From 7be14109a6bd493a2e881da4f9e30adaf3e7e5d5 Mon Sep 17 00:00:00 2001 From: InvalidUsernameException Date: Thu, 1 May 2025 00:27:42 +0200 Subject: [PATCH] [ie/zdf] Fix extractors (#12779) Closes #12647 Authored by: InvalidUsernameException, bashonly Co-authored-by: bashonly <88596187+bashonly@users.noreply.github.com> --- yt_dlp/extractor/dreisat.py | 114 +++-- yt_dlp/extractor/phoenix.py | 61 +-- yt_dlp/extractor/zdf.py | 979 ++++++++++++++++++++++-------------- 3 files changed, 713 insertions(+), 441 deletions(-) diff --git a/yt_dlp/extractor/dreisat.py b/yt_dlp/extractor/dreisat.py index 376ff672df..edd66e46cc 100644 --- a/yt_dlp/extractor/dreisat.py +++ b/yt_dlp/extractor/dreisat.py @@ -1,9 +1,15 @@ from .zdf import ZDFBaseIE +from ..utils import ( + int_or_none, + merge_dicts, + parse_iso8601, +) +from ..utils.traversal import require, traverse_obj class DreiSatIE(ZDFBaseIE): IE_NAME = '3sat' - _VALID_URL = r'https?://(?:www\.)?3sat\.de/(?:[^/]+/)*(?P[^/?#&]+)\.html' + _VALID_URL = r'https?://(?:www\.)?3sat\.de/(?:[^/?#]+/)*(?P[^/?#&]+)\.html' _TESTS = [{ 'url': 'https://www.3sat.de/dokumentation/reise/traumziele-suedostasiens-die-philippinen-und-vietnam-102.html', 'info_dict': { @@ -12,40 +18,59 @@ class DreiSatIE(ZDFBaseIE): 'title': 'Traumziele Südostasiens (1/2): Die Philippinen und Vietnam', 'description': 'md5:26329ce5197775b596773b939354079d', 'duration': 2625.0, - 'thumbnail': 'https://www.3sat.de/assets/traumziele-suedostasiens-die-philippinen-und-vietnam-100~2400x1350?cb=1699870351148', + 'thumbnail': 'https://www.3sat.de/assets/traumziele-suedostasiens-die-philippinen-und-vietnam-100~original?cb=1699870351148', 'episode': 'Traumziele Südostasiens (1/2): Die Philippinen und Vietnam', 'episode_id': 'POS_cc7ff51c-98cf-4d12-b99d-f7a551de1c95', - 'timestamp': 1738593000, - 'upload_date': '20250203', + 'timestamp': 1747920900, + 'upload_date': '20250522', }, }, { - # Same as https://www.zdf.de/dokumentation/ab-18/10-wochen-sommer-102.html - 'url': 'https://www.3sat.de/film/ab-18/10-wochen-sommer-108.html', - 'md5': '0aff3e7bc72c8813f5e0fae333316a1d', + 'url': 'https://www.3sat.de/film/ab-18/ab-18---mein-fremdes-ich-100.html', + 'md5': 'f92638413a11d759bdae95c9d8ec165c', 'info_dict': { - 'id': '141007_ab18_10wochensommer_film', + 'id': '221128_mein_fremdes_ich2_ab18', 'ext': 'mp4', - 'title': 'Ab 18! - 10 Wochen Sommer', - 'description': 'md5:8253f41dc99ce2c3ff892dac2d65fe26', - 'duration': 2660, - 'timestamp': 1608604200, - 'upload_date': '20201222', + 'title': 'Ab 18! - Mein fremdes Ich', + 'description': 'md5:cae0c0b27b7426d62ca0dda181738bf0', + 'duration': 2625.0, + 'thumbnail': 'https://www.3sat.de/assets/ab-18---mein-fremdes-ich-106~original?cb=1666081865812', + 'episode': 'Ab 18! - Mein fremdes Ich', + 'episode_id': 'POS_6225d1ca-a0d5-45e3-870b-e783ee6c8a3f', + 'timestamp': 1695081600, + 'upload_date': '20230919', }, - 'skip': '410 Gone', }, { - 'url': 'https://www.3sat.de/gesellschaft/schweizweit/waidmannsheil-100.html', + 'url': 'https://www.3sat.de/gesellschaft/37-grad-leben/aus-dem-leben-gerissen-102.html', + 'md5': 'a903eaf8d1fd635bd3317cd2ad87ec84', 'info_dict': { - 'id': '140913_sendung_schweizweit', + 'id': '250323_0903_sendung_sgl', 'ext': 'mp4', - 'title': 'Waidmannsheil', - 'description': 'md5:cce00ca1d70e21425e72c86a98a56817', - 'timestamp': 1410623100, - 'upload_date': '20140913', + 'title': 'Plötzlich ohne dich', + 'description': 'md5:380cc10659289dd91510ad8fa717c66b', + 'duration': 1620.0, + 'thumbnail': 'https://www.3sat.de/assets/37-grad-leben-106~original?cb=1645537156810', + 'episode': 'Plötzlich ohne dich', + 'episode_id': 'POS_faa7a93c-c0f2-4d51-823f-ce2ac3ee191b', + 'timestamp': 1743162540, + 'upload_date': '20250328', }, - 'params': { - 'skip_download': True, + }, { + # Video with chapters + 'url': 'https://www.3sat.de/kultur/buchmesse/dein-buch-das-beste-von-der-leipziger-buchmesse-2025-teil-1-100.html', + 'md5': '6b95790ce52e75f0d050adcdd2711ee6', + 'info_dict': { + 'id': '250330_dein_buch1_bum', + 'ext': 'mp4', + 'title': 'dein buch - Das Beste von der Leipziger Buchmesse 2025 - Teil 1', + 'description': 'md5:bae51bfc22f15563ce3acbf97d2e8844', + 'duration': 5399.0, + 'thumbnail': 'https://www.3sat.de/assets/buchmesse-kerkeling-100~original?cb=1743329640903', + 'chapters': 'count:24', + 'episode': 'dein buch - Das Beste von der Leipziger Buchmesse 2025 - Teil 1', + 'episode_id': 'POS_1ef236cc-b390-401e-acd0-4fb4b04315fb', + 'timestamp': 1743327000, + 'upload_date': '20250330', }, - 'skip': '404 Not Found', }, { # Same as https://www.zdf.de/filme/filme-sonstige/der-hauptmann-112.html 'url': 'https://www.3sat.de/film/spielfilm/der-hauptmann-100.html', @@ -58,11 +83,42 @@ class DreiSatIE(ZDFBaseIE): def _real_extract(self, url): video_id = self._match_id(url) + webpage = self._download_webpage(url, video_id) + player = self._search_json( + r'data-zdfplayer-jsb=(["\'])', webpage, 'player JSON', video_id) + player_url = player['content'] + api_token = f'Bearer {player["apiToken"]}' - webpage = self._download_webpage(url, video_id, fatal=False) - if webpage: - player = self._extract_player(webpage, url, fatal=False) - if player: - return self._extract_regular(url, player, video_id) + content = self._call_api(player_url, video_id, 'video metadata', api_token) - return self._extract_mobile(video_id) + video_target = content['mainVideoContent']['http://zdf.de/rels/target'] + ptmd_path = traverse_obj(video_target, ( + (('streams', 'default'), None), + ('http://zdf.de/rels/streams/ptmd', 'http://zdf.de/rels/streams/ptmd-template'), + {str}, any, {require('ptmd path')})) + ptmd_url = self._expand_ptmd_template(player_url, ptmd_path) + aspect_ratio = self._parse_aspect_ratio(video_target.get('aspectRatio')) + info = self._extract_ptmd(ptmd_url, video_id, api_token, aspect_ratio) + + return merge_dicts(info, { + **traverse_obj(content, { + 'title': (('title', 'teaserHeadline'), {str}, any), + 'episode': (('title', 'teaserHeadline'), {str}, any), + 'description': (('leadParagraph', 'teasertext'), {str}, any), + 'timestamp': ('editorialDate', {parse_iso8601}), + }), + **traverse_obj(video_target, { + 'duration': ('duration', {int_or_none}), + 'chapters': ('streamAnchorTag', {self._extract_chapters}), + }), + 'thumbnails': self._extract_thumbnails(traverse_obj(content, ('teaserImageRef', 'layouts', {dict}))), + **traverse_obj(content, ('programmeItem', 0, 'http://zdf.de/rels/target', { + 'series_id': ('http://zdf.de/rels/cmdm/series', 'seriesUuid', {str}), + 'series': ('http://zdf.de/rels/cmdm/series', 'seriesTitle', {str}), + 'season': ('http://zdf.de/rels/cmdm/season', 'seasonTitle', {str}), + 'season_number': ('http://zdf.de/rels/cmdm/season', 'seasonNumber', {int_or_none}), + 'season_id': ('http://zdf.de/rels/cmdm/season', 'seasonUuid', {str}), + 'episode_number': ('episodeNumber', {int_or_none}), + 'episode_id': ('contentId', {str}), + })), + }) diff --git a/yt_dlp/extractor/phoenix.py b/yt_dlp/extractor/phoenix.py index 63c256019e..9df34f8c91 100644 --- a/yt_dlp/extractor/phoenix.py +++ b/yt_dlp/extractor/phoenix.py @@ -1,5 +1,3 @@ -import re - from .youtube import YoutubeIE from .zdf import ZDFBaseIE from ..utils import ( @@ -7,44 +5,27 @@ from ..utils import ( merge_dicts, try_get, unified_timestamp, - urljoin, ) class PhoenixIE(ZDFBaseIE): IE_NAME = 'phoenix.de' - _VALID_URL = r'https?://(?:www\.)?phoenix\.de/(?:[^/]+/)*[^/?#&]*-a-(?P\d+)\.html' + _VALID_URL = r'https?://(?:www\.)?phoenix\.de/(?:[^/?#]+/)*[^/?#&]*-a-(?P\d+)\.html' _TESTS = [{ - # Same as https://www.zdf.de/politik/phoenix-sendungen/wohin-fuehrt-der-protest-in-der-pandemie-100.html - 'url': 'https://www.phoenix.de/sendungen/ereignisse/corona-nachgehakt/wohin-fuehrt-der-protest-in-der-pandemie-a-2050630.html', - 'md5': '34ec321e7eb34231fd88616c65c92db0', + 'url': 'https://www.phoenix.de/sendungen/dokumentationen/spitzbergen-a-893349.html', + 'md5': 'a79e86d9774d0b3f2102aff988a0bd32', 'info_dict': { - 'id': '210222_phx_nachgehakt_corona_protest', + 'id': '221215_phx_spitzbergen', 'ext': 'mp4', - 'title': 'Wohin führt der Protest in der Pandemie?', - 'description': 'md5:7d643fe7f565e53a24aac036b2122fbd', - 'duration': 1691, - 'timestamp': 1613902500, - 'upload_date': '20210221', + 'title': 'Spitzbergen', + 'description': 'Film von Tilmann Bünz', + 'duration': 728.0, + 'timestamp': 1555600500, + 'upload_date': '20190418', 'uploader': 'Phoenix', - 'series': 'corona nachgehakt', - 'episode': 'Wohin führt der Protest in der Pandemie?', - }, - }, { - # Youtube embed - 'url': 'https://www.phoenix.de/sendungen/gespraeche/phoenix-streitgut-brennglas-corona-a-1965505.html', - 'info_dict': { - 'id': 'hMQtqFYjomk', - 'ext': 'mp4', - 'title': 'phoenix streitgut: Brennglas Corona - Wie gerecht ist unsere Gesellschaft?', - 'description': 'md5:ac7a02e2eb3cb17600bc372e4ab28fdd', - 'duration': 3509, - 'upload_date': '20201219', - 'uploader': 'phoenix', - 'uploader_id': 'phoenix', - }, - 'params': { - 'skip_download': True, + 'thumbnail': 'https://www.phoenix.de/sixcms/media.php/21/Bergspitzen1.png', + 'series': 'Dokumentationen', + 'episode': 'Spitzbergen', }, }, { 'url': 'https://www.phoenix.de/entwicklungen-in-russland-a-2044720.html', @@ -90,8 +71,8 @@ class PhoenixIE(ZDFBaseIE): content_id = details['tracking']['nielsen']['content']['assetid'] info = self._extract_ptmd( - f'https://tmd.phoenix.de/tmd/2/ngplayer_2_3/vod/ptmd/phoenix/{content_id}', - content_id, None, url) + f'https://tmd.phoenix.de/tmd/2/android_native_6/vod/ptmd/phoenix/{content_id}', + content_id) duration = int_or_none(try_get( details, lambda x: x['tracking']['nielsen']['content']['length'])) @@ -101,20 +82,8 @@ class PhoenixIE(ZDFBaseIE): str) episode = title if details.get('contentType') == 'episode' else None - thumbnails = [] teaser_images = try_get(details, lambda x: x['teaserImageRef']['layouts'], dict) or {} - for thumbnail_key, thumbnail_url in teaser_images.items(): - thumbnail_url = urljoin(url, thumbnail_url) - if not thumbnail_url: - continue - thumbnail = { - 'url': thumbnail_url, - } - m = re.match('^([0-9]+)x([0-9]+)$', thumbnail_key) - if m: - thumbnail['width'] = int(m.group(1)) - thumbnail['height'] = int(m.group(2)) - thumbnails.append(thumbnail) + thumbnails = self._extract_thumbnails(teaser_images) return merge_dicts(info, { 'id': content_id, diff --git a/yt_dlp/extractor/zdf.py b/yt_dlp/extractor/zdf.py index 4bef1017cc..10be582a33 100644 --- a/yt_dlp/extractor/zdf.py +++ b/yt_dlp/extractor/zdf.py @@ -1,331 +1,410 @@ +import itertools +import json import re +import time from .common import InfoExtractor from ..utils import ( - NO_DEFAULT, ExtractorError, determine_ext, + filter_dict, float_or_none, int_or_none, join_nonempty, - merge_dicts, + make_archive_id, parse_codecs, - qualities, - traverse_obj, - try_get, + parse_iso8601, + parse_qs, + smuggle_url, unified_timestamp, - update_url_query, + unsmuggle_url, url_or_none, urljoin, + variadic, ) +from ..utils.traversal import require, traverse_obj class ZDFBaseIE(InfoExtractor): _GEO_COUNTRIES = ['DE'] - _QUALITIES = ('auto', 'low', 'med', 'high', 'veryhigh', 'hd', 'fhd', 'uhd') + _TOKEN_CACHE_PARAMS = ('zdf', 'api-token') + _token_cache = {} - def _download_v2_doc(self, document_id): - return self._download_json( - f'https://zdf-prod-futura.zdf.de/mediathekV2/document/{document_id}', - document_id) + def _get_api_token(self): + # As of 2025-03, this API is used by the Android app for getting tokens. + # An equivalent token could be extracted from the webpage should the API become unavailable. + # For now this allows the extractor to avoid dealing with Next.js hydration data. + if not self._token_cache: + self._token_cache.update(self.cache.load(*self._TOKEN_CACHE_PARAMS, default={})) - def _call_api(self, url, video_id, item, api_token=None, referrer=None): - headers = {} - if api_token: - headers['Api-Auth'] = f'Bearer {api_token}' - if referrer: - headers['Referer'] = referrer + if traverse_obj(self._token_cache, ('expires', {int_or_none}), default=0) < int(time.time()): + self._token_cache.update(self._download_json( + 'https://zdf-prod-futura.zdf.de/mediathekV2/token', None, + 'Downloading API token', 'Failed to download API token')) + self.cache.store(*self._TOKEN_CACHE_PARAMS, self._token_cache) + + return f'{self._token_cache["type"]} {self._token_cache["token"]}' + + def _call_api(self, url, video_id, item, api_token=None): return self._download_json( - url, video_id, f'Downloading JSON {item}', headers=headers) + url, video_id, f'Downloading {item}', f'Failed to download {item}', + headers=filter_dict({'Api-Auth': api_token})) + + def _parse_aspect_ratio(self, aspect_ratio): + if not aspect_ratio or not isinstance(aspect_ratio, str): + return None + mobj = re.match(r'(?P\d+):(?P\d+)', aspect_ratio) + return int(mobj.group('width')) / int(mobj.group('height')) if mobj else None + + def _extract_chapters(self, data): + return traverse_obj(data, (lambda _, v: v['anchorOffset'], { + 'start_time': ('anchorOffset', {float_or_none}), + 'title': ('anchorLabel', {str}), + })) or None @staticmethod def _extract_subtitles(src): + seen_urls = set() subtitles = {} - for caption in try_get(src, lambda x: x['captions'], list) or []: + for caption in src: subtitle_url = url_or_none(caption.get('uri')) - if subtitle_url: - lang = caption.get('language', 'deu') - subtitles.setdefault(lang, []).append({ - 'url': subtitle_url, - }) + if not subtitle_url or subtitle_url in seen_urls: + continue + seen_urls.add(subtitle_url) + lang = caption.get('language') or 'deu' + subtitles.setdefault(lang, []).append({ + 'url': subtitle_url, + }) return subtitles - def _extract_format(self, video_id, formats, format_urls, meta): - format_url = url_or_none(meta.get('url')) - if not format_url or format_url in format_urls: - return - format_urls.add(format_url) + def _expand_ptmd_template(self, api_base_url, template): + return urljoin(api_base_url, template.replace('{playerId}', 'android_native_6')) - mime_type, ext = meta.get('mimeType'), determine_ext(format_url) - if mime_type == 'application/x-mpegURL' or ext == 'm3u8': - new_formats = self._extract_m3u8_formats( - format_url, video_id, 'mp4', m3u8_id='hls', - entry_protocol='m3u8_native', fatal=False) - elif mime_type == 'application/f4m+xml' or ext == 'f4m': - new_formats = self._extract_f4m_formats( - update_url_query(format_url, {'hdcore': '3.7.0'}), video_id, f4m_id='hds', fatal=False) - elif ext == 'mpd': - new_formats = self._extract_mpd_formats( - format_url, video_id, mpd_id='dash', fatal=False) - else: - f = parse_codecs(meta.get('mimeCodec')) - if not f and meta.get('type'): - data = meta['type'].split('_') - if try_get(data, lambda x: x[2]) == ext: - f = {'vcodec': data[0], 'acodec': data[1]} - f.update({ - 'url': format_url, - 'format_id': join_nonempty('http', meta.get('type'), meta.get('quality')), - 'tbr': int_or_none(self._search_regex(r'_(\d+)k_', format_url, 'tbr', default=None)), - }) - new_formats = [f] - formats.extend(merge_dicts(f, { - 'format_note': join_nonempty('quality', 'class', from_dict=meta, delim=', '), - 'language': meta.get('language'), - 'language_preference': 10 if meta.get('class') == 'main' else -10 if meta.get('class') == 'ad' else -1, - 'quality': qualities(self._QUALITIES)(meta.get('quality')), - }) for f in new_formats) + def _extract_ptmd(self, ptmd_urls, video_id, api_token=None, aspect_ratio=None): + content_id = None + duration = None + formats, src_captions = [], [] + seen_urls = set() - def _extract_ptmd(self, ptmd_url, video_id, api_token, referrer): - ptmd = self._call_api( - ptmd_url, video_id, 'metadata', api_token, referrer) + for ptmd_url in variadic(ptmd_urls): + ptmd_url, smuggled_data = unsmuggle_url(ptmd_url, {}) + # Is it a DGS variant? (*D*eutsche *G*ebärden*s*prache' / German Sign Language) + is_dgs = smuggled_data.get('vod_media_type') == 'DGS' + ptmd = self._call_api(ptmd_url, video_id, 'PTMD data', api_token) - content_id = ptmd.get('basename') or ptmd_url.split('/')[-1] + basename = ( + ptmd.get('basename') + # ptmd_url examples: + # https://api.zdf.de/tmd/2/android_native_6/vod/ptmd/mediathek/250328_sendung_hsh/3 + # https://tmd.phoenix.de/tmd/2/android_native_6/vod/ptmd/phoenix/221215_phx_spitzbergen + or self._search_regex(r'/vod/ptmd/[^/?#]+/(\w+)', ptmd_url, 'content ID', default=None)) + # If this is_dgs, then it's from ZDFIE and it only uses content_id for _old_archive_ids, + # and the old version of the extractor didn't extract DGS variants, so ignore basename + if not content_id and not is_dgs: + content_id = basename - formats = [] - track_uris = set() - for p in ptmd['priorityList']: - formitaeten = p.get('formitaeten') - if not isinstance(formitaeten, list): - continue - for f in formitaeten: - f_qualities = f.get('qualities') - if not isinstance(f_qualities, list): - continue - for quality in f_qualities: - tracks = try_get(quality, lambda x: x['audio']['tracks'], list) - if not tracks: - continue - for track in tracks: - self._extract_format( - content_id, formats, track_uris, { - 'url': track.get('uri'), - 'type': f.get('type'), - 'mimeType': f.get('mimeType'), - 'quality': quality.get('quality'), - 'class': track.get('class'), - 'language': track.get('language'), + if not duration: + duration = traverse_obj(ptmd, ('attributes', 'duration', 'value', {float_or_none(scale=1000)})) + src_captions += traverse_obj(ptmd, ('captions', ..., {dict})) + + for stream in traverse_obj(ptmd, ('priorityList', ..., 'formitaeten', ..., {dict})): + for quality in traverse_obj(stream, ('qualities', ..., {dict})): + for variant in traverse_obj(quality, ('audio', 'tracks', lambda _, v: url_or_none(v['uri']))): + format_url = variant['uri'] + if format_url in seen_urls: + continue + seen_urls.add(format_url) + ext = determine_ext(format_url) + if ext == 'm3u8': + fmts = self._extract_m3u8_formats( + format_url, video_id, 'mp4', m3u8_id='hls', fatal=False) + elif ext == 'mpd': + fmts = self._extract_mpd_formats( + format_url, video_id, mpd_id='dash', fatal=False) + else: + height = int_or_none(quality.get('highestVerticalResolution')) + width = round(aspect_ratio * height) if aspect_ratio and height else None + fmts = [{ + 'url': format_url, + **parse_codecs(quality.get('mimeCodec')), + 'height': height, + 'width': width, + 'format_id': join_nonempty('http', stream.get('type')), + 'tbr': int_or_none(self._search_regex(r'_(\d+)k_', format_url, 'tbr', default=None)), + }] + f_class = variant.get('class') + for f in fmts: + formats.append({ + **f, + 'format_id': join_nonempty(f.get('format_id'), is_dgs and 'dgs'), + 'format_note': join_nonempty( + f_class, is_dgs and 'German Sign Language', f.get('format_note'), delim=', '), + 'language': variant.get('language') or f.get('language'), + 'preference': -2 if is_dgs else -1, + 'language_preference': 10 if f_class == 'main' else -10 if f_class == 'ad' else -1, }) - duration = float_or_none(try_get( - ptmd, lambda x: x['attributes']['duration']['value']), scale=1000) - return { - 'extractor_key': ZDFIE.ie_key(), - 'id': content_id, + 'id': content_id or video_id, 'duration': duration, 'formats': formats, - 'subtitles': self._extract_subtitles(ptmd), - '_format_sort_fields': ('tbr', 'res', 'quality', 'language_preference'), + 'subtitles': self._extract_subtitles(src_captions), } - def _extract_player(self, webpage, video_id, fatal=True): - return self._parse_json( - self._search_regex( - r'(?s)data-zdfplayer-jsb=(["\'])(?P{.+?})\1', webpage, - 'player JSON', default='{}' if not fatal else NO_DEFAULT, - group='json'), - video_id) + def _download_graphql(self, item_id, data_desc, query=None, body=None): + assert query or body, 'One of query or body is required' - def _extract_entry(self, url, player, content, video_id): - title = content.get('title') or content['teaserHeadline'] + return self._download_json( + 'https://api.zdf.de/graphql', item_id, + f'Downloading {data_desc}', f'Failed to download {data_desc}', + query=query, data=json.dumps(body).encode() if body else None, + headers=filter_dict({ + 'Api-Auth': self._get_api_token(), + 'Apollo-Require-Preflight': True, + 'Content-Type': 'application/json' if body else None, + })) - t = content['mainVideoContent']['http://zdf.de/rels/target'] - ptmd_path = traverse_obj(t, ( - (('streams', 'default'), None), - ('http://zdf.de/rels/streams/ptmd', 'http://zdf.de/rels/streams/ptmd-template'), - ), get_all=False) - if not ptmd_path: - raise ExtractorError('Could not extract ptmd_path') - - info = self._extract_ptmd( - urljoin(url, ptmd_path.replace('{playerId}', 'android_native_5')), video_id, player['apiToken'], url) - - thumbnails = [] - layouts = try_get( - content, lambda x: x['teaserImageRef']['layouts'], dict) - if layouts: - for layout_key, layout_url in layouts.items(): - layout_url = url_or_none(layout_url) - if not layout_url: - continue - thumbnail = { - 'url': layout_url, - 'format_id': layout_key, - } - mobj = re.search(r'(?P\d+)x(?P\d+)', layout_key) - if mobj: - thumbnail.update({ - 'width': int(mobj.group('width')), - 'height': int(mobj.group('height')), - }) - thumbnails.append(thumbnail) - - chapter_marks = t.get('streamAnchorTag') or [] - chapter_marks.append({'anchorOffset': int_or_none(t.get('duration'))}) - chapters = [{ - 'start_time': chap.get('anchorOffset'), - 'end_time': next_chap.get('anchorOffset'), - 'title': chap.get('anchorLabel'), - } for chap, next_chap in zip(chapter_marks, chapter_marks[1:])] - - return merge_dicts(info, { - 'title': title, - 'description': content.get('leadParagraph') or content.get('teasertext'), - 'duration': int_or_none(t.get('duration')), - 'timestamp': unified_timestamp(content.get('editorialDate')), - 'thumbnails': thumbnails, - 'chapters': chapters or None, - 'episode': title, - **traverse_obj(content, ('programmeItem', 0, 'http://zdf.de/rels/target', { - 'series_id': ('http://zdf.de/rels/cmdm/series', 'seriesUuid', {str}), - 'series': ('http://zdf.de/rels/cmdm/series', 'seriesTitle', {str}), - 'season': ('http://zdf.de/rels/cmdm/season', 'seasonTitle', {str}), - 'season_number': ('http://zdf.de/rels/cmdm/season', 'seasonNumber', {int_or_none}), - 'season_id': ('http://zdf.de/rels/cmdm/season', 'seasonUuid', {str}), - 'episode_number': ('episodeNumber', {int_or_none}), - 'episode_id': ('contentId', {str}), - })), - }) - - def _extract_regular(self, url, player, video_id, query=None): - player_url = player['content'] - - content = self._call_api( - update_url_query(player_url, query), - video_id, 'content', player['apiToken'], url) - - return self._extract_entry(player_url, player, content, video_id) - - def _extract_mobile(self, video_id): - video = self._download_v2_doc(video_id) - - formats = [] - formitaeten = try_get(video, lambda x: x['document']['formitaeten'], list) - document = formitaeten and video['document'] - if formitaeten: - title = document['titel'] - content_id = document['basename'] - - format_urls = set() - for f in formitaeten or []: - self._extract_format(content_id, formats, format_urls, f) - - thumbnails = [] - teaser_bild = document.get('teaserBild') - if isinstance(teaser_bild, dict): - for thumbnail_key, thumbnail in teaser_bild.items(): - thumbnail_url = try_get( - thumbnail, lambda x: x['url'], str) - if thumbnail_url: - thumbnails.append({ - 'url': thumbnail_url, - 'id': thumbnail_key, - 'width': int_or_none(thumbnail.get('width')), - 'height': int_or_none(thumbnail.get('height')), - }) - - return { - 'id': content_id, - 'title': title, - 'description': document.get('beschreibung'), - 'duration': int_or_none(document.get('length')), - 'timestamp': unified_timestamp(document.get('date')) or unified_timestamp( - try_get(video, lambda x: x['meta']['editorialDate'], str)), - 'thumbnails': thumbnails, - 'subtitles': self._extract_subtitles(document), - 'formats': formats, - } + @staticmethod + def _extract_thumbnails(source): + return [{ + 'id': str(format_id), + 'url': url, + 'preference': 1 if format_id == 'original' else 0, + **traverse_obj(re.search(r'(?P\d+|auto)[Xx](?P\d+|auto)', str(format_id)), { + 'width': ('width', {int_or_none}), + 'height': ('height', {int_or_none}), + }), + } for format_id, url in traverse_obj(source, ({dict.items}, lambda _, v: url_or_none(v[1])))] class ZDFIE(ZDFBaseIE): - _VALID_URL = r'https?://www\.zdf\.de/(?:[^/]+/)*(?P[^/?#&]+)\.html' + _VALID_URL = [ + r'https?://(?:www\.)?zdf\.de/(?:video|play)/(?:[^/?#]+/)*(?P[^/?#]+)', + # /nachrichten/ sub-site URLs and legacy redirects from before the redesign in 2025-03 + r'https?://(?:www\.)?zdf\.de/(?:[^/?#]+/)*(?P[^/?#]+)\.html', + ] + IE_NAME = 'zdf' _TESTS = [{ - # Same as https://www.phoenix.de/sendungen/ereignisse/corona-nachgehakt/wohin-fuehrt-der-protest-in-der-pandemie-a-2050630.html - 'url': 'https://www.zdf.de/politik/phoenix-sendungen/wohin-fuehrt-der-protest-in-der-pandemie-100.html', - 'md5': '34ec321e7eb34231fd88616c65c92db0', + # Standalone video (i.e. not part of a playlist), video URL + 'url': 'https://www.zdf.de/video/dokus/sylt---deutschlands-edles-nordlicht-movie-100/sylt-deutschlands-edles-nordlicht-100', 'info_dict': { - 'id': '210222_phx_nachgehakt_corona_protest', + 'id': 'sylt-deutschlands-edles-nordlicht-100', 'ext': 'mp4', - 'title': 'Wohin führt der Protest in der Pandemie?', - 'description': 'md5:7d643fe7f565e53a24aac036b2122fbd', - 'duration': 1691, - 'timestamp': 1613948400, - 'upload_date': '20210221', + 'title': 'Sylt - Deutschlands edles Nordlicht', + 'description': 'md5:35407b810c2e1e33efbe15ef6e4c06c3', + 'duration': 810.0, + 'thumbnail': 'https://www.zdf.de/assets/sylt-118~original?cb=1613992485011', + 'series': 'Sylt - Deutschlands edles Nordlicht', + 'series_id': 'sylt---deutschlands-edles-nordlicht-movie-100', + 'timestamp': 1612462500, + 'upload_date': '20210204', + '_old_archive_ids': ['zdf 210402_1915_sendung_dok'], }, - 'skip': 'No longer available: "Diese Seite wurde leider nicht gefunden"', }, { - # Same as https://www.3sat.de/film/ab-18/10-wochen-sommer-108.html - 'url': 'https://www.zdf.de/dokumentation/ab-18/10-wochen-sommer-102.html', - 'md5': '0aff3e7bc72c8813f5e0fae333316a1d', + # Standalone video (i.e. not part of a playlist), play URL + 'url': 'https://www.zdf.de/play/dokus/sylt---deutschlands-edles-nordlicht-movie-100/sylt-deutschlands-edles-nordlicht-100', 'info_dict': { - 'id': '141007_ab18_10wochensommer_film', + 'id': 'sylt-deutschlands-edles-nordlicht-100', 'ext': 'mp4', - 'title': 'Ab 18! - 10 Wochen Sommer', - 'description': 'md5:8253f41dc99ce2c3ff892dac2d65fe26', - 'duration': 2660, - 'timestamp': 1608604200, - 'upload_date': '20201222', + 'title': 'Sylt - Deutschlands edles Nordlicht', + 'description': 'md5:35407b810c2e1e33efbe15ef6e4c06c3', + 'duration': 810.0, + 'thumbnail': 'https://www.zdf.de/assets/sylt-118~original?cb=1613992485011', + 'series': 'Sylt - Deutschlands edles Nordlicht', + 'series_id': 'sylt---deutschlands-edles-nordlicht-movie-100', + 'timestamp': 1612462500, + 'upload_date': '20210204', + '_old_archive_ids': ['zdf 210402_1915_sendung_dok'], }, - 'skip': 'No longer available: "Diese Seite wurde leider nicht gefunden"', + 'params': {'skip_download': True}, }, { - 'url': 'https://www.zdf.de/nachrichten/heute-journal/heute-journal-vom-30-12-2021-100.html', + # Standalone video (i.e. not part of a playlist), legacy URL before website redesign in 2025-03 + 'url': 'https://www.zdf.de/dokumentation/dokumentation-sonstige/sylt-deutschlands-edles-nordlicht-100.html', 'info_dict': { - 'id': '211230_sendung_hjo', + 'id': 'sylt-deutschlands-edles-nordlicht-100', 'ext': 'mp4', - 'description': 'md5:47dff85977bde9fb8cba9e9c9b929839', - 'duration': 1890.0, - 'upload_date': '20211230', - 'chapters': list, - 'thumbnail': 'md5:e65f459f741be5455c952cd820eb188e', - 'title': 'heute journal vom 30.12.2021', - 'timestamp': 1640897100, + 'title': 'Sylt - Deutschlands edles Nordlicht', + 'description': 'md5:35407b810c2e1e33efbe15ef6e4c06c3', + 'duration': 810.0, + 'thumbnail': 'https://www.zdf.de/assets/sylt-118~original?cb=1613992485011', + 'series': 'Sylt - Deutschlands edles Nordlicht', + 'series_id': 'sylt---deutschlands-edles-nordlicht-movie-100', + 'timestamp': 1612462500, + 'upload_date': '20210204', + '_old_archive_ids': ['zdf 210402_1915_sendung_dok'], }, - 'skip': 'No longer available: "Diese Seite wurde leider nicht gefunden"', + 'params': {'skip_download': True}, }, { - 'url': 'https://www.zdf.de/dokumentation/terra-x/die-magie-der-farben-von-koenigspurpur-und-jeansblau-100.html', + # Video belongs to a playlist, video URL + 'url': 'https://www.zdf.de/video/dokus/die-magie-der-farben-116/die-magie-der-farben-von-koenigspurpur-und-jeansblau-100', + 'md5': '1eda17eb40a9ead3046326e10b9c5973', 'info_dict': { - 'id': '151025_magie_farben2_tex', + 'id': 'die-magie-der-farben-von-koenigspurpur-und-jeansblau-100', 'ext': 'mp4', - 'duration': 2615.0, - 'title': 'Die Magie der Farben (2/2)', + 'title': 'Von Königspurpur bis Jeansblau', 'description': 'md5:a89da10c928c6235401066b60a6d5c1a', - 'timestamp': 1465021200, - 'thumbnail': 'https://www.zdf.de/assets/mauve-im-labor-100~768x432?cb=1464909117806', - 'upload_date': '20160604', - 'episode': 'Die Magie der Farben (2/2)', - 'episode_id': 'POS_954f4170-36a5-4a41-a6cf-78f1f3b1f127', - 'season': 'Staffel 1', + 'duration': 2615.0, + 'thumbnail': 'https://www.zdf.de/assets/koenigspurpur-bis-jeansblau-100~original?cb=1741857765971', 'series': 'Die Magie der Farben', + 'series_id': 'die-magie-der-farben-116', + 'season': 'Season 1', 'season_number': 1, - 'series_id': 'a39900dd-cdbd-4a6a-a413-44e8c6ae18bc', - 'season_id': '5a92e619-8a0f-4410-a3d5-19c76fbebb37', + 'episode': 'Episode 2', 'episode_number': 2, + 'timestamp': 1445797800, + 'upload_date': '20151025', + '_old_archive_ids': ['zdf 151025_magie_farben2_tex'], + }, + }, { + # Video belongs to a playlist, play URL + 'url': 'https://www.zdf.de/play/dokus/die-magie-der-farben-116/die-magie-der-farben-von-koenigspurpur-und-jeansblau-100', + 'md5': '1eda17eb40a9ead3046326e10b9c5973', + 'info_dict': { + 'id': 'die-magie-der-farben-von-koenigspurpur-und-jeansblau-100', + 'ext': 'mp4', + 'title': 'Von Königspurpur bis Jeansblau', + 'description': 'md5:a89da10c928c6235401066b60a6d5c1a', + 'duration': 2615.0, + 'thumbnail': 'https://www.zdf.de/assets/koenigspurpur-bis-jeansblau-100~original?cb=1741857765971', + 'series': 'Die Magie der Farben', + 'series_id': 'die-magie-der-farben-116', + 'season': 'Season 1', + 'season_number': 1, + 'episode': 'Episode 2', + 'episode_number': 2, + 'timestamp': 1445797800, + 'upload_date': '20151025', + '_old_archive_ids': ['zdf 151025_magie_farben2_tex'], + }, + 'params': {'skip_download': True}, + }, { + # Video belongs to a playlist, legacy URL before website redesign in 2025-03 + 'url': 'https://www.zdf.de/dokumentation/terra-x/die-magie-der-farben-von-koenigspurpur-und-jeansblau-100.html', + 'md5': '1eda17eb40a9ead3046326e10b9c5973', + 'info_dict': { + 'id': 'die-magie-der-farben-von-koenigspurpur-und-jeansblau-100', + 'ext': 'mp4', + 'title': 'Von Königspurpur bis Jeansblau', + 'description': 'md5:a89da10c928c6235401066b60a6d5c1a', + 'duration': 2615.0, + 'thumbnail': 'https://www.zdf.de/assets/koenigspurpur-bis-jeansblau-100~original?cb=1741857765971', + 'series': 'Die Magie der Farben', + 'series_id': 'die-magie-der-farben-116', + 'season': 'Season 1', + 'season_number': 1, + 'episode': 'Episode 2', + 'episode_number': 2, + 'timestamp': 1445797800, + 'upload_date': '20151025', + '_old_archive_ids': ['zdf 151025_magie_farben2_tex'], + }, + 'params': {'skip_download': True}, + }, { + # Video with chapters + # Also: video with sign-language variant + 'url': 'https://www.zdf.de/video/magazine/heute-journal-104/heute-journal-vom-19-12-2021-100', + 'md5': '6ada39465497a84fb98d48ffff69e7b7', + 'info_dict': { + 'id': 'heute-journal-vom-19-12-2021-100', + 'ext': 'mp4', + 'title': 'heute journal vom 19.12.2021', + 'description': 'md5:02504cf3b03777ff32fcc927d260c5dd', + 'duration': 1770.0, + 'thumbnail': 'https://epg-image.zdf.de/fotobase-webdelivery/images/273e5545-16e7-4ca3-898e-52fe9e06d964?layout=1920x1080', + 'chapters': 'count:11', + 'series': 'heute journal', + 'series_id': 'heute-journal-104', + 'season': 'Season 2021', + 'season_number': 2021, + 'episode': 'Episode 370', + 'episode_number': 370, + 'timestamp': 1639946700, + 'upload_date': '20211219', + # Videos with sign language variants must not have a 'dgs' suffix on their old archive IDs. + '_old_archive_ids': ['zdf 211219_sendung_hjo'], + }, + }, { + # Video that requires fallback extraction + 'url': 'https://www.zdf.de/nachrichten/politik/deutschland/koalitionsverhandlungen-spd-cdu-csu-dobrindt-100.html', + 'md5': 'c3a78514dd993a5781aa3afe50db51e2', + 'info_dict': { + 'id': 'koalitionsverhandlungen-spd-cdu-csu-dobrindt-100', + 'ext': 'mp4', + 'title': 'Dobrindt schließt Steuererhöhungen aus', + 'description': 'md5:9a117646d7b8df6bc902eb543a9c9023', + 'duration': 325, + 'thumbnail': 'https://www.zdf.de/assets/dobrindt-csu-berlin-direkt-100~1920x1080?cb=1743357653736', + 'timestamp': 1743374520, + 'upload_date': '20250330', + '_old_archive_ids': ['zdf 250330_clip_2_bdi'], }, }, { 'url': 'https://www.zdf.de/funk/druck-11790/funk-alles-ist-verzaubert-102.html', 'md5': '57af4423db0455a3975d2dc4578536bc', 'info_dict': { + 'id': 'funk-alles-ist-verzaubert-102', 'ext': 'mp4', - 'id': 'video_funk_1770473', - 'duration': 1278.0, 'title': 'Alles ist verzaubert', 'description': 'Die Neue an der Schule verdreht Ismail den Kopf.', + 'duration': 1278.0, + 'thumbnail': 'https://www.zdf.de/assets/teaser-funk-alles-ist-verzaubert-102~original?cb=1663848412907', + 'series': 'DRUCK', + 'series_id': 'funk-collection-funk-11790-1590', + 'season': 'Season 7', + 'season_number': 7, + 'episode': 'Episode 1', + 'episode_number': 1, 'timestamp': 1635520560, - 'thumbnail': 'https://www.zdf.de/assets/teaser-funk-alles-ist-verzaubert-102~1920x1080?cb=1663848412907', 'upload_date': '20211029', - 'episode': 'Alles ist verzaubert', + '_old_archive_ids': ['zdf video_funk_1770473'], }, + }, { + 'url': 'https://www.zdf.de/serien/soko-stuttgart/das-geld-anderer-leute-100.html', + 'info_dict': { + 'id': 'das-geld-anderer-leute-100', + 'ext': 'mp4', + 'title': 'Das Geld anderer Leute', + 'description': 'md5:cb6f660850dc5eb7d1ab776ea094959d', + 'duration': 2581.0, + 'thumbnail': 'https://epg-image.zdf.de/fotobase-webdelivery/images/e2d7e55a-09f0-424e-ac73-6cac4dd65f35?layout=1920x1080', + 'series': 'SOKO Stuttgart', + 'series_id': 'soko-stuttgart-104', + 'season': 'Season 11', + 'season_number': 11, + 'episode': 'Episode 10', + 'episode_number': 10, + 'timestamp': 1728983700, + 'upload_date': '20241015', + '_old_archive_ids': ['zdf 191205_1800_sendung_sok8'], + }, + }, { + 'url': 'https://www.zdf.de/serien/northern-lights/begegnung-auf-der-bruecke-100.html', + 'info_dict': { + 'id': 'begegnung-auf-der-bruecke-100', + 'ext': 'webm', + 'title': 'Begegnung auf der Brücke', + 'description': 'md5:e53a555da87447f7f1207f10353f8e45', + 'duration': 3083.0, + 'thumbnail': 'https://epg-image.zdf.de/fotobase-webdelivery/images/c5ff1d1f-f5c8-4468-86ac-1b2f1dbecc76?layout=1920x1080', + 'series': 'Northern Lights', + 'series_id': 'northern-lights-100', + 'season': 'Season 1', + 'season_number': 1, + 'episode': 'Episode 1', + 'episode_number': 1, + 'timestamp': 1738546500, + 'upload_date': '20250203', + '_old_archive_ids': ['zdf 240319_2310_sendung_not'], + }, + 'params': {'skip_download': 'geo-restricted http format'}, + }, { + # Same as https://www.phoenix.de/sendungen/ereignisse/corona-nachgehakt/wohin-fuehrt-der-protest-in-der-pandemie-a-2050630.html + 'url': 'https://www.zdf.de/politik/phoenix-sendungen/wohin-fuehrt-der-protest-in-der-pandemie-100.html', + 'only_matching': True, + }, { + # Same as https://www.3sat.de/film/ab-18/10-wochen-sommer-108.html + 'url': 'https://www.zdf.de/dokumentation/ab-18/10-wochen-sommer-102.html', + 'only_matching': True, }, { # Same as https://www.phoenix.de/sendungen/dokumentationen/gesten-der-maechtigen-i-a-89468.html?ref=suche 'url': 'https://www.zdf.de/politik/phoenix-sendungen/die-gesten-der-maechtigen-100.html', @@ -349,155 +428,323 @@ class ZDFIE(ZDFBaseIE): 'only_matching': True, }, { 'url': 'https://www.zdf.de/arte/todliche-flucht/page-video-artede-toedliche-flucht-16-100.html', - 'info_dict': { - 'id': 'video_artede_083871-001-A', - 'ext': 'mp4', - 'title': 'Tödliche Flucht (1/6)', - 'description': 'md5:e34f96a9a5f8abd839ccfcebad3d5315', - 'duration': 3193.0, - 'timestamp': 1641355200, - 'upload_date': '20220105', - }, - 'skip': 'No longer available "Diese Seite wurde leider nicht gefunden"', - }, { - 'url': 'https://www.zdf.de/serien/soko-stuttgart/das-geld-anderer-leute-100.html', - 'info_dict': { - 'id': '191205_1800_sendung_sok8', - 'ext': 'mp4', - 'title': 'Das Geld anderer Leute', - 'description': 'md5:cb6f660850dc5eb7d1ab776ea094959d', - 'duration': 2581.0, - 'timestamp': 1728983700, - 'upload_date': '20241015', - 'thumbnail': 'https://epg-image.zdf.de/fotobase-webdelivery/images/e2d7e55a-09f0-424e-ac73-6cac4dd65f35?layout=2400x1350', - 'series': 'SOKO Stuttgart', - 'series_id': 'f862ce9a-6dd1-4388-a698-22b36ac4c9e9', - 'season': 'Staffel 11', - 'season_number': 11, - 'season_id': 'ae1b4990-6d87-4970-a571-caccf1ba2879', - 'episode': 'Das Geld anderer Leute', - 'episode_number': 10, - 'episode_id': 'POS_7f367934-f2f0-45cb-9081-736781ff2d23', - }, + 'only_matching': True, }, { 'url': 'https://www.zdf.de/dokumentation/terra-x/unser-gruener-planet-wuesten-doku-100.html', - 'info_dict': { - 'id': '220525_green_planet_makingof_1_tropen_tex', - 'ext': 'mp4', - 'title': 'Making-of Unser grüner Planet - Tropen', - 'description': 'md5:d7c6949dc7c75c73c4ad51c785fb0b79', - 'duration': 435.0, - 'timestamp': 1653811200, - 'upload_date': '20220529', - 'format_note': 'hd, main', - 'thumbnail': 'https://www.zdf.de/assets/unser-gruener-planet-making-of-1-tropen-100~3840x2160?cb=1653493335577', - 'episode': 'Making-of Unser grüner Planet - Tropen', - }, - 'skip': 'No longer available: "Leider kein Video verfügbar"', - }, { - 'url': 'https://www.zdf.de/serien/northern-lights/begegnung-auf-der-bruecke-100.html', - 'info_dict': { - 'id': '240319_2310_sendung_not', - 'ext': 'mp4', - 'title': 'Begegnung auf der Brücke', - 'description': 'md5:e53a555da87447f7f1207f10353f8e45', - 'thumbnail': 'https://epg-image.zdf.de/fotobase-webdelivery/images/c5ff1d1f-f5c8-4468-86ac-1b2f1dbecc76?layout=2400x1350', - 'upload_date': '20250203', - 'duration': 3083.0, - 'timestamp': 1738546500, - 'series_id': '1d7a1879-01ee-4468-8237-c6b4ecd633c7', - 'series': 'Northern Lights', - 'season': 'Staffel 1', - 'season_number': 1, - 'season_id': '22ac26a2-4ea2-4055-ac0b-98b755cdf718', - 'episode': 'Begegnung auf der Brücke', - 'episode_number': 1, - 'episode_id': 'POS_71049438-024b-471f-b472-4fe2e490d1fb', - }, + 'only_matching': True, }] + _GRAPHQL_QUERY = ''' +query VideoByCanonical($canonical: String!) { + videoByCanonical(canonical: $canonical) { + canonical + title + leadParagraph + editorialDate + teaser { + description + image { + list + } + } + episodeInfo { + episodeNumber + seasonNumber + } + smartCollection { + canonical + title + } + currentMedia { + nodes { + ptmdTemplate + ... on VodMedia { + duration + aspectRatio + streamAnchorTags { + nodes { + anchorOffset + anchorLabel + } + } + vodMediaType + label + } + ... on LiveMedia { + start + stop + encryption + liveMediaType + label + } + id + } + } + } +} + ''' + + def _extract_ptmd(self, *args, **kwargs): + ptmd_data = super()._extract_ptmd(*args, **kwargs) + # This was the video id before the graphql redesign, other extractors still use it as such + old_archive_id = ptmd_data.pop('id') + ptmd_data['_old_archive_ids'] = [make_archive_id(self, old_archive_id)] + return ptmd_data + + # This fallback should generally only happen for pages under `zdf.de/nachrichten`. + # They are on a separate website for which GraphQL often doesn't return results. + # The API used here is no longer in use by official clients and likely deprecated. + # Long-term, news documents probably should use the API used by the mobile apps: + # https://zdf-prod-futura.zdf.de/news/documents/ (note 'news' vs 'mediathekV2') + def _extract_fallback(self, document_id): + video = self._download_json( + f'https://zdf-prod-futura.zdf.de/mediathekV2/document/{document_id}', + document_id, note='Downloading fallback metadata', + errnote='Failed to download fallback metadata') + document = video['document'] + + ptmd_url = traverse_obj(document, ( + ('streamApiUrlAndroid', ('streams', 0, 'streamApiUrlAndroid')), + {url_or_none}, any, {require('PTMD URL')})) + + thumbnails = [] + for thumbnail_key, thumbnail in traverse_obj(document, ('teaserBild', {dict.items}, ...)): + thumbnail_url = traverse_obj(thumbnail, ('url', {url_or_none})) + if not thumbnail_url: + continue + thumbnails.append({ + 'url': thumbnail_url, + 'id': thumbnail_key, + 'width': int_or_none(thumbnail.get('width')), + 'height': int_or_none(thumbnail.get('height')), + }) + + return { + 'thumbnails': thumbnails, + **traverse_obj(video, { + 'title': ('document', 'titel', {str}), + 'description': ('document', 'beschreibung', {str}), + 'timestamp': ( + (('document', 'date'), ('meta', 'editorialDate')), + {unified_timestamp}, any), + 'subtitles': ('document', 'captions', {self._extract_subtitles}), + }), + **self._extract_ptmd(ptmd_url, document_id, self._get_api_token()), + 'id': document_id, + } + def _real_extract(self, url): video_id = self._match_id(url) + video_data = self._download_graphql(video_id, 'video metadata', body={ + 'operationName': 'VideoByCanonical', + 'query': self._GRAPHQL_QUERY, + 'variables': {'canonical': video_id}, + })['data']['videoByCanonical'] - webpage = self._download_webpage(url, video_id, fatal=False) - if webpage: - player = self._extract_player(webpage, url, fatal=False) - if player: - return self._extract_regular(url, player, video_id, query={'profile': 'player-3'}) + if not video_data: + return self._extract_fallback(video_id) - return self._extract_mobile(video_id) + aspect_ratio = None + ptmd_urls = [] + for node in traverse_obj(video_data, ('currentMedia', 'nodes', lambda _, v: v['ptmdTemplate'])): + ptmd_url = self._expand_ptmd_template('https://api.zdf.de', node['ptmdTemplate']) + # Smuggle vod_media_type so that _extract_ptmd is aware of 'DGS' variants + if vod_media_type := node.get('vodMediaType'): + ptmd_url = smuggle_url(ptmd_url, {'vod_media_type': vod_media_type}) + ptmd_urls.append(ptmd_url) + if not aspect_ratio: + aspect_ratio = self._parse_aspect_ratio(node.get('aspectRatio')) + + return { + **traverse_obj(video_data, { + 'title': ('title', {str}), + 'description': (('leadParagraph', ('teaser', 'description')), any, {str}), + 'timestamp': ('editorialDate', {parse_iso8601}), + 'thumbnails': ('teaser', 'image', 'list', {self._extract_thumbnails}), + 'episode_number': ('episodeInfo', 'episodeNumber', {int_or_none}), + 'season_number': ('episodeInfo', 'seasonNumber', {int_or_none}), + 'series': ('smartCollection', 'title', {str}), + 'series_id': ('smartCollection', 'canonical', {str}), + 'chapters': ('currentMedia', 'nodes', 0, 'streamAnchorTags', 'nodes', {self._extract_chapters}), + }), + **self._extract_ptmd(ptmd_urls, video_id, self._get_api_token(), aspect_ratio), + 'id': video_id, + } class ZDFChannelIE(ZDFBaseIE): - _VALID_URL = r'https?://www\.zdf\.de/(?:[^/?#]+/)*(?P[^/?#&]+)' + _VALID_URL = r'https?://www\.zdf\.de/(?:[^/?#]+/)*(?P[^/?#]+)' + IE_NAME = 'zdf:channel' _TESTS = [{ + # Playlist, legacy URL before website redesign in 2025-03 'url': 'https://www.zdf.de/sport/das-aktuelle-sportstudio', 'info_dict': { - 'id': 'das-aktuelle-sportstudio', + 'id': 'das-aktuelle-sportstudio-220', 'title': 'das aktuelle sportstudio', + 'description': 'md5:e46c785324238a03edcf8b301c5fd5dc', }, - 'playlist_mincount': 18, + 'playlist_mincount': 25, }, { - 'url': 'https://www.zdf.de/dokumentation/planet-e', + # Playlist, current URL + 'url': 'https://www.zdf.de/sport/das-aktuelle-sportstudio-220', 'info_dict': { - 'id': 'planet-e', - 'title': 'planet e.', - 'description': 'md5:87e3b9c66a63cf1407ee443d2c4eb88e', + 'id': 'das-aktuelle-sportstudio-220', + 'title': 'das aktuelle sportstudio', + 'description': 'md5:e46c785324238a03edcf8b301c5fd5dc', }, - 'playlist_mincount': 50, + 'playlist_mincount': 25, + }, { + # Standalone video (i.e. not part of a playlist), collection URL + 'add_ie': [ZDFIE.ie_key()], + 'url': 'https://www.zdf.de/dokus/sylt---deutschlands-edles-nordlicht-movie-100', + 'info_dict': { + 'id': 'sylt-deutschlands-edles-nordlicht-100', + 'ext': 'mp4', + 'title': 'Sylt - Deutschlands edles Nordlicht', + 'description': 'md5:35407b810c2e1e33efbe15ef6e4c06c3', + 'duration': 810.0, + 'thumbnail': 'https://www.zdf.de/assets/sylt-118~original?cb=1613992485011', + 'series': 'Sylt - Deutschlands edles Nordlicht', + 'series_id': 'sylt---deutschlands-edles-nordlicht-movie-100', + 'timestamp': 1612462500, + 'upload_date': '20210204', + '_old_archive_ids': ['zdf 210402_1915_sendung_dok'], + }, + 'params': {'skip_download': True}, }, { 'url': 'https://www.zdf.de/gesellschaft/aktenzeichen-xy-ungeloest', 'info_dict': { - 'id': 'aktenzeichen-xy-ungeloest', + 'id': 'aktenzeichen-xy-ungeloest-110', 'title': 'Aktenzeichen XY... Ungelöst', - 'description': 'md5:623ede5819c400c6d04943fa8100e6e7', + 'description': 'md5:b79ac0d64b979e53cbe510c0ca9cb7be', }, 'playlist_mincount': 2, }, { 'url': 'https://www.zdf.de/serien/taunuskrimi/', - 'only_matching': True, + 'info_dict': { + 'id': 'taunuskrimi-100', + 'title': 'Taunuskrimi', + 'description': 'md5:ee7204e9c625c3b611d1274f9d0e3070', + }, + 'playlist_mincount': 8, + }, { + 'url': 'https://www.zdf.de/serien/taunuskrimi/?staffel=1', + 'info_dict': { + 'id': 'taunuskrimi-100-s1', + 'title': 'Taunuskrimi - Season 1', + 'description': 'md5:ee7204e9c625c3b611d1274f9d0e3070', + }, + 'playlist_count': 7, + }, { + 'url': 'https://www.zdf.de/magazine/heute-journal-104', + 'info_dict': { + 'id': 'heute-journal-104', + 'title': 'heute journal', + 'description': 'md5:6edad39189abf8431795d3d6d7f986b3', + }, + 'playlist_mincount': 500, + }, { + 'url': 'https://www.zdf.de/magazine/heute-journal-104?staffel=2024', + 'info_dict': { + 'id': 'heute-journal-104-s2024', + 'title': 'heute journal - Season 2024', + 'description': 'md5:6edad39189abf8431795d3d6d7f986b3', + }, + 'playlist_count': 242, }] + _PAGE_SIZE = 24 + @classmethod def suitable(cls, url): return False if ZDFIE.suitable(url) else super().suitable(url) - def _extract_entry(self, entry): - return self.url_result( - entry['sharingUrl'], ZDFIE, **traverse_obj(entry, { - 'id': ('basename', {str}), - 'title': ('titel', {str}), - 'description': ('beschreibung', {str}), - 'duration': ('length', {float_or_none}), - 'season_number': ('seasonNumber', {int_or_none}), - 'episode_number': ('episodeNumber', {int_or_none}), - })) + def _fetch_page(self, playlist_id, canonical_id, season_idx, season_number, page_number, cursor=None): + return self._download_graphql( + playlist_id, f'season {season_number} page {page_number} JSON', query={ + 'operationName': 'seasonByCanonical', + 'variables': json.dumps(filter_dict({ + 'seasonIndex': season_idx, + 'canonical': canonical_id, + 'episodesPageSize': self._PAGE_SIZE, + 'episodesAfter': cursor, + })), + 'extensions': json.dumps({ + 'persistedQuery': { + 'version': 1, + 'sha256Hash': '9412a0f4ac55dc37d46975d461ec64bfd14380d815df843a1492348f77b5c99a', + }, + }), + })['data']['smartCollectionByCanonical'] - def _entries(self, data, document_id): - for entry in traverse_obj(data, ( - 'cluster', lambda _, v: v['type'] == 'teaser', - # If 'brandId' differs, it is a 'You might also like' video. Filter these out - 'teaser', lambda _, v: v['type'] == 'video' and v['brandId'] == document_id and v['sharingUrl'], - )): - yield self._extract_entry(entry) + def _entries(self, playlist_id, canonical_id, season_numbers, requested_season_number): + for season_idx, season_number in enumerate(season_numbers): + if requested_season_number is not None and requested_season_number != season_number: + continue + + cursor = None + for page_number in itertools.count(1): + page = self._fetch_page( + playlist_id, canonical_id, season_idx, season_number, page_number, cursor) + + nodes = traverse_obj(page, ('seasons', 'nodes', ...)) + + for episode in traverse_obj(nodes, ( + ..., 'episodes', 'nodes', lambda _, v: url_or_none(v['sharingUrl']), + )): + yield self.url_result( + episode['sharingUrl'], ZDFIE, + **traverse_obj(episode, { + 'id': ('canonical', {str}), + 'title': ('teaser', 'title', {str}), + 'description': (('leadParagraph', ('teaser', 'description')), any, {str}), + 'timestamp': ('editorialDate', {parse_iso8601}), + 'episode_number': ('episodeInfo', 'episodeNumber', {int_or_none}), + 'season_number': ('episodeInfo', 'seasonNumber', {int_or_none}), + })) + + page_info = traverse_obj(nodes, (-1, 'episodes', 'pageInfo', {dict})) or {} + if not page_info.get('hasNextPage') or not page_info.get('endCursor'): + break + cursor = page_info['endCursor'] def _real_extract(self, url): - channel_id = self._match_id(url) - webpage = self._download_webpage(url, channel_id) - document_id = self._search_regex( - r'docId\s*:\s*(["\'])(?P(?:(?!\1).)+)\1', webpage, 'document id', group='doc_id') - data = self._download_v2_doc(document_id) + canonical_id = self._match_id(url) + # Make sure to get the correct ID in case of redirects + urlh = self._request_webpage(url, canonical_id) + canonical_id = self._search_regex(self._VALID_URL, urlh.url, 'channel id', group='id') + season_number = traverse_obj(parse_qs(url), ('staffel', -1, {int_or_none})) + playlist_id = join_nonempty(canonical_id, season_number and f's{season_number}') - main_video = traverse_obj(data, ( - 'cluster', lambda _, v: v['type'] == 'teaserContent', - 'teaser', lambda _, v: v['type'] == 'video' and v['basename'] and v['sharingUrl'], any)) or {} + collection_data = self._download_graphql( + playlist_id, 'smart collection data', query={ + 'operationName': 'GetSmartCollectionByCanonical', + 'variables': json.dumps({ + 'canonical': canonical_id, + 'videoPageSize': 100, # Use max page size to get episodes from all seasons + }), + 'extensions': json.dumps({ + 'persistedQuery': { + 'version': 1, + 'sha256Hash': 'cb49420e133bd668ad895a8cea0e65cba6aa11ac1cacb02341ff5cf32a17cd02', + }, + }), + })['data']['smartCollectionByCanonical'] + video_data = traverse_obj(collection_data, ('video', {dict})) or {} + season_numbers = traverse_obj(collection_data, ('seasons', 'seasons', ..., 'number', {int_or_none})) - if not self._yes_playlist(channel_id, main_video.get('basename')): - return self._extract_entry(main_video) + if not self._yes_playlist( + season_numbers and playlist_id, + url_or_none(video_data.get('sharingUrl')) and video_data.get('canonical'), + ): + return self.url_result(video_data['sharingUrl'], ZDFIE, video_data['canonical']) + + if season_number is not None and season_number not in season_numbers: + raise ExtractorError(f'Season {season_number} was not found in the collection data') return self.playlist_result( - self._entries(data, document_id), channel_id, - re.split(r'\s+[-|]\s+ZDF(?:mediathek)?$', self._og_search_title(webpage) or '')[0] or None, - join_nonempty( - 'headline', 'text', delim='\n\n', - from_dict=traverse_obj(data, ('shortText', {dict}), default={})) or None) + self._entries(playlist_id, canonical_id, season_numbers, season_number), + playlist_id, join_nonempty( + traverse_obj(collection_data, ('title', {str})), + season_number and f'Season {season_number}', delim=' - '), + traverse_obj(collection_data, ('infoText', {str})))