From 86587204142fca8081f02a23145633d9bb268022 Mon Sep 17 00:00:00 2001 From: DTrombett Date: Sun, 29 Dec 2024 19:05:40 +0100 Subject: [PATCH 1/3] [ie/plutotv] Fix extractor --- yt_dlp/extractor/plutotv.py | 257 ++++++++++++++++++------------------ 1 file changed, 125 insertions(+), 132 deletions(-) diff --git a/yt_dlp/extractor/plutotv.py b/yt_dlp/extractor/plutotv.py index 234ee987b6..98154d0337 100644 --- a/yt_dlp/extractor/plutotv.py +++ b/yt_dlp/extractor/plutotv.py @@ -3,29 +3,11 @@ import urllib.parse import uuid from .common import InfoExtractor -from ..utils import ( - ExtractorError, - float_or_none, - int_or_none, - try_get, - url_or_none, -) +from ..utils import ExtractorError, float_or_none, int_or_none, traverse_obj -class PlutoTVIE(InfoExtractor): - _WORKING = False - _VALID_URL = r'''(?x) - https?://(?:www\.)?pluto\.tv(?:/[^/]+)?/on-demand - /(?Pmovies|series) - /(?P[^/]+) - (?: - (?:/seasons?/(?P\d+))? - (?:/episode/(?P[^/]+))? - )? - /?(?:$|[#?])''' - - _INFO_URL = 'https://service-vod.clusters.pluto.tv/v3/vod/slugs/' - _INFO_QUERY_PARAMS = { +class PlutoTVBase(InfoExtractor): + _START_QUERY = { 'appName': 'web', 'appVersion': 'na', 'clientID': str(uuid.uuid1()), @@ -35,66 +17,80 @@ class PlutoTVIE(InfoExtractor): 'deviceModel': 'web', 'deviceType': 'web', 'deviceVersion': 'unknown', - 'sid': str(uuid.uuid1()), } - _TESTS = [ - { - 'url': 'https://pluto.tv/on-demand/series/i-love-money/season/2/episode/its-in-the-cards-2009-2-3', - 'md5': 'ebcdd8ed89aaace9df37924f722fd9bd', - 'info_dict': { - 'id': '5de6c598e9379ae4912df0a8', - 'ext': 'mp4', - 'title': 'It\'s In The Cards', - 'episode': 'It\'s In The Cards', - 'description': 'The teams face off against each other in a 3-on-2 soccer showdown. Strategy comes into play, though, as each team gets to select their opposing teams’ two defenders.', - 'series': 'I Love Money', - 'season_number': 2, - 'episode_number': 3, - 'duration': 3600, - }, - }, { - 'url': 'https://pluto.tv/on-demand/series/i-love-money/season/1/', - 'playlist_count': 11, - 'info_dict': { - 'id': '5de6c582e9379ae4912dedbd', - 'title': 'I Love Money - Season 1', - }, - }, { - 'url': 'https://pluto.tv/on-demand/series/i-love-money/', - 'playlist_count': 26, - 'info_dict': { - 'id': '5de6c582e9379ae4912dedbd', - 'title': 'I Love Money', - }, - }, { - 'url': 'https://pluto.tv/on-demand/movies/arrival-2015-1-1', - 'md5': '3cead001d317a018bf856a896dee1762', - 'info_dict': { - 'id': '5e83ac701fa6a9001bb9df24', - 'ext': 'mp4', - 'title': 'Arrival', - 'description': 'When mysterious spacecraft touch down across the globe, an elite team - led by expert translator Louise Banks (Academy Award® nominee Amy Adams) – races against time to decipher their intent.', - 'duration': 9000, - }, - }, { - 'url': 'https://pluto.tv/en/on-demand/series/manhunters-fugitive-task-force/seasons/1/episode/third-times-the-charm-1-1', - 'only_matching': True, - }, { - 'url': 'https://pluto.tv/it/on-demand/series/csi-vegas/episode/legacy-2021-1-1', - 'only_matching': True, + + def _extract_formats(self, start, element): + formats, subtitles = self._extract_m3u8_formats_and_subtitles(f"{start['servers']['stitcher']}/v2{element['stitched']['path']}?{start['stitcherParams']}&jwt={start['sessionToken']}", element.get('id') or element.get('_id')) + for f in formats: + f['url'] += f"&jwt={start['sessionToken']}" + f.setdefault('vcodec', 'avc1.64001f') + f.setdefault('acodec', 'mp4a.40.2') + f.setdefault('fps', 30) + return formats, subtitles + + +class PlutoTVIE(PlutoTVBase): + _VALID_URL = r'''(?x) + https?://(?:www\.)?pluto\.tv(?:/[^/]+)?/on-demand + /(movies|series) + /(?P[^/]+) + (?: + (?:/seasons?/(?P\d+))? + (?:/episode/(?P[^/]+))? + )? + /?(?:$|[#?])''' + _TESTS = [{ + 'url': 'https://pluto.tv/it/on-demand/movies/6246b0adef11000014d220c3', + 'md5': '5efe37ad6c1085a4ad4684b9b82cb1c1', + 'info_dict': { + 'id': '6246b0adef11000014d220c3', + 'ext': 'mp4', + 'episode_id': '6246b0adef11000014d220c3', + 'description': 'md5:c9a412d330d3d73a527e9ba981c0ddb8', + 'episode': 'Non Bussate A Quella Porta', + 'thumbnail': 'http://images.pluto.tv/episodes/6246b0adef11000014d220c3/poster.jpg?fm=png&q=100', + 'display_id': 'dont-knock-twice-it-2016-1-1', + 'genres': ['Horror'], + 'title': 'Non Bussate A Quella Porta', + 'duration': 5940, + }, + }, { + 'url': 'https://pluto.tv/on-demand/movies/6246b0adef11000014d220c3', + 'only_matching': True, + }, { + 'url': 'https://pluto.tv/on-demand/series/6655b0c5cceea000134aee27', + 'info_dict': { + 'id': '6655b0c5cceea000134aee27', + 'title': 'Mission Impossible', }, - { - 'url': 'https://pluto.tv/en/on-demand/movies/attack-of-the-killer-tomatoes-1977-1-1-ptv1', - 'md5': '7db56369c0da626a32d505ec6eb3f89f', - 'info_dict': { - 'id': '5b190c7bb0875c36c90c29c4', - 'ext': 'mp4', - 'title': 'Attack of the Killer Tomatoes', - 'description': 'A group of scientists band together to save the world from mutated tomatoes that KILL! (1978)', - 'duration': 5700, - }, + 'playlist_mincount': 100, + }, { + 'url': 'https://pluto.tv/on-demand/series/66ab6d80b20e79001338fe4c/season/5', + 'info_dict': { + 'id': '66ab6d80b20e79001338fe4c-5', + 'title': 'Squadra Speciale Cobra 11 - Season 5', }, - ] + 'playlist_count': 17, + }, { + 'url': 'https://pluto.tv/on-demand/series/62f5fb1b51b268001a993666/season/2/episode/62fb6e32946347001bf008ca', + 'md5': '7e100cd3b771e9dd9b6af81abdd812af', + 'info_dict': { + 'id': '62fb6e32946347001bf008ca', + 'ext': 'mp4', + 'display_id': 'serie-2-episodio-5-2006-2-5', + 'episode_id': '62fb6e32946347001bf008ca', + 'genres': ['Sci-Fi & Fantasy'], + 'thumbnail': 'http://images.pluto.tv/episodes/62fb6e32946347001bf008ca/poster.jpg?fm=png&q=100', + 'season_number': 2, + 'series_id': '62f5fb1b51b268001a993666', + 'title': 'Serie 2, Episodio 5', + 'duration': 3120.0, + 'season': 'Season 2', + 'series': 'Doctor Who', + 'description': 'md5:66965714ba60b02996bd1b551475f6fa', + 'episode': 'Serie 2, Episodio 5', + }, + }] def _to_ad_free_formats(self, video_id, formats, subtitles): ad_free_formats, ad_free_subtitles, m3u8_urls = [], {}, set() @@ -130,63 +126,60 @@ class PlutoTVIE(InfoExtractor): self.report_warning('Unable to find ad-free formats') return formats, subtitles - def _get_video_info(self, video_json, slug, series_name=None): - video_id = video_json.get('_id', slug) - formats, subtitles = [], {} - for video_url in try_get(video_json, lambda x: x['stitched']['urls'], list) or []: - if video_url.get('type') != 'hls': - continue - url = url_or_none(video_url.get('url')) - - fmts, subs = self._extract_m3u8_formats_and_subtitles( - url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False) - formats.extend(fmts) - subtitles = self._merge_subtitles(subtitles, subs) - - formats, subtitles = self._to_ad_free_formats(video_id, formats, subtitles) - - info = { - 'id': video_id, + def _get_video_info(self, video_json, video, series=None, season_number=None): + formats, subtitles = self._extract_formats(video_json, video) + thumbnails = [{ + 'url': cover['url'], + 'width': int(m.group(1)) if (m := re.search(r'w=(\d+)&h=(\d+)', cover['url'])) else None, + 'height': int(m.group(2)) if m else None, + } for cover in video.get('covers', [])] + first_cover = traverse_obj(video, ('covers', 0, 'url')) + if first_cover: + thumbnails.append({ + 'id': 'original', + 'url': re.sub(r'\?.*$', '?fm=png&q=100', first_cover), + 'preference': 1, + }) + return { + 'id': video.get('id') or video.get('_id'), + 'title': video.get('name'), + 'display_id': video.get('slug'), + 'thumbnails': thumbnails, 'formats': formats, 'subtitles': subtitles, - 'title': video_json.get('name'), - 'description': video_json.get('description'), - 'duration': float_or_none(video_json.get('duration'), scale=1000), + 'description': video.get('description'), + 'duration': float_or_none(video.get('duration'), scale=1000), + 'genres': [video.get('genre')], + 'series_id': series and series.get('id'), + 'series': series and series.get('name'), + 'episode': video.get('name'), + 'episode_id': video.get('id') or video.get('_id'), + 'season_number': int_or_none(season_number), } - if series_name: - info.update({ - 'series': series_name, - 'episode': video_json.get('name'), - 'season_number': int_or_none(video_json.get('season')), - 'episode_number': int_or_none(video_json.get('number')), - }) - return info - def _real_extract(self, url): - mobj = self._match_valid_url(url).groupdict() - info_slug = mobj['series_or_movie_slug'] - video_json = self._download_json(self._INFO_URL + info_slug, info_slug, query=self._INFO_QUERY_PARAMS) + def _season_playlist(self, video_json, series, season_number, season): + for ep in season['episodes']: + yield self._get_video_info(video_json, ep, series, season_number) - if mobj['video_type'] == 'series': - series_name = video_json.get('name', info_slug) - season_number, episode_slug = mobj.get('season_number'), mobj.get('episode_slug') + def _series_playlist(self, video_json, series): + for season in series['seasons']: + for ep in season['episodes']: + yield self._get_video_info(video_json, ep, series, season['number']) - videos = [] - for season in video_json['seasons']: - if season_number is not None and season_number != int_or_none(season.get('number')): - continue - for episode in season['episodes']: - if episode_slug is not None and episode_slug != episode.get('slug'): - continue - videos.append(self._get_video_info(episode, episode_slug, series_name)) - if not videos: - raise ExtractorError('Failed to find any videos to extract') - if episode_slug is not None and len(videos) == 1: - return videos[0] - playlist_title = series_name - if season_number is not None: - playlist_title += ' - Season %d' % season_number - return self.playlist_result(videos, - playlist_id=video_json.get('_id', info_slug), - playlist_title=playlist_title) - return self._get_video_info(video_json, info_slug) + def _real_extract(self, url): + mobj = self._match_valid_url(url).groupdict() + slug = mobj['slug'] + season_number, episode_id = mobj.get('season'), mobj.get('episode') + query = {**self._START_QUERY, 'seriesIDs': slug} + if episode_id: + query['episodeIDs'] = episode_id + video_json = self._download_json('https://boot.pluto.tv/v4/start', slug, 'Downloading info json', query=query) + series = video_json['VOD'][0] + if episode_id: + return self._get_video_info(video_json, video_json['VOD'][1], series, season_number) + if season_number: + for season in series['seasons']: + if season['number'] == int(season_number): + return self.playlist_result(self._season_playlist(video_json, series, season_number, season), f"{series['id']}-{season_number}", f"{series['name']} - Season {season_number}") + raise ExtractorError(f'Failed to find season {season_number}') + return self.playlist_result(self._series_playlist(video_json, series), series['id'], series['name'], series.get('description')) if 'seasons' in series else self._get_video_info(video_json, series) From ea8ed5d52de51462363067a7dc5c2784feb49169 Mon Sep 17 00:00:00 2001 From: DTrombett Date: Mon, 30 Dec 2024 13:52:00 +0100 Subject: [PATCH 2/3] [ie/plutotv] Improve playlist extraction --- yt_dlp/extractor/plutotv.py | 85 +++++++++++++++++++++++++------------ 1 file changed, 57 insertions(+), 28 deletions(-) diff --git a/yt_dlp/extractor/plutotv.py b/yt_dlp/extractor/plutotv.py index 98154d0337..9b39577e90 100644 --- a/yt_dlp/extractor/plutotv.py +++ b/yt_dlp/extractor/plutotv.py @@ -3,7 +3,7 @@ import urllib.parse import uuid from .common import InfoExtractor -from ..utils import ExtractorError, float_or_none, int_or_none, traverse_obj +from ..utils import ExtractorError, float_or_none, int_or_none, smuggle_url, traverse_obj, unsmuggle_url class PlutoTVBase(InfoExtractor): @@ -19,14 +19,26 @@ class PlutoTVBase(InfoExtractor): 'deviceVersion': 'unknown', } - def _extract_formats(self, start, element): - formats, subtitles = self._extract_m3u8_formats_and_subtitles(f"{start['servers']['stitcher']}/v2{element['stitched']['path']}?{start['stitcherParams']}&jwt={start['sessionToken']}", element.get('id') or element.get('_id')) + def _resolve_data(self, start, element): + return { + 'stitcher': start['servers']['stitcher'], + 'path': element['stitched']['path'], + 'stitcherParams': start['stitcherParams'], + 'sessionToken': start['sessionToken'], + 'id': element.get('id') or element['_id'], + } + + def _extract_formats(self, video_data): + formats, subtitles = self._extract_m3u8_formats_and_subtitles(f"{video_data['stitcher']}/v2{video_data['path']}?{video_data['stitcherParams']}&jwt={video_data['sessionToken']}", video_data['id']) for f in formats: - f['url'] += f"&jwt={start['sessionToken']}" + f['url'] += f"&jwt={video_data['sessionToken']}" f.setdefault('vcodec', 'avc1.64001f') f.setdefault('acodec', 'mp4a.40.2') f.setdefault('fps', 30) - return formats, subtitles + return { + 'formats': formats, + 'subtitles': subtitles, + } class PlutoTVIE(PlutoTVBase): @@ -37,8 +49,7 @@ class PlutoTVIE(PlutoTVBase): (?: (?:/seasons?/(?P\d+))? (?:/episode/(?P[^/]+))? - )? - /?(?:$|[#?])''' + )?''' _TESTS = [{ 'url': 'https://pluto.tv/it/on-demand/movies/6246b0adef11000014d220c3', 'md5': '5efe37ad6c1085a4ad4684b9b82cb1c1', @@ -62,8 +73,9 @@ class PlutoTVIE(PlutoTVBase): 'info_dict': { 'id': '6655b0c5cceea000134aee27', 'title': 'Mission Impossible', + 'description': 'md5:21604bf9971528825c359e0f4977d572', }, - 'playlist_mincount': 100, + 'playlist_mincount': 113, }, { 'url': 'https://pluto.tv/on-demand/series/66ab6d80b20e79001338fe4c/season/5', 'info_dict': { @@ -126,8 +138,7 @@ class PlutoTVIE(PlutoTVBase): self.report_warning('Unable to find ad-free formats') return formats, subtitles - def _get_video_info(self, video_json, video, series=None, season_number=None): - formats, subtitles = self._extract_formats(video_json, video) + def _get_video_info(self, video, series=None, season_number=None): thumbnails = [{ 'url': cover['url'], 'width': int(m.group(1)) if (m := re.search(r'w=(\d+)&h=(\d+)', cover['url'])) else None, @@ -141,45 +152,63 @@ class PlutoTVIE(PlutoTVBase): 'preference': 1, }) return { - 'id': video.get('id') or video.get('_id'), + 'id': video.get('id') or video['_id'], 'title': video.get('name'), 'display_id': video.get('slug'), 'thumbnails': thumbnails, - 'formats': formats, - 'subtitles': subtitles, 'description': video.get('description'), 'duration': float_or_none(video.get('duration'), scale=1000), 'genres': [video.get('genre')], 'series_id': series and series.get('id'), 'series': series and series.get('name'), 'episode': video.get('name'), - 'episode_id': video.get('id') or video.get('_id'), + 'episode_id': video.get('id') or video['_id'], 'season_number': int_or_none(season_number), } - def _season_playlist(self, video_json, series, season_number, season): - for ep in season['episodes']: - yield self._get_video_info(video_json, ep, series, season_number) - - def _series_playlist(self, video_json, series): - for season in series['seasons']: - for ep in season['episodes']: - yield self._get_video_info(video_json, ep, series, season['number']) + def _playlist_entry(self, video_json, series, season, ep): + episode_id = ep.get('id') or ep['_id'] + return self.url_result( + smuggle_url( + f"https://pluto.tv/on-demand/series/{series.get('id') or series['_id']}/season/{season['number']}/episode/{episode_id}", + self._resolve_data(video_json, ep), + ), + PlutoTVIE, + episode_id, + ep.get('name'), + url_transparent=True, + **self._get_video_info(ep, series, season.get('number')), + ) def _real_extract(self, url): + url, video_data = unsmuggle_url(url) + if video_data: + return {**self._extract_formats(video_data), 'id': video_data['id']} + mobj = self._match_valid_url(url).groupdict() slug = mobj['slug'] season_number, episode_id = mobj.get('season'), mobj.get('episode') query = {**self._START_QUERY, 'seriesIDs': slug} if episode_id: query['episodeIDs'] = episode_id + video_json = self._download_json('https://boot.pluto.tv/v4/start', slug, 'Downloading info json', query=query) series = video_json['VOD'][0] + if episode_id: - return self._get_video_info(video_json, video_json['VOD'][1], series, season_number) + episode = video_json['VOD'][1] + return {**self._get_video_info(episode, series, season_number), **self._extract_formats(self._resolve_data(video_json, episode))} + if season_number: - for season in series['seasons']: - if season['number'] == int(season_number): - return self.playlist_result(self._season_playlist(video_json, series, season_number, season), f"{series['id']}-{season_number}", f"{series['name']} - Season {season_number}") - raise ExtractorError(f'Failed to find season {season_number}') - return self.playlist_result(self._series_playlist(video_json, series), series['id'], series['name'], series.get('description')) if 'seasons' in series else self._get_video_info(video_json, series) + season = next((s for s in series['seasons'] if s['number'] == int(season_number)), None) + if not season: + raise ExtractorError(f'Failed to find season {season_number}') + return self.playlist_result( + [self._playlist_entry(video_json, series, season, ep) for ep in season['episodes']], + f"{series['id']}-{season_number}", f"{series['name']} - Season {season_number}", + ) + + return self.playlist_result( + [self._playlist_entry(video_json, series, season, ep) for season in series.get('seasons', []) for ep in season['episodes']], + series['id'], series['name'], series.get('description'), + ) if 'seasons' in series else {**self._get_video_info(series), **self._extract_formats(self._resolve_data(video_json, series))} From 562083e557cb36200493f8285acd429dcc6ff783 Mon Sep 17 00:00:00 2001 From: DTrombett Date: Wed, 12 Mar 2025 16:10:23 +0100 Subject: [PATCH 3/3] fix: new titles --- yt_dlp/extractor/plutotv.py | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/yt_dlp/extractor/plutotv.py b/yt_dlp/extractor/plutotv.py index 9b39577e90..4afc641e3d 100644 --- a/yt_dlp/extractor/plutotv.py +++ b/yt_dlp/extractor/plutotv.py @@ -95,12 +95,12 @@ class PlutoTVIE(PlutoTVBase): 'thumbnail': 'http://images.pluto.tv/episodes/62fb6e32946347001bf008ca/poster.jpg?fm=png&q=100', 'season_number': 2, 'series_id': '62f5fb1b51b268001a993666', - 'title': 'Serie 2, Episodio 5', + 'title': 'L\'ascesa dei Cyber-uomini (parte 1)', 'duration': 3120.0, 'season': 'Season 2', 'series': 'Doctor Who', 'description': 'md5:66965714ba60b02996bd1b551475f6fa', - 'episode': 'Serie 2, Episodio 5', + 'episode': 'L\'ascesa dei Cyber-uomini (parte 1)', }, }] @@ -176,7 +176,6 @@ class PlutoTVIE(PlutoTVBase): PlutoTVIE, episode_id, ep.get('name'), - url_transparent=True, **self._get_video_info(ep, series, season.get('number')), ) @@ -196,7 +195,9 @@ class PlutoTVIE(PlutoTVBase): series = video_json['VOD'][0] if episode_id: - episode = video_json['VOD'][1] + episode = traverse_obj(video_json, ('VOD', 1)) + if not episode: + raise ExtractorError('Failed to find episode') return {**self._get_video_info(episode, series, season_number), **self._extract_formats(self._resolve_data(video_json, episode))} if season_number: