory video title field,
this field should denote the exact title of the video episode
without any kind of decoration.
episode_number: Number of the video episode within a season, as an integer.
episode_id: Id of the video episode, as a unicode string.
The following fields should only be used when the media is a track or a part of
a music album:
track: Title of the track.
track_number: Number of the track within an album or a disc, as an integer.
track_id: Id of the track (useful in case of custom indexing, e.g. 6.iii),
as a unicode string.
artist: Artist(s) of the track.
genre: Genre(s) of the track.
album: Title of the album the track belongs to.
album_type: Type of the album (e.g. "Demo", "Full-length", "Split", "Compilation", etc).
album_artist: List of all artists appeared on the album (e.g.
"Ash Borer / Fell Voices" or "Various Artists", useful for splits
and compilations).
disc_number: Number of the disc or other physical medium the track belongs to,
as an integer.
release_year: Year (YYYY) when the album was released.
composer: Composer of the piece
The following fields should only be set for clips that should be cut from the original video:
section_start: Start time of the section in seconds
section_end: End time of the section in seconds
Unless mentioned otherwise, the fields should be Unicode strings.
Unless mentioned otherwise, None is equivalent to absence of information.
_type "playlist" indicates multiple videos.
There must be a key "entries", which is a list, an iterable, or a PagedList
object, each element of which is a valid dictionary by this specification.
Additionally, playlists can have "id", "title", and any other relevant
attributes with the same semantics as videos (see above).
It can also have the following optional fields:
playlist_count: The total number of videos in a playlist. If not given,
YoutubeDL tries to calculate it from "entries"
_type "multi_video" indicates that there are multiple videos that
form a single show, for examples multiple acts of an opera or TV episode.
It must have an entries key like a playlist and contain all the keys
required for a video at the same time.
_type "url" indicates that the video must be extracted from another
location, possibly by a different extractor. Its only required key is:
"url" - the next URL to extract.
The key "ie_key" can be set to the class name (minus the trailing "IE",
e.g. "Youtube") if the extractor class is known in advance.
Additionally, the dictionary may have any properties of the resolved entity
known in advance, for example "title" if the title of the referred video is
known ahead of time.
_type "url_transparent" entities have the same specification as "url", but
indicate that the given additional information is more precise than the one
associated with the resolved URL.
This is useful when a site employs a video service that hosts the video and
its technical metadata, but that video service does not embed a useful
title, description etc.
Subclasses of this should define a _VALID_URL regexp and, re-define the
_real_extract() and (optionally) _real_initialize() methods.
Probably, they should also be added to the list of extractors.
Subclasses may also override suitable() if necessary, but ensure the function
signature is preserved and that this function imports everything it needs
(except other extractors), so that lazy_extractors works correctly.
To support username + password (or netrc) login, the extractor must define a
_NETRC_MACHINE and re-define _perform_login(username, password) and
(optionally) _initialize_pre_login() methods. The _perform_login method will
be called between _initialize_pre_login and _real_initialize if credentials
are passed by the user. In cases where it is necessary to have the login
process as part of the extraction rather than initialization, _perform_login
can be left undefined.
_GEO_BYPASS attribute may be set to False in order to disable
geo restriction bypass mechanisms for a particular extractor.
Though it won't disable explicit geo restriction bypass based on
country code provided with geo_bypass_country.
_GEO_COUNTRIES attribute may contain a list of presumably geo unrestricted
countries for this extractor. One of these countries will be used by
geo restriction bypass mechanism right away in order to bypass
geo restriction, of course, if the mechanism is not disabled.
_GEO_IP_BLOCKS attribute may contain a list of presumably geo unrestricted
IP blocks in CIDR notation for this extractor. One of these IP blocks
will be used by geo restriction bypass mechanism similarly
The _WORKING attribute should be set to False for broken IEs
in order to warn the users and skip the tests.
_ready = False
_downloader = None
_x_forwarded_for_ip = None
IE_DESC = None
def _login_hint(self, method=NO_DEFAULT, netrc=None):
password_hint = f'--username and --password, or --netrc ({netrc or self._NETRC_MACHINE}) to provide account credentials'
return {
None: '',
'any': f'Use --cookies, --cookies-from-browser, {password_hint}',
'password': f'Use {password_hint}',
'cookies': (
'Use --cookies-from-browser or --cookies for the authentication. '
'See for how to manually pass cookies'),
}[method if method is not NO_DEFAULT else 'any' if self.supports_login() else 'cookies']
def __init__(self, downloader=None):
"""Constructor. Receives an optional downloader (a YoutubeDL instance).
If a downloader is not passed during initialization,
it must be set using "set_downloader()" before "extract()" is called"""
self._ready = False
self._x_forwarded_for_ip = None
self._printed_messages = set()
def _match_valid_url(cls, url):
# This does not use has/getattr intentionally - we want to know whether
# we have cached the regexp for *this* class, whereas getattr would also
# match the superclass
if '_VALID_URL_RE' not in cls.__dict__:
if '_VALID_URL' not in cls.__dict__:
cls._VALID_URL = cls._make_valid_url()
cls._VALID_URL_RE = re.compile(cls._VALID_URL)
return cls._VALID_URL_RE.match(url)
def suitable(cls, url):
"""Receives a URL and returns True if suitable for this IE."""
# This function must import everything it needs (except other extractors),
# so that lazy_extractors works correctly
return cls._match_valid_url(url) is not None
def _match_id(cls, url):
return cls._match_valid_url(url).group('id')
def get_temp_id(cls, url):
return cls._match_id(url)
except (IndexError, AttributeError):
return None
def working(cls):
"""Getter method for _WORKING."""
return cls._WORKING
def supports_login(cls):
return bool(cls._NETRC_MACHINE)
def initialize(self):
"""Initializes an instance (authentication, etc)."""
self._printed_messages = set()
'countries': self._GEO_COUNTRIES,
'ip_blocks': self._GEO_IP_BLOCKS,
if not self._ready:
if self.supports_login():
username, password = self._get_login_info()
if username:
self._perform_login(username, password)
elif self.get_param('username') and False not in (self.IE_DESC, self._NETRC_MACHINE):
self.report_warning(f'Login with password is not supported for this website. {self._login_hint("cookies")}')
self._ready = True
def _initialize_geo_bypass(self, geo_bypass_context):
Initialize geo restriction bypass mechanism.
This method is used to initialize geo bypass mechanism based on faking
X-Forwarded-For HTTP header. A random country from provided country list
is selected and a random IP belonging to this country is generated. This
IP will be passed as X-Forwarded-For HTTP header in all subsequent
HTTP requests.
This method will be used for initial geo bypass mechanism initialization
during the instance initialization with _GEO_COUNTRIES and
You may also manually call it from extractor's code if geo bypass
information is not available beforehand (e.g. obtained during
extraction) or due to some other reason. In this case you should pass
this information in geo bypass context passed as first argument. It may
contain following fields:
countries: List of geo unrestricted countries (similar
ip_blocks: List of geo unrestricted IP blocks in CIDR notation
(similar to _GEO_IP_BLOCKS)
if not self._x_forwarded_for_ip:
# Geo bypass mechanism is explicitly disabled by user
if not self.get_param('geo_bypass', True):
if not geo_bypass_context:
geo_bypass_context = {}
# Backward compatibility: previously _initialize_geo_bypass
# expected a list of countries, some 3rd party code may still use
# it this way
if isinstance(geo_bypass_context, (list, tuple)):
geo_bypass_context = {
'countries': geo_bypass_context,
# The whole point of geo bypass mechanism is to fake IP
# as X-Forwarded-For HTTP header based on some IP block or
# country code.
# Path 1: bypassing based on IP block in CIDR notation
# Explicit IP block specified by user, use it right away
# regardless of whether extractor is geo bypassable or not
ip_block = self.get_param('geo_bypass_ip_block', None)
# Otherwise use random IP block from geo bypass context but only
# if extractor is known as geo bypassable
if not ip_block:
ip_blocks = geo_bypass_context.get('ip_blocks')
if self._GEO_BYPASS and ip_blocks:
ip_block = random.choice(ip_blocks)
if ip_block:
self._x_forwarded_for_ip = GeoUtils.random_ipv4(ip_block)
self.write_debug(f'Using fake IP {self._x_forwarded_for_ip} as X-Forwarded-For')
# Path 2: bypassing based on country code
# Explicit country code specified by user, use it right away
# regardless of whether extractor is geo bypassable or not
country = self.get_param('geo_bypass_country', None)
# Otherwise use random country code from geo bypass context but
# only if extractor is known as geo bypassable
if not country:
countries = geo_bypass_context.get('countries')
if self._GEO_BYPASS and countries:
country = random.choice(countries)
if country:
self._x_forwarded_for_ip = GeoUtils.random_ipv4(country)
f'Using fake IP {self._x_forwarded_for_ip} ({country.upper()}) as X-Forwarded-For')
def extract(self, url):
"""Extracts URL information and returns it in list of dicts."""
for _ in range(2):
self.write_debug('Extracting URL: %s' % url)
ie_result = self._real_extract(url)
if ie_result is None:
return None
if self._x_forwarded_for_ip:
ie_result['__x_forwarded_for_ip'] = self._x_forwarded_for_ip
subtitles = ie_result.get('subtitles')
if (subtitles and 'live_chat' in subtitles
and 'no-live-chat' in self.get_param('compat_opts', [])):
del subtitles['live_chat']
return ie_result
except GeoRestrictedError as e:
if self.__maybe_fake_ip_and_retry(e.countries):
except UnsupportedError:
except ExtractorError as e:
kwargs = {
'video_id': e.video_id or self.get_temp_id(url),
'ie': self.IE_NAME,
'tb': e.traceback or sys.exc_info()[2],
'expected': e.expected,
'cause': e.cause
if hasattr(e, 'countries'):
kwargs['countries'] = e.countries
raise type(e)(e.orig_msg, **kwargs)
except http.client.IncompleteRead as e:
raise ExtractorError('A network error has occurred.', cause=e, expected=True, video_id=self.get_temp_id(url))
except (KeyError, StopIteration) as e:
raise ExtractorError('An extractor error has occurred.', cause=e, video_id=self.get_temp_id(url))
def __maybe_fake_ip_and_retry(self, countries):
if (not self.get_param('geo_bypass_country', None)
and self._GEO_BYPASS
and self.get_param('geo_bypass', True)
and not self._x_forwarded_for_ip
and countries):
country_code = random.choice(countries)
self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code)
if self._x_forwarded_for_ip:
'Video is geo restricted. Retrying extraction with fake IP %s (%s) as X-Forwarded-For.'
% (self._x_forwarded_for_ip, country_code.upper()))
return True
return False
def set_downloader(self, downloader):
"""Sets a YoutubeDL instance as the downloader for this IE."""
self._downloader = downloader
def cache(self):
return self._downloader.cache
def cookiejar(self):
return self._downloader.cookiejar
def _initialize_pre_login(self):
""" Initialization before login. Redefine in subclasses."""
def _perform_login(self, username, password):
""" Login with username and password. Redefine in subclasses."""
def _real_initialize(self):
"""Real initialization process. Redefine in subclasses."""
def _real_extract(self, url):
"""Real extraction process. Redefine in subclasses."""
raise NotImplementedError('This method must be implemented by subclasses')
def ie_key(cls):
"""A string for getting the InfoExtractor with get_info_extractor"""
return cls.__name__[:-2]
def IE_NAME(cls):
return cls.__name__[:-2]
def __can_accept_status_code(err, expected_status):
assert isinstance(err, urllib.error.HTTPError)
if expected_status is None:
return False
elif callable(expected_status):
return expected_status(err.code) is True
return err.code in variadic(expected_status)
def _create_request(self, url_or_request, data=None, headers=None, query=None):
if isinstance(url_or_request, urllib.request.Request):
return update_Request(url_or_request, data=data, headers=headers, query=query)
if query:
url_or_request = update_url_query(url_or_request, query)
return sanitized_Request(url_or_request, data, headers or {})
def _request_webpage(self, url_or_request, video_id, note=None, errnote=None, fatal=True, data=None, headers=None, query=None, expected_status=None):
Return the response handle.
See _download_webpage docstring for arguments specification.
if not self._downloader._first_webpage_request:
sleep_interval = self.get_param('sleep_interval_requests') or 0
if sleep_interval > 0:
self.to_screen('Sleeping %s seconds ...' % sleep_interval)
self._downloader._first_webpage_request = False
if note is None:
elif note is not False:
if video_id is None:
self.to_screen(f'{video_id}: {note}')
# Some sites check X-Forwarded-For HTTP header in order to figure out
# the origin of the client behind proxy. This allows bypassing geo
# restriction by faking this header's value to IP that belongs to some
# geo unrestricted country. We will do so once we encounter any
# geo restriction error.
if self._x_forwarded_for_ip:
headers = (headers or {}).copy()
headers.setdefault('X-Forwarded-For', self._x_forwarded_for_ip)
return self._downloader.urlopen(self._create_request(url_or_request, data, headers, query))
except network_exceptions as err:
if isinstance(err, urllib.error.HTTPError):
if self.__can_accept_status_code(err, expected_status):
# Retain reference to error to prevent file object from
# being closed before it can be read. Works around the
# effects of <>
# introduced in Python 3.4.1.
err.fp._error = err
return err.fp
if errnote is False:
return False
if errnote is None:
errnote = 'Unable to download webpage'
errmsg = f'{errnote}: {error_to_compat_str(err)}'
if fatal:
raise ExtractorError(errmsg, cause=err)
return False
def _download_webpage_handle(self, url_or_request, video_id, note=None, errnote=None, fatal=True,
encoding=None, data=None, headers={}, query={}, expected_status=None):
Return a tuple (page content as string, URL handle).
url_or_request -- plain text URL as a string or
a urllib.request.Request object
video_id -- Video/playlist/item identifier (string)
Keyword arguments:
note -- note printed before downloading (string)
errnote -- note printed in case of an error (string)
fatal -- flag denoting whether error should be considered fatal,
i.e. whether it should cause ExtractionError to be raised,
otherwise a warning will be reported and extraction continued
encoding -- encoding for a page content decoding, guessed automatically
when not explicitly specified
data -- POST data (bytes)
headers -- HTTP headers (dict)
query -- URL query (dict)
expected_status -- allows to accept failed HTTP requests (non 2xx
status code) by explicitly specifying a set of accepted status
codes. Can be any of the following entities:
- an integer type specifying an exact failed status code to
- a list or a tuple of integer types specifying a list of
failed status codes to accept
- a callable accepting an actual failed status code and
returning True if it should be accepted
Note that this argument does not affect success status codes (2xx)
which are always accepted.
# Strip hashes from the URL (#1038)
if isinstance(url_or_request, str):
url_or_request = url_or_request.partition('#')[0]
urlh = self._request_webpage(url_or_request, video_id, note, errnote, fatal, data=data, headers=headers, query=query, expected_status=expected_status)
if urlh is False:
assert not fatal
return False
content = self._webpage_read_content(urlh, url_or_request, video_id, note, errnote, fatal, encoding=encoding)
return (content, urlh)
def _guess_encoding_from_content(content_type, webpage_bytes):
m = re.match(r'[a-zA-Z0-9_.-]+/[a-zA-Z0-9_.-]+\s*;\s*charset=(.+)', content_type)
if m:
encoding =
m ='<meta[^>]+charset=[\'"]?([^\'")]+)[ /\'">]',
if m:
encoding ='ascii')
elif webpage_bytes.startswith(b'\xff\xfe'):
encoding = 'utf-16'
encoding = 'utf-8'
return encoding
def __check_blocked(self, content):
first_block = content[:512]
if ('<title>Access to this site is blocked</title>' in content
and 'Websense' in first_block):
msg = 'Access to this webpage has been blocked by Websense filtering software in your network.'
blocked_iframe = self._html_search_regex(
r'<iframe src="([^"]+)"', content,
'Websense information URL', default=None)
if blocked_iframe:
msg += ' Visit %s for more details' % blocked_iframe
raise ExtractorError(msg, expected=True)
if '<title>The URL you requested has been blocked</title>' in first_block:
msg = (
'Access to this webpage has been blocked by Indian censorship. '
'Use a VPN or proxy server (with --proxy) to route around it.')
block_msg = self._html_search_regex(
content, 'block message', default=None)
if block_msg:
msg += ' (Message: "%s")' % block_msg.replace('\n', ' ')
raise ExtractorError(msg, expected=True)
if ('<title>TTK :: Доступ к ресурсу ограничен</title>' in content
and '' in content):
raise ExtractorError(
'Access to this webpage has been blocked by decision of the Russian government. '
'Visit for a block reason.',
def _request_dump_filename(self, url, video_id):
basen = f'{video_id}_{url}'
trim_length = self.get_param('trim_file_name') or 240
if len(basen) > trim_length:
h = '___' + hashlib.md5(basen.encode('utf-8')).hexdigest()
basen = basen[:trim_length - len(h)] + h
filename = sanitize_filename(f'{basen}.dump', restricted=True)
# Working around MAX_PATH limitation on Windows (see
if compat_os_name == 'nt':
absfilepath = os.path.abspath(filename)
if len(absfilepath) > 259:
filename = fR'\\?\{absfilepath}'
return filename
def __decode_webpage(self, webpage_bytes, encoding, headers):
if not encoding:
encoding = self._guess_encoding_from_content(headers.get('Content-Type', ''), webpage_bytes)
return webpage_bytes.decode(encoding, 'replace')
except LookupError:
return webpage_bytes.decode('utf-8', 'replace')
def _webpage_read_content(self, urlh, url_or_request, video_id, note=None, errnote=None, fatal=True, prefix=None, encoding=None):
webpage_bytes =
if prefix is not None:
webpage_bytes = prefix + webpage_bytes
if self.get_param('dump_intermediate_pages', False):
self.to_screen('Dumping request to ' + urlh.geturl())
dump = base64.b64encode(webpage_bytes).decode('ascii')
if self.get_param('write_pages'):
filename = self._request_dump_filename(urlh.geturl(), video_id)
self.to_screen(f'Saving request to {filename}')
with open(filename, 'wb') as outf:
content = self.__decode_webpage(webpage_bytes, encoding, urlh.headers)
return content
def _parse_xml(self, xml_string, video_id, transform_source=None, fatal=True):
if transform_source:
xml_string = transform_source(xml_string)
return compat_etree_fromstring(xml_string.encode('utf-8'))
except xml.etree.ElementTree.ParseError as ve:
errmsg = '%s: Failed to parse XML ' % video_id
if fatal:
raise ExtractorError(errmsg, cause=ve)
self.report_warning(errmsg + str(ve))
def _parse_json(self, json_string, video_id, transform_source=None, fatal=True, **parser_kwargs):
return json.loads(
json_string, cls=LenientJSONDecoder, strict=False, transform_source=transform_source, **parser_kwargs)
except ValueError as ve:
errmsg = f'{video_id}: Failed to parse JSON'
if fatal:
raise ExtractorError(errmsg, cause=ve)
self.report_warning(f'{errmsg}: {ve}')
def _parse_socket_response_as_json(self, data, video_id, transform_source=None, fatal=True):
return self._parse_json(
data[data.find('{'):data.rfind('}') + 1],
video_id, transform_source, fatal)
def __create_download_methods(name, parser, note, errnote, return_value):
def parse(ie, content, *args, **kwargs):
if parser is None:
return content
# parser is fetched by name so subclasses can override it
return getattr(ie, parser)(content, *args, **kwargs)
def download_handle(self, url_or_request, video_id, note=note, errnote=errnote, transform_source=None,
fatal=True, encoding=None, data=None, headers={}, query={}, expected_status=None):
res = self._download_webpage_handle(
url_or_request, video_id, note=note, errnote=errnote, fatal=fatal, encoding=encoding,
data=data, headers=headers, query=query, expected_status=expected_status)
if res is False:
return res
content, urlh = res
return parse(self, content, video_id, transform_source=transform_source, fatal=fatal), urlh
def download_content(self, url_or_request, video_id, note=note, errnote=errnote, transform_source=None,
fatal=True, encoding=None, data=None, headers={}, query={}, expected_status=None):
if self.get_param('load_pages'):
url_or_request = self._create_request(url_or_request, data, headers, query)
filename = self._request_dump_filename(url_or_request.full_url, video_id)
self.to_screen(f'Loading request from {filename}')
with open(filename, 'rb') as dumpf:
webpage_bytes =
except OSError as e:
self.report_warning(f'Unable to load request from disk: {e}')
content = self.__decode_webpage(webpage_bytes, encoding, url_or_request.headers)
return parse(self, content, video_id, transform_source, fatal)
kwargs = {
'note': note,
'errnote': errnote,
'transform_source': transform_source,
'fatal': fatal,
'encoding': encoding,
'data': data,
'headers': headers,
'query': query,
'expected_status': expected_status,
if parser is None:
# The method is fetched by name so subclasses can override _download_..._handle
res = getattr(self, download_handle.__name__)(url_or_request, video_id, **kwargs)
return res if res is False else res[0]
def impersonate(func, name, return_value):
func.__name__, func.__qualname__ = name, f'InfoExtractor.{name}'
func.__doc__ = f'''
@param transform_source Apply this transformation before parsing
@returns {return_value}
See _download_webpage_handle docstring for other arguments specification
impersonate(download_handle, f'_download_{name}_handle', f'({return_value}, URL handle)')
impersonate(download_content, f'_download_{name}', f'{return_value}')
return download_handle, download_content
_download_xml_handle, _download_xml = __create_download_methods(
'xml', '_parse_xml', 'Downloading XML', 'Unable to download XML', 'xml as an xml.etree.ElementTree.Element')
_download_json_handle, _download_json = __create_download_methods(
'json', '_parse_json', 'Downloading JSON metadata', 'Unable to download JSON metadata', 'JSON object as a dict')
_download_socket_json_handle, _download_socket_json = __create_download_methods(
'socket_json', '_parse_socket_response_as_json', 'Polling socket', 'Unable to poll socket', 'JSON object as a dict')
__download_webpage = __create_download_methods('webpage', None, None, None, 'data of the page as a string')[1]
def _download_webpage(
self, url_or_request, video_id, note=None, errnote=None,
fatal=True, tries=1, timeout=NO_DEFAULT, *args, **kwargs):
Return the data of the page as a string.
Keyword arguments:
tries -- number of tries
timeout -- sleep interval between tries
See _download_webpage_handle docstring for other arguments specification.
R''' # NB: These are unused; should they be deprecated?
if tries != 1:
self._downloader.deprecation_warning('tries argument is deprecated in InfoExtractor._download_webpage')
if timeout is NO_DEFAULT:
timeout = 5
self._downloader.deprecation_warning('timeout argument is deprecated in InfoExtractor._download_webpage')
try_count = 0
while True:
return self.__download_webpage(url_or_request, video_id, note, errnote, None, fatal, *args, **kwargs)
except http.client.IncompleteRead as e:
try_count += 1
if try_count >= tries:
raise e
self._sleep(timeout, video_id)
def report_warning(self, msg, video_id=None, *args, only_once=False, **kwargs):
idstr = format_field(video_id, None, '%s: ')
msg = f'[{self.IE_NAME}] {idstr}{msg}'
if only_once:
if f'WARNING: {msg}' in self._printed_messages:
self._printed_messages.add(f'WARNING: {msg}')
self._downloader.report_warning(msg, *args, **kwargs)
def to_screen(self, msg, *args, **kwargs):
"""Print msg to screen, prefixing it with '[ie_name]'"""
self._downloader.to_screen(f'[{self.IE_NAME}] {msg}', *args, **kwargs)
def write_debug(self, msg, *args, **kwargs):
self._downloader.write_debug(f'[{self.IE_NAME}] {msg}', *args, **kwargs)
def get_param(self, name, default=None, *args, **kwargs):
if self._downloader:
return self._downloader.params.get(name, default, *args, **kwargs)
return default
def report_drm(self, video_id, partial=False):
self.raise_no_formats('This video is DRM protected', expected=True, video_id=video_id)
def report_extraction(self, id_or_name):
"""Report information extraction."""
self.to_screen('%s: Extracting information' % id_or_name)
def report_download_webpage(self, video_id):
"""Report webpage download."""
self.to_screen('%s: Downloading webpage' % video_id)
def report_age_confirmation(self):
"""Report attempt to confirm age."""
self.to_screen('Confirming age')
def report_login(self):
"""Report attempt to log in."""
self.to_screen('Logging in')
def raise_login_required(
self, msg='This video is only available for registered users',
metadata_available=False, method=NO_DEFAULT):
if metadata_available and (
self.get_param('ignore_no_formats_error') or self.get_param('wait_for_video')):
msg += format_field(self._login_hint(method), None, '. %s')
raise ExtractorError(msg, expected=True)
def raise_geo_restricted(
self, msg='This video is not available from your location due to geo restriction',
countries=None, metadata_available=False):
if metadata_available and (
self.get_param('ignore_no_formats_error') or self.get_param('wait_for_video')):
raise GeoRestrictedError(msg, countries=countries)
def raise_no_formats(self, msg, expected=False, video_id=None):
if expected and (
self.get_param('ignore_no_formats_error') or self.get_param('wait_for_video')):
self.report_warning(msg, video_id)
elif isinstance(msg, ExtractorError):
raise msg
raise ExtractorError(msg, expected=expected, video_id=video_id)
# Methods for following #608
def url_result(url, ie=None, video_id=None, video_title=None, *, url_transparent=False, **kwargs):
"""Returns a URL that points to a page that should be processed"""
if ie is not None:
kwargs['ie_key'] = ie if isinstance(ie, str) else ie.ie_key()
if video_id is not None:
kwargs['id'] = video_id
if video_title is not None:
kwargs['title'] = video_title
return {
'_type': 'url_transparent' if url_transparent else 'url',
'url': url,
def playlist_from_matches(self, matches, playlist_id=None, playlist_title=None, getter=None, ie=None, video_kwargs=None, **kwargs):
urls = (self.url_result(self._proto_relative_url(m), ie, **(video_kwargs or {}))
for m in orderedSet(map(getter, matches) if getter else matches))
return self.playlist_result(urls, playlist_id, playlist_title, **kwargs)
def playlist_result(entries, playlist_id=None, playlist_title=None, playlist_description=None, *, multi_video=False, **kwargs):
"""Returns a playlist"""
if playlist_id:
kwargs['id'] = playlist_id
if playlist_title:
kwargs['title'] = playlist_title
if playlist_description is not None:
kwargs['description'] = playlist_description
return {
'_type': 'multi_video' if multi_video else 'playlist',
'entries': entries,
def _search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
Perform a regex search on the given string, using a single or a list of
patterns returning the first matching group.
In case of failure return a default value or raise a WARNING or a
RegexNotFoundError, depending on fatal, specifying the field name.
if string is None:
mobj = None
elif isinstance(pattern, (str, re.Pattern)):
mobj =, string, flags)
for p in pattern:
mobj =, string, flags)
if mobj:
_name = self._downloader._format_err(name, self._downloader.Styles.EMPHASIS)
if mobj:
if group is None:
# return the first matching group
return next(g for g in mobj.groups() if g is not None)
elif isinstance(group, (list, tuple)):
return tuple( for g in group)
elif default is not NO_DEFAULT:
return default
elif fatal:
raise RegexNotFoundError('Unable to extract %s' % _name)
self.report_warning('unable to extract %s' % _name + bug_reports_message())
return None
def _search_json(self, start_pattern, string, name, video_id, *, end_pattern='',
contains_pattern='(?s:.+)', fatal=True, default=NO_DEFAULT, **kwargs):
"""Searches string for the JSON object specified by start_pattern"""
# NB: end_pattern is only used to reduce the size of the initial match
if default is NO_DEFAULT:
default, has_default = {}, False
fatal, has_default = False, True
json_string = self._search_regex(
string, name, group='json', fatal=fatal, default=None if has_default else NO_DEFAULT)
if not json_string:
return default
_name = self._downloader._format_err(name, self._downloader.Styles.EMPHASIS)
return self._parse_json(json_string, video_id, ignore_extra=True, **kwargs)
except ExtractorError as e:
if fatal:
raise ExtractorError(
f'Unable to extract {_name} - Failed to parse JSON', cause=e.cause, video_id=video_id)
elif not has_default:
f'Unable to extract {_name} - Failed to parse JSON: {e}', video_id=video_id)
return default
def _html_search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
Like _search_regex, but strips HTML tags and unescapes entities.
res = self._search_regex(pattern, string, name, default, fatal, flags, group)
if res:
return clean_html(res).strip()
return res
def _get_netrc_login_info(self, netrc_machine=None):
username = None
password = None
netrc_machine = netrc_machine or self._NETRC_MACHINE
if self.get_param('usenetrc', False):
netrc_file = compat_expanduser(self.get_param('netrc_location') or '~')
if os.path.isdir(netrc_file):
netrc_file = os.path.join(netrc_file, '.netrc')
info = netrc.netrc(file=netrc_file).authenticators(netrc_machine)
if info is not None:
username = info[0]
password = info[2]
raise netrc.NetrcParseError(
'No authenticators for %s' % netrc_machine)
except (OSError, netrc.NetrcParseError) as err:
'parsing .netrc: %s' % error_to_compat_str(err))
return username, password
def _get_login_info(self, username_option='username', password_option='password', netrc_machine=None):
Get the login info as (username, password)
First look for the manually specified credentials using username_option
and password_option as keys in params dictionary. If no such credentials
available look in the netrc file using the netrc_machine or _NETRC_MACHINE
If there's no info available, return (None, None)
# Attempt to use provided username and password or .netrc data
username = self.get_param(username_option)
if username is not None:
password = self.get_param(password_option)
username, password = self._get_netrc_login_info(netrc_machine)
return username, password
def _get_tfa_info(self, note='two-factor verification code'):
Get the two-factor authentication info
TODO - asking the user will be required for sms/phone verify
currently just uses the command line option
If there's no info available, return None
tfa = self.get_param('twofactor')
if tfa is not None:
return tfa
return getpass.getpass('Type %s and press [Return]: ' % note)
# Helper functions for extracting OpenGraph info
def _og_regexes(prop):
content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))'
property_re = (r'(?:name|property)=(?:\'og%(sep)s%(prop)s\'|"og%(sep)s%(prop)s"|\s*og%(sep)s%(prop)s\b)'
% {'prop': re.escape(prop), 'sep': '(?:&#x3A;|[:-])'})
template = r'<meta[^>]+?%s[^>]+?%s'
return [
template % (property_re, content_re),
template % (content_re, property_re),
def _meta_regex(prop):
return r'''(?isx)<meta
[^>]+?content=(["\'])(?P<content>.*?)\2''' % re.escape(prop)
def _og_search_property(self, prop, html, name=None, **kargs):
prop = variadic(prop)
if name is None:
name = 'OpenGraph %s' % prop[0]
og_regexes = []
for p in prop:
escaped = self._search_regex(og_regexes, html, name, flags=re.DOTALL, **kargs)
if escaped is None:
return None
return unescapeHTML(escaped)
def _og_search_thumbnail(self, html, **kargs):
return self._og_search_property('image', html, 'thumbnail URL', fatal=False, **kargs)
def _og_search_description(self, html, **kargs):
return self._og_search_property('description', html, fatal=False, **kargs)
def _og_search_title(self, html, *, fatal=False, **kargs):
return self._og_search_property('title', html, fatal=fatal, **kargs)
def _og_search_video_url(self, html, name='video url', secure=True, **kargs):
regexes = self._og_regexes('video') + self._og_regexes('video:url')
if secure:
regexes = self._og_regexes('video:secure_url') + regexes
return self._html_search_regex(regexes, html, name, **kargs)
def _og_search_url(self, html, **kargs):
return self._og_search_property('url', html, **kargs)
def _html_extract_title(self, html, name='title', *, fatal=False, **kwargs):
return self._html_search_regex(r'(?s)<title\b[^>]*>([^<]+)</title>', html, name, fatal=fatal, **kwargs)
def _html_search_meta(self, name, html, display_name=None, fatal=False, **kwargs):
name = variadic(name)
if display_name is None:
display_name = name[0]
return self._html_search_regex(
[self._meta_regex(n) for n in name],
html, display_name, fatal=fatal, group='content', **kwargs)
def _dc_search_uploader(self, html):
return self._html_search_meta('dc.creator', html, 'uploader')
def _rta_search(self, html):
# See
r' content="RTA-5042-1996-1400-1577-RTA"',
return 18
return 0
def _media_rating_search(self, html):
# See
rating = self._html_search_meta('rating', html)
if not rating:
return None
'safe for kids': 0,
'general': 8,
'14 years': 14,
'mature': 17,
'restricted': 19,
return RATING_TABLE.get(rating.lower())
def _family_friendly_search(self, html):
# See
family_friendly = self._html_search_meta(
'isFamilyFriendly', html, default=None)
if not family_friendly:
return None
'1': 0,
'true': 0,
'0': 18,
'false': 18,
return RATING_TABLE.get(family_friendly.lower())
def _twitter_search_player(self, html):
return self._html_search_meta('twitter:player', html,
'twitter card player')
def _yield_json_ld(self, html, video_id, *, fatal=True, default=NO_DEFAULT):
"""Yield all json ld objects in the html"""
if default is not NO_DEFAULT:
fatal = False
for mobj in re.finditer(JSON_LD_RE, html):
json_ld_item = self._parse_json('json_ld'), video_id, fatal=fatal)
for json_ld in variadic(json_ld_item):
if isinstance(json_ld, dict):
yield json_ld
def _search_json_ld(self, html, video_id, expected_type=None, *, fatal=True, default=NO_DEFAULT):
"""Search for a video in any json ld in the html"""
if default is not NO_DEFAULT:
fatal = False
info = self._json_ld(
list(self._yield_json_ld(html, video_id, fatal=fatal, default=default)),
video_id, fatal=fatal, expected_type=expected_type)
if info:
return info
if default is not NO_DEFAULT:
return default
elif fatal:
raise RegexNotFoundError('Unable to extract JSON-LD')
self.report_warning('unable to extract JSON-LD %s' % bug_reports_message())
return {}
def _json_ld(self, json_ld, video_id, fatal=True, expected_type=None):
if isinstance(json_ld, str):
json_ld = self._parse_json(json_ld, video_id, fatal=fatal)
if not json_ld:
return {}
info = {}
if not isinstance(json_ld, (list, tuple, dict)):
return info
if isinstance(json_ld, dict):
json_ld = [json_ld]
'CommentAction': 'comment',
'AgreeAction': 'like',
'DisagreeAction': 'dislike',
'LikeAction': 'like',
'DislikeAction': 'dislike',
'ListenAction': 'view',
'WatchAction': 'view',
'ViewAction': 'view',
def is_type(e, *expected_types):
type = variadic(traverse_obj(e, '@type'))
return any(x in type for x in expected_types)
def extract_interaction_type(e):
interaction_type = e.get('interactionType')
if isinstance(interaction_type, dict):
interaction_type = interaction_type.get('@type')
return str_or_none(interaction_type)
def extract_interaction_statistic(e):
interaction_statistic = e.get('interactionStatistic')
if isinstance(interaction_statistic, dict):
interaction_statistic = [interaction_statistic]
if not isinstance(interaction_statistic, list):
for is_e in interaction_statistic:
if not is_type(is_e, 'InteractionCounter'):
interaction_type = extract_interaction_type(is_e)
if not interaction_type:
# For interaction count some sites provide string instead of
# an integer (as per spec) with non digit characters (e.g. ",")
# so extracting count with more relaxed str_to_int
interaction_count = str_to_int(is_e.get('userInteractionCount'))
if interaction_count is None:
count_kind = INTERACTION_TYPE_MAP.get(interaction_type.split('/')[-1])
if not count_kind:
count_key = '%s_count' % count_kind
if info.get(count_key) is not None:
info[count_key] = interaction_count
def extract_chapter_information(e):
chapters = [{
'title': part.get('name'),
'start_time': part.get('startOffset'),
'end_time': part.get('endOffset'),
} for part in variadic(e.get('hasPart') or []) if part.get('@type') == 'Clip']
for idx, (last_c, current_c, next_c) in enumerate(zip(
[{'end_time': 0}] + chapters, chapters, chapters[1:])):
current_c['end_time'] = current_c['end_time'] or next_c['start_time']
current_c['start_time'] = current_c['start_time'] or last_c['end_time']
if None in current_c.values():
self.report_warning(f'Chapter {idx} contains broken data. Not extracting chapters')
if chapters:
chapters[-1]['end_time'] = chapters[-1]['end_time'] or info['duration']
info['chapters'] = chapters
def extract_video_object(e):
assert is_type(e, 'VideoObject')
author = e.get('author')
'url': url_or_none(e.get('contentUrl')),
'title': unescapeHTML(e.get('name')),
'description': unescapeHTML(e.get('description')),
'thumbnails': [{'url': url}
for url in variadic(traverse_obj(e, 'thumbnailUrl', 'thumbnailURL'))
if url_or_none(url)],
'duration': parse_duration(e.get('duration')),
'timestamp': unified_timestamp(e.get('uploadDate')),
# author can be an instance of 'Organization' or 'Person' types.
# both types can have 'name' property(inherited from 'Thing' type). [1]
# however some websites are using 'Text' type instead.
# 1.
'uploader': author.get('name') if isinstance(author, dict) else author if isinstance(author, str) else None,
'filesize': int_or_none(float_or_none(e.get('contentSize'))),
'tbr': int_or_none(e.get('bitrate')),
'width': int_or_none(e.get('width')),
'height': int_or_none(e.get('height')),
'view_count': int_or_none(e.get('interactionCount')),
def traverse_json_ld(json_ld, at_top_level=True):
for e in json_ld:
if at_top_level and '@context' not in e:
if at_top_level and set(e.keys()) == {'@context', '@graph'}:
traverse_json_ld(variadic(e['@graph'], allowed_types=(dict,)), at_top_level=False)
if expected_type is not None and not is_type(e, expected_type):
rating = traverse_obj(e, ('aggregateRating', 'ratingValue'), expected_type=float_or_none)
if rating is not None:
info['average_rating'] = rating
if is_type(e, 'TVEpisode', 'Episode'):
episode_name = unescapeHTML(e.get('name'))
'episode': episode_name,
'episode_number': int_or_none(e.get('episodeNumber')),
'description': unescapeHTML(e.get('description')),
if not info.get('title') and episode_name:
info['title'] = episode_name
part_of_season = e.get('partOfSeason')
if is_type(part_of_season, 'TVSeason', 'Season', 'CreativeWorkSeason'):
'season': unescapeHTML(part_of_season.get('name')),
'season_number': int_or_none(part_of_season.get('seasonNumber')),
part_of_series = e.get('partOfSeries') or e.get('partOfTVSeries')
if is_type(part_of_series, 'TVSeries', 'Series', 'CreativeWorkSeries'):
info['series'] = unescapeHTML(part_of_series.get('name'))
elif is_type(e, 'Movie'):
'title': unescapeHTML(e.get('name')),
'description': unescapeHTML(e.get('description')),
'duration': parse_duration(e.get('duration')),
'timestamp': unified_timestamp(e.get('dateCreated')),
elif is_type(e, 'Article', 'NewsArticle'):
'timestamp': parse_iso8601(e.get('datePublished')),
'title': unescapeHTML(e.get('headline')),
'description': unescapeHTML(e.get('articleBody') or e.get('description')),
if is_type(traverse_obj(e, ('video', 0)), 'VideoObject'):
elif is_type(traverse_obj(e, ('subjectOf', 0)), 'VideoObject'):
elif is_type(e, 'VideoObject'):
if expected_type is None:
video = e.get('video')
if is_type(video, 'VideoObject'):
if expected_type is None:
return filter_dict(info)
def _search_nextjs_data(self, webpage, video_id, *, transform_source=None, fatal=True, **kw):
return self._parse_json(
webpage, 'next.js data', fatal=fatal, **kw),
video_id, transform_source=transform_source, fatal=fatal)
def _search_nuxt_data(self, webpage, video_id, context_name='__NUXT__', *, fatal=True, traverse=('data', 0)):
"""Parses Nuxt.js metadata. This works as long as the function __NUXT__ invokes is a pure function"""
rectx = re.escape(context_name)
FUNCTION_RE = r'\(function\((?P<arg_keys>.*?)\){return\s+(?P<js>{.*?})\s*;?\s*}\((?P<arg_vals>.*?)\)'
js, arg_keys, arg_vals = self._search_regex(
(rf'<script>\s*window\.{rectx}={FUNCTION_RE}\s*\)\s*;?\s*</script>', rf'{rectx}\(.*?{FUNCTION_RE}'),
webpage, context_name, group=('js', 'arg_keys', 'arg_vals'), fatal=fatal)
args = dict(zip(arg_keys.split(','), arg_vals.split(',')))
for key, val in args.items():
if val in ('undefined', 'void 0'):
args[key] = 'null'
ret = self._parse_json(js, video_id, transform_source=functools.partial(js_to_json, vars=args), fatal=fatal)
return traverse_obj(ret, traverse) or {}
def _hidden_inputs(html):
html = re.sub(r'<!--(?:(?!<!--).)*-->', '', html)
hidden_inputs = {}
for input in re.findall(r'(?i)(<input[^>]+>)', html):
attrs = extract_attributes(input)
if not input:
if attrs.get('type') not in ('hidden', 'submit'):
name = attrs.get('name') or attrs.get('id')
value = attrs.get('value')
if name and value is not None:
hidden_inputs[name] = value
return hidden_inputs
def _form_hidden_inputs(self, form_id, html):
form = self._search_regex(
r'(?is)<form[^>]+?id=(["\'])%s\1[^>]*>(?P<form>.+?)</form>' % form_id,
html, '%s form' % form_id, group='form')
return self._hidden_inputs(form)
class FormatSort:
regex = r' *((?P<reverse>\+)?(?P<field>[a-zA-Z0-9_]+)((?P<separator>[~:])(?P<limit>.*?))?)? *$'
default = ('hidden', 'aud_or_vid', 'hasvid', 'ie_pref', 'lang', 'quality',
'res', 'fps', 'hdr:12', 'codec:vp9.2', 'size', 'br', 'asr',
'proto', 'ext', 'hasaud', 'source', 'id') # These must not be aliases
ytdl_default = ('hasaud', 'lang', 'quality', 'tbr', 'filesize', 'vbr',
'height', 'width', 'proto', 'vext', 'abr', 'aext',
'fps', 'fs_approx', 'source', 'id')
settings = {
'vcodec': {'type': 'ordered', 'regex': True,
'order': ['av0?1', 'vp0?9.2', 'vp0?9', '[hx]265|he?vc?', '[hx]264|avc', 'vp0?8', 'mp4v|h263', 'theora', '', None, 'none']},
'acodec': {'type': 'ordered', 'regex': True,
'order': ['[af]lac', 'wav|aiff', 'opus', 'vorbis|ogg', 'aac', 'mp?4a?', 'mp3', 'e-?a?c-?3', 'ac-?3', 'dts', '', None, 'none']},
'hdr': {'type': 'ordered', 'regex': True, 'field': 'dynamic_range',
'order': ['dv', '(hdr)?12', r'(hdr)?10\+', '(hdr)?10', 'hlg', '', 'sdr', None]},
'proto': {'type': 'ordered', 'regex': True, 'field': 'protocol',
'order': ['(ht|f)tps', '(ht|f)tp$', 'm3u8.*', '.*dash', 'websocket_frag', 'rtmpe?', '', 'mms|rtsp', 'ws|websocket', 'f4']},
'vext': {'type': 'ordered', 'field': 'video_ext',
'order': ('mp4', 'webm', 'flv', '', 'none'),
'order_free': ('webm', 'mp4', 'flv', '', 'none')},
'aext': {'type': 'ordered', 'field': 'audio_ext',
'order': ('m4a', 'aac', 'mp3', 'ogg', 'opus', 'webm', '', 'none'),
'order_free': ('opus', 'ogg', 'webm', 'm4a', 'mp3', 'aac', '', 'none')},
'hidden': {'visible': False, 'forced': True, 'type': 'extractor', 'max': -1000},
'aud_or_vid': {'visible': False, 'forced': True, 'type': 'multiple',
'field': ('vcodec', 'acodec'),
'function': lambda it: int(any(v != 'none' for v in it))},
'ie_pref': {'priority': True, 'type': 'extractor'},
'hasvid': {'priority': True, 'field': 'vcodec', 'type': 'boolean', 'not_in_list': ('none',)},
'hasaud': {'field': 'acodec', 'type': 'boolean', 'not_in_list': ('none',)},
'lang': {'convert': 'float', 'field': 'language_preference', 'default': -1},
'quality': {'convert': 'float', 'default': -1},
'filesize': {'convert': 'bytes'},
'fs_approx': {'convert': 'bytes', 'field': 'filesize_approx'},
'id': {'convert': 'string', 'field': 'format_id'},
'height': {'convert': 'float_none'},
'width': {'convert': 'float_none'},
'fps': {'convert': 'float_none'},
'tbr': {'convert': 'float_none'},
'vbr': {'convert': 'float_none'},
'abr': {'convert': 'float_none'},
'asr': {'convert': 'float_none'},
'source': {'convert': 'float', 'field': 'source_preference', 'default': -1},
'codec': {'type': 'combined', 'field': ('vcodec', 'acodec')},
'br': {'type': 'combined', 'field': ('tbr', 'vbr', 'abr'), 'same_limit': True},
'size': {'type': 'combined', 'same_limit': True, 'field': ('filesize', 'fs_approx')},
'ext': {'type': 'combined', 'field': ('vext', 'aext')},
'res': {'type': 'multiple', 'field': ('height', 'width'),
'function': lambda it: (lambda l: min(l) if l else 0)(tuple(filter(None, it)))},
# For compatibility with youtube-dl
'format_id': {'type': 'alias', 'field': 'id'},
'preference': {'type': 'alias', 'field': 'ie_pref'},
'language_preference': {'type': 'alias', 'field': 'lang'},
'source_preference': {'type': 'alias', 'field': 'source'},
'protocol': {'type': 'alias', 'field': 'proto'},
'filesize_approx': {'type': 'alias', 'field': 'fs_approx'},
# Deprecated
'dimension': {'type': 'alias', 'field': 'res', 'deprecated': True},
'resolution': {'type': 'alias', 'field': 'res', 'deprecated': True},
'extension': {'type': 'alias', 'field': 'ext', 'deprecated': True},
'bitrate': {'type': 'alias', 'field': 'br', 'deprecated': True},
'total_bitrate': {'type': 'alias', 'field': 'tbr', 'deprecated': True},
'video_bitrate': {'type': 'alias', 'field': 'vbr', 'deprecated': True},
'audio_bitrate': {'type': 'alias', 'field': 'abr', 'deprecated': True},
'framerate': {'type': 'alias', 'field': 'fps', 'deprecated': True},
'filesize_estimate': {'type': 'alias', 'field': 'size', 'deprecated': True},
'samplerate': {'type': 'alias', 'field': 'asr', 'deprecated': True},
'video_ext': {'type': 'alias', 'field': 'vext', 'deprecated': True},
'audio_ext': {'type': 'alias', 'field': 'aext', 'deprecated': True},
'video_codec': {'type': 'alias', 'field': 'vcodec', 'deprecated': True},
'audio_codec': {'type': 'alias', 'field': 'acodec', 'deprecated': True},
'video': {'type': 'alias', 'field': 'hasvid', 'deprecated': True},
'has_video': {'type': 'alias', 'field': 'hasvid', 'deprecated': True},
'audio': {'type': 'alias', 'field': 'hasaud', 'deprecated': True},
'has_audio': {'type': 'alias', 'field': 'hasaud', 'deprecated': True},
'extractor': {'type': 'alias', 'field': 'ie_pref', 'deprecated': True},
'extractor_preference': {'type': 'alias', 'field': 'ie_pref', 'deprecated': True},
def __init__(self, ie, field_preference):
self._order = []
self.ydl = ie._downloader
self.evaluate_params(self.ydl.params, field_preference)
if ie.get_param('verbose'):
def _get_field_setting(self, field, key):
if field not in self.settings:
if key in ('forced', 'priority'):
return False
f'Using arbitrary fields ({field}) for format sorting is deprecated '
'and may be removed in a future version')
self.settings[field] = {}
propObj = self.settings[field]
if key not in propObj:
type = propObj.get('type')
if key == 'field':
default = 'preference' if type == 'extractor' else (field,) if type in ('combined', 'multiple') else field
elif key == 'convert':
default = 'order' if type == 'ordered' else 'float_string' if field else 'ignore'
default = {'type': 'field', 'visible': True, 'order': [], 'not_in_list': (None,)}.get(key, None)
propObj[key] = default
return propObj[key]
def _resolve_field_value(self, field, value, convertNone=False):
if value is None:
if not convertNone:
return None
value = value.lower()
conversion = self._get_field_setting(field, 'convert')
if conversion == 'ignore':
return None
if conversion == 'string':
return value
elif conversion == 'float_none':
return float_or_none(value)
elif conversion == 'bytes':
return FileDownloader.parse_bytes(value)
elif conversion == 'order':
order_list = (self._use_free_order and self._get_field_setting(field, 'order_free')) or self._get_field_setting(field, 'order')
use_regex = self._get_field_setting(field, 'regex')
list_length = len(order_list)
empty_pos = order_list.index('') if '' in order_list else list_length + 1
if use_regex and value is not None:
for i, regex in enumerate(order_list):
if regex and re.match(regex, value):
return list_length - i
return list_length - empty_pos # not in list
else: # not regex or value = None
return list_length - (order_list.index(value) if value in order_list else empty_pos)
if value.isnumeric():
return float(value)
self.settings[field]['convert'] = 'string'
return value
def evaluate_params(self, params, sort_extractor):
self._use_free_order = params.get('prefer_free_formats', False)
self._sort_user = params.get('format_sort', [])
self._sort_extractor = sort_extractor
def add_item(field, reverse, closest, limit_text):
field = field.lower()
if field in self._order:
limit = self._resolve_field_value(field, limit_text)
data = {
'reverse': reverse,
'closest': False if limit is None else closest,
'limit_text': limit_text,
'limit': limit}
if field in self.settings:
self.settings[field] = data
sort_list = (
tuple(field for field in self.default if self._get_field_setting(field, 'forced'))
+ (tuple() if params.get('format_sort_force', False)
else tuple(field for field in self.default if self._get_field_setting(field, 'priority')))
+ tuple(self._sort_user) + tuple(sort_extractor) + self.default)
for item in sort_list:
match = re.match(self.regex, item)
if match is None:
raise ExtractorError('Invalid format sort string "%s" given by extractor' % item)
field ='field')
if field is None:
if self._get_field_setting(field, 'type') == 'alias':
alias, field = field, self._get_field_setting(field, 'field')
if self._get_field_setting(alias, 'deprecated'):
f'Format sorting alias {alias} is deprecated '
f'and may be removed in a future version. Please use {field} instead')
reverse ='reverse') is not None
closest ='separator') == '~'
limit_text ='limit')
has_limit = limit_text is not None
has_multiple_fields = self._get_field_setting(field, 'type') == 'combined'
has_multiple_limits = has_limit and has_multiple_fields and not self._get_field_setting(field, 'same_limit')
fields = self._get_field_setting(field, 'field') if has_multiple_fields else (field,)
limits = limit_text.split(':') if has_multiple_limits else (limit_text,) if has_limit else tuple()
limit_count = len(limits)
for (i, f) in enumerate(fields):
add_item(f, reverse, closest,
limits[i] if i < limit_count
else limits[0] if has_limit and not has_multiple_limits
else None)
def print_verbose_info(self, write_debug):
if self._sort_user:
write_debug('Sort order given by user: %s' % ', '.join(self._sort_user))
if self._sort_extractor:
write_debug('Sort order given by extractor: %s' % ', '.join(self._sort_extractor))
write_debug('Formats sorted by: %s' % ', '.join(['%s%s%s' % (
'+' if self._get_field_setting(field, 'reverse') else '', field,
'%s%s(%s)' % ('~' if self._get_field_setting(field, 'closest') else ':',
self._get_field_setting(field, 'limit_text'),
self._get_field_setting(field, 'limit'))
if self._get_field_setting(field, 'limit_text') is not None else '')
for field in self._order if self._get_field_setting(field, 'visible')]))
def _calculate_field_preference_from_value(self, format, field, type, value):
reverse = self._get_field_setting(field, 'reverse')
closest = self._get_field_setting(field, 'closest')
limit = self._get_field_setting(field, 'limit')
if type == 'extractor':
maximum = self._get_field_setting(field, 'max')
if value is None or (maximum is not None and value >= maximum):
value = -1
elif type == 'boolean':
in_list = self._get_field_setting(field, 'in_list')
not_in_list = self._get_field_setting(field, 'not_in_list')
value = 0 if ((in_list is None or value in in_list) and (not_in_list is None or value not in not_in_list)) else -1
elif type == 'ordered':
value = self._resolve_field_value(field, value, True)
# try to convert to number
val_num = float_or_none(value, default=self._get_field_setting(field, 'default'))
is_num = self._get_field_setting(field, 'convert') != 'string' and val_num is not None
if is_num:
value = val_num
return ((-10, 0) if value is None
else (1, value, 0) if not is_num # if a field has mixed strings and numbers, strings are sorted higher
else (0, -abs(value - limit), value - limit if reverse else limit - value) if closest
else (0, value, 0) if not reverse and (limit is None or value <= limit)
else (0, -value, 0) if limit is None or (reverse and value == limit) or value > limit
else (-1, value, 0))
def _calculate_field_preference(self, format, field):
type = self._get_field_setting(field, 'type') # extractor, boolean, ordered, field, multiple
get_value = lambda f: format.get(self._get_field_setting(f, 'field'))
if type == 'multiple':
type = 'field' # Only 'field' is allowed in multiple for now
actual_fields = self._get_field_setting(field, 'field')
value = self._get_field_setting(field, 'function')(get_value(f) for f in actual_fields)
value = get_value(field)
return self._calculate_field_preference_from_value(format, field, type, value)
def calculate_preference(self, format):
# Determine missing protocol
if not format.get('protocol'):
format['protocol'] = determine_protocol(format)
# Determine missing ext
if not format.get('ext') and 'url' in format:
format['ext'] = determine_ext(format['url'])
if format.get('vcodec') == 'none':
format['audio_ext'] = format['ext'] if format.get('acodec') != 'none' else 'none'
format['video_ext'] = 'none'
format['video_ext'] = format['ext']
format['audio_ext'] = 'none'
# if format.get('preference') is None and format.get('ext') in ('f4f', 'f4m'): # Not supported?
# format['preference'] = -1000
# Determine missing bitrates
if format.get('tbr') is None:
if format.get('vbr') is not None and format.get('abr') is not None:
format['tbr'] = format.get('vbr', 0) + format.get('abr', 0)
if format.get('vcodec') != 'none' and format.get('vbr') is None:
format['vbr'] = format.get('tbr') - format.get('abr', 0)
if format.get('acodec') != 'none' and format.get('abr') is None:
format['abr'] = format.get('tbr') - format.get('vbr', 0)
return tuple(self._calculate_field_preference(format, field) for field in self._order)
def _sort_formats(self, formats, field_preference=[]):
if not formats:
formats.sort(key=self.FormatSort(self, field_preference).calculate_preference)
def _check_formats(self, formats, video_id):
if formats:
formats[:] = filter(
lambda f: self._is_valid_url(
f['url'], video_id,
item='%s video format' % f.get('format_id') if f.get('format_id') else 'video'),
def _remove_duplicate_formats(formats):
format_urls = set()
unique_formats = []
for f in formats:
if f['url'] not in format_urls:
formats[:] = unique_formats
def _is_valid_url(self, url, video_id, item='video', headers={}):
url = self._proto_relative_url(url, scheme='http:')
# For now assume non HTTP(S) URLs always valid
if not (url.startswith('http://') or url.startswith('https://')):
return True
self._request_webpage(url, video_id, 'Checking %s URL' % item, headers=headers)
return True
except ExtractorError as e:
'%s: %s URL is invalid, skipping: %s'
% (video_id, item, error_to_compat_str(e.cause)))
return False
def http_scheme(self):
""" Either "http:" or "https:", depending on the user's preferences """
return (
if self.get_param('prefer_insecure', False)
else 'https:')
def _proto_relative_url(self, url, scheme=None):
if url is None:
return url
if url.startswith('//'):
if scheme is None:
scheme = self.http_scheme()
return scheme + url
return url
def _sleep(self, timeout, video_id, msg_template=None):
if msg_template is None:
msg_template = '%(video_id)s: Waiting for %(timeout)s seconds'
msg = msg_template % {'video_id': video_id, 'timeout': timeout}
def _extract_f4m_formats(self, manifest_url, video_id, preference=None, quality=None, f4m_id=None,
transform_source=lambda s: fix_xml_ampersands(s).strip(),
fatal=True, m3u8_id=None, data=None, headers={}, query={}):
res = self._download_xml_handle(
manifest_url, video_id, 'Downloading f4m manifest',
'Unable to download f4m manifest',
# Some manifests may be malformed, e.g. prosiebensat1 generated manifests
# (see
fatal=fatal, data=data, headers=headers, query=query)
if res is False:
return []
manifest, urlh = res
manifest_url = urlh.geturl()
return self._parse_f4m_formats(
manifest, manifest_url, video_id, preference=preference, quality=quality, f4m_id=f4m_id,
transform_source=transform_source, fatal=fatal, m3u8_id=m3u8_id)
def _parse_f4m_formats(self, manifest, manifest_url, video_id, preference=None, quality=None, f4m_id=None,
transform_source=lambda s: fix_xml_ampersands(s).strip(),
fatal=True, m3u8_id=None):
if not isinstance(manifest, xml.etree.ElementTree.Element) and not fatal:
return []
# currently yt-dlp cannot decode the playerVerificationChallenge as Akamai uses Adobe Alchemy
akamai_pv = manifest.find('{}pv-2.0')
if akamai_pv is not None and ';' in akamai_pv.text:
playerVerificationChallenge = akamai_pv.text.split(';')[0]
if playerVerificationChallenge.strip() != '':
return []
formats = []
manifest_version = '1.0'
media_nodes = manifest.findall('{}media')
if not media_nodes:
manifest_version = '2.0'
media_nodes = manifest.findall('{}media')
# Remove unsupported DRM protected media from final formats
# rendition (see
media_nodes = remove_encrypted_media(media_nodes)
if not media_nodes:
return formats
manifest_base_url = get_base_url(manifest)
bootstrap_info = xpath_element(
manifest, ['{}bootstrapInfo', '{}bootstrapInfo'],
'bootstrap info', default=None)
vcodec = None
mime_type = xpath_text(
manifest, ['{}mimeType', '{}mimeType'],
'base URL', default=None)
if mime_type and mime_type.startswith('audio/'):
vcodec = 'none'
for i, media_el in enumerate(media_nodes):
tbr = int_or_none(media_el.attrib.get('bitrate'))
width = int_or_none(media_el.attrib.get('width'))
height = int_or_none(media_el.attrib.get('height'))
format_id = join_nonempty(f4m_id, tbr or i)
# If <bootstrapInfo> is present, the specified f4m is a
# stream-level manifest, and only set-level manifests may refer to
# external resources. See section 11.4 and section 4 of F4M spec
if bootstrap_info is None:
media_url = None
# @href is introduced in 2.0, see section 11.6 of F4M spec
if manifest_version == '2.0':
media_url = media_el.attrib.get('href')
if media_url is None:
media_url = media_el.attrib.get('url')
if not media_url:
manifest_url = (
media_url if media_url.startswith('http://') or media_url.startswith('https://')
else ((manifest_base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url))
# If media_url is itself a f4m manifest do the recursive extraction
# since bitrates in parent manifest (this one) and media_url manifest
# may differ leading to inability to resolve the format by requested
# bitrate in f4m downloader
ext = determine_ext(manifest_url)
if ext == 'f4m':
f4m_formats = self._extract_f4m_formats(
manifest_url, video_id, preference=preference, quality=quality, f4m_id=f4m_id,
transform_source=transform_source, fatal=fatal)
# Sometimes stream-level manifest contains single media entry that
# does not contain any quality metadata (e.g.
# At the same time parent's media entry in set-level manifest may
# contain it. We will copy it from parent in such cases.
if len(f4m_formats) == 1:
f = f4m_formats[0]
'tbr': f.get('tbr') or tbr,
'width': f.get('width') or width,
'height': f.get('height') or height,
'format_id': f.get('format_id') if not tbr else format_id,
'vcodec': vcodec,
elif ext == 'm3u8':
manifest_url, video_id, 'mp4', preference=preference,
quality=quality, m3u8_id=m3u8_id, fatal=fatal))
'format_id': format_id,
'url': manifest_url,
'manifest_url': manifest_url,
'ext': 'flv' if bootstrap_info is not None else None,
'protocol': 'f4m',
'tbr': tbr,
'width': width,
'height': height,
'vcodec': vcodec,
'preference': preference,
'quality': quality,
return formats
def _m3u8_meta_format(self, m3u8_url, ext=None, preference=None, quality=None, m3u8_id=None):
return {
'format_id': join_nonempty(m3u8_id, 'meta'),
'url': m3u8_url,
'ext': ext,
'protocol': 'm3u8',
'preference': preference - 100 if preference else -100,
'quality': quality,
'resolution': 'multiple',
'format_note': 'Quality selection URL',
def _report_ignoring_subs(self, name):
f'Ignoring subtitle tracks found in the {name} manifest; '
'if any subtitle tracks are missing,'
), only_once=True)
def _extract_m3u8_formats(self, *args, **kwargs):
fmts, subs = self._extract_m3u8_formats_and_subtitles(*args, **kwargs)
if subs:
return fmts
def _extract_m3u8_formats_and_subtitles(
self, m3u8_url, video_id, ext=None, entry_protocol='m3u8_native',
preference=None, quality=None, m3u8_id=None, note=None,
errnote=None, fatal=True, live=False, data=None, headers={},
res = self._download_webpage_handle(
m3u8_url, video_id,
note='Downloading m3u8 information' if note is None else note,
errnote='Failed to download m3u8 information' if errnote is None else errnote,
fatal=fatal, data=data, headers=headers, query=query)
if res is False:
return [], {}
m3u8_doc, urlh = res
m3u8_url = urlh.geturl()
return self._parse_m3u8_formats_and_subtitles(
m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol,
preference=preference, quality=quality, m3u8_id=m3u8_id,
note=note, errnote=errnote, fatal=fatal, live=live, data=data,
headers=headers, query=query, video_id=video_id)
def _parse_m3u8_formats_and_subtitles(
self, m3u8_doc, m3u8_url=None, ext=None, entry_protocol='m3u8_native',
preference=None, quality=None, m3u8_id=None, live=False, note=None,
errnote=None, fatal=True, data=None, headers={}, query={},
formats, subtitles = [], {}
has_drm ='|'.join([
r'#EXT-X-FAXS-CM:', # Adobe Flash Access
r'#EXT-X-(?:SESSION-)?KEY:.*?URI="skd://', # Apple FairPlay
]), m3u8_doc)
def format_url(url):
return url if re.match(r'^https?://', url) else urllib.parse.urljoin(m3u8_url, url)
if self.get_param('hls_split_discontinuity', False):
def _extract_m3u8_playlist_indices(manifest_url=None, m3u8_doc=None):
if not m3u8_doc:
if not manifest_url:
return []
m3u8_doc = self._download_webpage(
manifest_url, video_id, fatal=fatal, data=data, headers=headers,
note=False, errnote='Failed to download m3u8 playlist information')
if m3u8_doc is False:
return []
return range(1 + sum(line.startswith('#EXT-X-DISCONTINUITY') for line in m3u8_doc.splitlines()))
def _extract_m3u8_playlist_indices(*args, **kwargs):
return [None]
# References:
# 1.
# 2.
# 3.
# We should try extracting formats only from master playlists [1, 4.3.4],
# i.e. playlists that describe available qualities. On the other hand
# media playlists [1, 4.3.3] should be returned as is since they contain
# just the media without qualities renditions.
# Fortunately, master playlist can be easily distinguished from media
# playlist based on particular tags availability. As of [1, 4.3.3, 4.3.4]
# master playlist tags MUST NOT appear in a media playlist and vice versa.
# As of [1,] #EXT-X-TARGETDURATION tag is REQUIRED for every
# media playlist and MUST NOT appear in master playlist thus we can
# clearly detect media playlist with this criterion.
if '#EXT-X-TARGETDURATION' in m3u8_doc: # media playlist, return as is
formats = [{
'format_id': join_nonempty(m3u8_id, idx),
'format_index': idx,
'url': m3u8_url or encode_data_uri(m3u8_doc.encode('utf-8'), 'application/x-mpegurl'),
'ext': ext,
'protocol': entry_protocol,
'preference': preference,
'quality': quality,
'has_drm': has_drm,
} for idx in _extract_m3u8_playlist_indices(m3u8_doc=m3u8_doc)]
return formats, subtitles
groups = {}
last_stream_inf = {}
def extract_media(x_media_line):
media = parse_m3u8_attributes(x_media_line)
# As per [1,] TYPE, GROUP-ID and NAME are REQUIRED
media_type, group_id, name = media.get('TYPE'), media.get('GROUP-ID'), media.get('NAME')
if not (media_type and group_id and name):
groups.setdefault(group_id, []).append(media)
# <>
if media_type == 'SUBTITLES':
# According to RFC 8216 §, URI is REQUIRED in the
# EXT-X-MEDIA tag if the media type is SUBTITLES.
# However, lack of URI has been spotted in the wild.