From e9efb99f66b57a2e25937dff3dd834b5e6c381e3 Mon Sep 17 00:00:00 2001 From: Sonic <47434066+TwoThousandHedgehogs@users.noreply.github.com> Date: Fri, 24 Dec 2021 06:49:33 -0500 Subject: [PATCH] [dropout] Add extractor (#2045) Authored-by: TwoThousandHedgehogs, pukkandan --- yt_dlp/extractor/dropout.py | 212 +++++++++++++++++++++++++++++++++ yt_dlp/extractor/extractors.py | 4 + 2 files changed, 216 insertions(+) create mode 100644 yt_dlp/extractor/dropout.py diff --git a/yt_dlp/extractor/dropout.py b/yt_dlp/extractor/dropout.py new file mode 100644 index 000000000..a7442d8f0 --- /dev/null +++ b/yt_dlp/extractor/dropout.py @@ -0,0 +1,212 @@ +# coding: utf-8 +from .common import InfoExtractor +from .vimeo import VHXEmbedIE +from ..utils import ( + clean_html, + ExtractorError, + get_element_by_class, + get_element_by_id, + get_elements_by_class, + int_or_none, + join_nonempty, + unified_strdate, + urlencode_postdata, +) + + +class DropoutIE(InfoExtractor): + _LOGIN_URL = 'https://www.dropout.tv/login' + _NETRC_MACHINE = 'dropout' + + _VALID_URL = r'https?://(?:www\.)?dropout\.tv/(?:[^/]+/)*videos/(?P[^/]+)/?$' + _TESTS = [ + { + 'url': 'https://www.dropout.tv/game-changer/season:2/videos/yes-or-no', + 'note': 'Episode in a series', + 'md5': '5e000fdfd8d8fa46ff40456f1c2af04a', + 'info_dict': { + 'id': '738153', + 'display_id': 'yes-or-no', + 'ext': 'mp4', + 'title': 'Yes or No', + 'description': 'Ally, Brennan, and Zac are asked a simple question, but is there a correct answer?', + 'release_date': '20200508', + 'thumbnail': 'https://vhx.imgix.net/chuncensoredstaging/assets/351e3f24-c4a3-459a-8b79-dc80f1e5b7fd.jpg', + 'series': 'Game Changer', + 'season_number': 2, + 'season': 'Season 2', + 'episode_number': 6, + 'episode': 'Yes or No', + 'duration': 1180, + 'uploader_id': 'user80538407', + 'uploader_url': 'https://vimeo.com/user80538407', + 'uploader': 'OTT Videos' + }, + 'expected_warnings': ['Ignoring subtitle tracks found in the HLS manifest'] + }, + { + 'url': 'https://www.dropout.tv/dimension-20-fantasy-high/season:1/videos/episode-1', + 'note': 'Episode in a series (missing release_date)', + 'md5': '712caf7c191f1c47c8f1879520c2fa5c', + 'info_dict': { + 'id': '320562', + 'display_id': 'episode-1', + 'ext': 'mp4', + 'title': 'The Beginning Begins', + 'description': 'The cast introduces their PCs, including a neurotic elf, a goblin PI, and a corn-worshipping cleric.', + 'thumbnail': 'https://vhx.imgix.net/chuncensoredstaging/assets/4421ed0d-f630-4c88-9004-5251b2b8adfa.jpg', + 'series': 'Dimension 20: Fantasy High', + 'season_number': 1, + 'season': 'Season 1', + 'episode_number': 1, + 'episode': 'The Beginning Begins', + 'duration': 6838, + 'uploader_id': 'user80538407', + 'uploader_url': 'https://vimeo.com/user80538407', + 'uploader': 'OTT Videos' + }, + 'expected_warnings': ['Ignoring subtitle tracks found in the HLS manifest'] + }, + { + 'url': 'https://www.dropout.tv/videos/misfits-magic-holiday-special', + 'note': 'Episode not in a series', + 'md5': 'c30fa18999c5880d156339f13c953a26', + 'info_dict': { + 'id': '1915774', + 'display_id': 'misfits-magic-holiday-special', + 'ext': 'mp4', + 'title': 'Misfits & Magic Holiday Special', + 'description': 'The magical misfits spend Christmas break at Gowpenny, with an unwelcome visitor.', + 'release_date': '20211215', + 'thumbnail': 'https://vhx.imgix.net/chuncensoredstaging/assets/d91ea8a6-b250-42ed-907e-b30fb1c65176-8e24b8e5.jpg', + 'duration': 11698, + 'uploader_id': 'user80538407', + 'uploader_url': 'https://vimeo.com/user80538407', + 'uploader': 'OTT Videos' + }, + 'expected_warnings': ['Ignoring subtitle tracks found in the HLS manifest'] + } + ] + + def _get_authenticity_token(self, display_id): + signin_page = self._download_webpage( + self._LOGIN_URL, display_id, note='Getting authenticity token') + return self._html_search_regex( + r'name=["\']authenticity_token["\'] value=["\'](.+?)["\']', + signin_page, 'authenticity_token') + + def _login(self, display_id): + username, password = self._get_login_info() + if not (username and password): + self.raise_login_required(method='password') + + response = self._download_webpage( + self._LOGIN_URL, display_id, note='Logging in', data=urlencode_postdata({ + 'email': username, + 'password': password, + 'authenticity_token': self._get_authenticity_token(display_id), + 'utf8': True + })) + + user_has_subscription = self._search_regex( + r'user_has_subscription:\s*["\'](.+?)["\']', response, 'subscription status', default='none') + if user_has_subscription.lower() == 'true': + return response + elif user_has_subscription.lower() == 'false': + raise ExtractorError('Account is not subscribed') + else: + raise ExtractorError('Incorrect username/password') + + def _real_extract(self, url): + display_id = self._match_id(url) + try: + self._login(display_id) + webpage = self._download_webpage(url, display_id, note='Downloading video webpage') + finally: + self._download_webpage('https://www.dropout.tv/logout', display_id, note='Logging out') + + embed_url = self._search_regex(r'embed_url:\s*["\'](.+?)["\']', webpage, 'embed url') + thumbnail = self._og_search_thumbnail(webpage) + watch_info = get_element_by_id('watch-info', webpage) or '' + + title = clean_html(get_element_by_class('video-title', watch_info)) + season_episode = get_element_by_class( + 'site-font-secondary-color', get_element_by_class('text', watch_info)) + episode_number = int_or_none(self._search_regex( + r'Episode (\d+)', season_episode or '', 'episode', default=None)) + + return { + '_type': 'url_transparent', + 'ie_key': VHXEmbedIE.ie_key(), + 'url': embed_url, + 'id': self._search_regex(r'embed.vhx.tv/videos/(.+?)\?', embed_url, 'id'), + 'display_id': display_id, + 'title': title, + 'description': self._html_search_meta('description', webpage, fatal=False), + 'thumbnail': thumbnail.split('?')[0] if thumbnail else None, # Ignore crop/downscale + 'series': clean_html(get_element_by_class('series-title', watch_info)), + 'episode_number': episode_number, + 'episode': title if episode_number else None, + 'season_number': int_or_none(self._search_regex( + r'Season (\d+),', season_episode or '', 'season', default=None)), + 'release_date': unified_strdate(self._search_regex( + r'data-meta-field-name=["\']release_dates["\'] data-meta-field-value=["\'](.+?)["\']', + watch_info, 'release date', default=None)), + } + + +class DropoutSeasonIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?dropout\.tv/(?P[^\/$&?#]+)(?:/?$|/season:[0-9]+/?$)' + _TESTS = [ + { + 'url': 'https://www.dropout.tv/dimension-20-fantasy-high/season:1', + 'note': 'Multi-season series with the season in the url', + 'playlist_count': 17, + 'info_dict': { + 'id': 'dimension-20-fantasy-high-season-1', + 'title': 'Dimension 20 Fantasy High - Season 1' + } + }, + { + 'url': 'https://www.dropout.tv/dimension-20-fantasy-high', + 'note': 'Multi-season series with the season not in the url', + 'playlist_count': 17, + 'info_dict': { + 'id': 'dimension-20-fantasy-high-season-1', + 'title': 'Dimension 20 Fantasy High - Season 1' + } + }, + { + 'url': 'https://www.dropout.tv/dimension-20-shriek-week', + 'note': 'Single-season series', + 'playlist_count': 4, + 'info_dict': { + 'id': 'dimension-20-shriek-week-season-1', + 'title': 'Dimension 20 Shriek Week - Season 1' + } + } + ] + + def _real_extract(self, url): + season_id = self._match_id(url) + season_title = season_id.replace('-', ' ').title() + webpage = self._download_webpage(url, season_id) + + entries = [ + self.url_result( + url=self._search_regex(r']+selected>([^<]+)', + seasons, 'current_season', default='').strip() + + return { + '_type': 'playlist', + 'id': join_nonempty(season_id, current_season.lower().replace(' ', '-')), + 'title': join_nonempty(season_title, current_season, delim=' - '), + 'entries': entries + } diff --git a/yt_dlp/extractor/extractors.py b/yt_dlp/extractor/extractors.py index 0b359a253..2a907bc57 100644 --- a/yt_dlp/extractor/extractors.py +++ b/yt_dlp/extractor/extractors.py @@ -385,6 +385,10 @@ from .disney import DisneyIE from .dispeak import DigitallySpeakingIE from .doodstream import DoodStreamIE from .dropbox import DropboxIE +from .dropout import ( + DropoutSeasonIE, + DropoutIE +) from .dw import ( DWIE, DWArticleIE,