]+\bclass="article_title">([^<]+)<', webpage, 'title')
@@ -146,7 +89,7 @@ class MediaKlikkIE(InfoExtractor):
upload_date = unified_strdate(self._html_search_regex(
r']+\bclass="article_date">([^<]+)<', webpage, 'upload date', default=None))
- player_data['video'] = player_data.pop('token')
+ player_data['video'] = urllib.parse.unquote(player_data.pop('token'))
player_page = self._download_webpage(
'https://player.mediaklikk.hu/playernew/player.php', video_id,
query=player_data, headers={'Referer': url})
diff --git a/yt_dlp/extractor/medialaan.py b/yt_dlp/extractor/medialaan.py
index bce20dcfd7..5aa5d4dfb0 100644
--- a/yt_dlp/extractor/medialaan.py
+++ b/yt_dlp/extractor/medialaan.py
@@ -1,15 +1,73 @@
-import re
-
from .common import InfoExtractor
from ..utils import (
+ clean_html,
+ determine_ext,
extract_attributes,
int_or_none,
- mimetype2ext,
- parse_iso8601,
+ parse_resolution,
+ str_or_none,
+ url_or_none,
)
+from ..utils.traversal import find_elements, traverse_obj
+
+
+class MedialaanBaseIE(InfoExtractor):
+ def _extract_from_mychannels_api(self, mychannels_id):
+ webpage = self._download_webpage(
+ f'https://mychannels.video/embed/{mychannels_id}', mychannels_id)
+ brand_config = self._search_json(
+ r'window\.mychannels\.brand_config\s*=', webpage, 'brand config', mychannels_id)
+ response = self._download_json(
+ f'https://api.mychannels.world/v1/embed/video/{mychannels_id}',
+ mychannels_id, headers={'X-Mychannels-Brand': brand_config['brand']})
+ formats = []
+ for stream in traverse_obj(response, (
+ 'streams', lambda _, v: url_or_none(v['url']),
+ )):
+ source_url = stream['url']
+ ext = determine_ext(source_url)
+ if ext == 'm3u8':
+ formats.extend(self._extract_m3u8_formats(
+ source_url, mychannels_id, 'mp4', m3u8_id='hls', fatal=False))
+ else:
+ format_id = traverse_obj(stream, ('quality', {str}))
+ formats.append({
+ 'ext': ext,
+ 'format_id': format_id,
+ 'url': source_url,
+ **parse_resolution(format_id),
+ })
-class MedialaanIE(InfoExtractor):
+ return {
+ 'id': mychannels_id,
+ 'formats': formats,
+ **traverse_obj(response, {
+ 'title': ('title', {clean_html}),
+ 'description': ('description', {clean_html}, filter),
+ 'duration': ('durationMs', {int_or_none(scale=1000)}, {lambda x: x if x >= 0 else None}),
+ 'genres': ('genre', 'title', {str}, filter, all, filter),
+ 'is_live': ('live', {bool}),
+ 'release_timestamp': ('publicationTimestampMs', {int_or_none(scale=1000)}),
+ 'tags': ('tags', ..., 'title', {str}, filter, all, filter),
+ 'thumbnail': ('image', 'baseUrl', {url_or_none}),
+ }),
+ **traverse_obj(response, ('channel', {
+ 'channel': ('title', {clean_html}),
+ 'channel_id': ('id', {str_or_none}),
+ })),
+ **traverse_obj(response, ('organisation', {
+ 'uploader': ('title', {clean_html}),
+ 'uploader_id': ('id', {str_or_none}),
+ })),
+ **traverse_obj(response, ('show', {
+ 'series': ('title', {clean_html}),
+ 'series_id': ('id', {str_or_none}),
+ })),
+ }
+
+
+class MedialaanIE(MedialaanBaseIE):
_VALID_URL = r'''(?x)
https?://
(?:
@@ -32,7 +90,7 @@ class MedialaanIE(InfoExtractor):
tubantia|
volkskrant
)\.nl
- )/video/(?:[^/]+/)*[^/?]+~p
+ )/videos?/(?:[^/?#]+/)*[^/?]+(?:-|~p)
)
(?P\d+)
'''
@@ -42,18 +100,83 @@ class MedialaanIE(InfoExtractor):
'id': '193993',
'ext': 'mp4',
'title': 'De terugkeer van Ally de Aap en wie vertrekt er nog bij NAC?',
- 'timestamp': 1611663540,
- 'upload_date': '20210126',
+ 'description': 'In een nieuwe Gegenpressing video bespreken Yadran Blanco en Dennis Kas het nieuws omrent NAC.',
'duration': 238,
- },
- 'params': {
- 'skip_download': True,
+ 'channel': 'BN DeStem',
+ 'channel_id': '418',
+ 'genres': ['Sports'],
+ 'release_date': '20210126',
+ 'release_timestamp': 1611663540,
+ 'series': 'Korte Reportage',
+ 'series_id': '972',
+ 'tags': 'count:2',
+ 'thumbnail': r're:https?://images\.mychannels\.video/imgix/.+\.(?:jpe?g|png)',
+ 'uploader': 'BN De Stem',
+ 'uploader_id': '26',
},
}, {
'url': 'https://www.gelderlander.nl/video/kanalen/degelderlander~c320/series/snel-nieuws~s984/noodbevel-in-doetinchem-politie-stuurt-mensen-centrum-uit~p194093',
- 'only_matching': True,
+ 'info_dict': {
+ 'id': '194093',
+ 'ext': 'mp4',
+ 'title': 'Noodbevel in Doetinchem: politie stuurt mensen centrum uit',
+ 'description': 'md5:77e85b2cb26cfff9dc1fe2b1db524001',
+ 'duration': 44,
+ 'channel': 'De Gelderlander',
+ 'channel_id': '320',
+ 'genres': ['News'],
+ 'release_date': '20210126',
+ 'release_timestamp': 1611690600,
+ 'series': 'Snel Nieuws',
+ 'series_id': '984',
+ 'tags': 'count:1',
+ 'thumbnail': r're:https?://images\.mychannels\.video/imgix/.+\.(?:jpe?g|png)',
+ 'uploader': 'De Gelderlander',
+ 'uploader_id': '25',
+ },
}, {
- 'url': 'https://embed.mychannels.video/sdk/production/193993?options=TFTFF_default',
+ 'url': 'https://www.7sur7.be/videos/production/lla-tendance-tiktok-qui-enflamme-lespagne-707650',
+ 'info_dict': {
+ 'id': '707650',
+ 'ext': 'mp4',
+ 'title': 'La tendance TikTok qui enflamme l’Espagne',
+ 'description': 'md5:c7ec4cb733190f227fc8935899f533b5',
+ 'duration': 70,
+ 'channel': 'Lifestyle',
+ 'channel_id': '770',
+ 'genres': ['Beauty & Lifestyle'],
+ 'release_date': '20240906',
+ 'release_timestamp': 1725617330,
+ 'series': 'Lifestyle',
+ 'series_id': '1848',
+ 'tags': 'count:1',
+ 'thumbnail': r're:https?://images\.mychannels\.video/imgix/.+\.(?:jpe?g|png)',
+ 'uploader': '7sur7',
+ 'uploader_id': '67',
+ },
+ }, {
+ 'url': 'https://mychannels.video/embed/313117',
+ 'info_dict': {
+ 'id': '313117',
+ 'ext': 'mp4',
+ 'title': str,
+ 'description': 'md5:255e2e52f6fe8a57103d06def438f016',
+ 'channel': 'AD',
+ 'channel_id': '238',
+ 'genres': ['News'],
+ 'live_status': 'is_live',
+ 'release_date': '20241225',
+ 'release_timestamp': 1735169425,
+ 'series': 'Nieuws Update',
+ 'series_id': '3337',
+ 'tags': 'count:1',
+ 'thumbnail': r're:https?://images\.mychannels\.video/imgix/.+\.(?:jpe?g|png)',
+ 'uploader': 'AD',
+ 'uploader_id': '1',
+ },
+ 'params': {'skip_download': 'Livestream'},
+ }, {
+ 'url': 'https://embed.mychannels.video/sdk/production/193993',
'only_matching': True,
}, {
'url': 'https://embed.mychannels.video/script/production/193993',
@@ -61,51 +184,41 @@ class MedialaanIE(InfoExtractor):
}, {
'url': 'https://embed.mychannels.video/production/193993',
'only_matching': True,
- }, {
- 'url': 'https://mychannels.video/embed/193993',
- 'only_matching': True,
}, {
'url': 'https://embed.mychannels.video/embed/193993',
'only_matching': True,
}]
+ _WEBPAGE_TESTS = [{
+ 'url': 'https://www.demorgen.be/snelnieuws/tom-waes-promoot-alcoholtesten-op-werchter-ik-ben-de-laatste-persoon-die-met-de-vinger-moet-wijzen~b7457c0d/',
+ 'info_dict': {
+ 'id': '1576607',
+ 'ext': 'mp4',
+ 'title': 'Tom Waes blaastest',
+ 'channel': 'De Morgen',
+ 'channel_id': '352',
+ 'description': 'Tom Waes werkt mee aan een alcoholcampagne op Werchter',
+ 'duration': 62,
+ 'genres': ['News'],
+ 'release_date': '20250705',
+ 'release_timestamp': 1751730795,
+ 'series': 'Nieuwsvideo\'s',
+ 'series_id': '1683',
+ 'tags': 'count:1',
+ 'thumbnail': r're:https?://video-images\.persgroep\.be/aws_generated.+\.jpg',
+ 'uploader': 'De Morgen',
+ 'uploader_id': '17',
+ },
+ 'params': {'extractor_args': {'generic': {'impersonate': ['chrome']}}},
+ }]
@classmethod
def _extract_embed_urls(cls, url, webpage):
- entries = []
- for element in re.findall(r'(]+data-mychannels-type="video"[^>]*>)', webpage):
- mychannels_id = extract_attributes(element).get('data-mychannels-id')
- if mychannels_id:
- entries.append('https://mychannels.video/embed/' + mychannels_id)
- return entries
+ yield from traverse_obj(webpage, (
+ {find_elements(tag='div', attr='data-mychannels-type', value='video', html=True)},
+ ..., {extract_attributes}, 'data-mychannels-id', {str}, filter,
+ {lambda x: f'https://mychannels.video/embed/{x}'}))
def _real_extract(self, url):
- production_id = self._match_id(url)
- production = self._download_json(
- 'https://embed.mychannels.video/sdk/production/' + production_id,
- production_id, query={'options': 'UUUU_default'})['productions'][0]
- title = production['title']
+ mychannels_id = self._match_id(url)
- formats = []
- for source in (production.get('sources') or []):
- src = source.get('src')
- if not src:
- continue
- ext = mimetype2ext(source.get('type'))
- if ext == 'm3u8':
- formats.extend(self._extract_m3u8_formats(
- src, production_id, 'mp4', 'm3u8_native',
- m3u8_id='hls', fatal=False))
- else:
- formats.append({
- 'ext': ext,
- 'url': src,
- })
-
- return {
- 'id': production_id,
- 'title': title,
- 'formats': formats,
- 'thumbnail': production.get('posterUrl'),
- 'timestamp': parse_iso8601(production.get('publicationDate'), ' '),
- 'duration': int_or_none(production.get('duration')) or None,
- }
+ return self._extract_from_mychannels_api(mychannels_id)
diff --git a/yt_dlp/extractor/megatvcom.py b/yt_dlp/extractor/megatvcom.py
index 93c7e7dc08..826a9a3b1f 100644
--- a/yt_dlp/extractor/megatvcom.py
+++ b/yt_dlp/extractor/megatvcom.py
@@ -31,10 +31,9 @@ class MegaTVComIE(MegaTVComBaseIE):
IE_NAME = 'megatvcom'
IE_DESC = 'megatv.com videos'
_VALID_URL = r'https?://(?:www\.)?megatv\.com/(?:\d{4}/\d{2}/\d{2}|[^/]+/(?P\d+))/(?P[^/]+)'
-
_TESTS = [{
+ # FIXME: Unable to extract article id
'url': 'https://www.megatv.com/2021/10/23/egkainia-gia-ti-nea-skini-omega-tou-dimotikou-theatrou-peiraia/',
- 'md5': '6546a1a37fff0dd51c9dce5f490b7d7d',
'info_dict': {
'id': '520979',
'ext': 'mp4',
@@ -43,20 +42,19 @@ class MegaTVComIE(MegaTVComBaseIE):
'timestamp': 1634975747,
'upload_date': '20211023',
'display_id': 'egkainia-gia-ti-nea-skini-omega-tou-dimotikou-theatrou-peiraia',
- 'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/10/ΠΕΙΡΑΙΑΣ-1024x450.jpg',
+ 'thumbnail': r're:https?://www\.megatv\.com/wp-content/uploads/.+\.jpg',
},
}, {
'url': 'https://www.megatv.com/tvshows/527800/epeisodio-65-12/',
- 'md5': 'cba2085d45c1abeb8e7e9b7e1d6c0072',
'info_dict': {
'id': '527800',
'ext': 'mp4',
- 'title': 'md5:fc322cb51f682eecfe2f54cd5ab3a157',
+ 'title': 'Η Γη της Ελιάς: Επεισόδιο 65 - A\' ΚΥΚΛΟΣ ',
'description': 'md5:b2b7ed3690a78f2a0156eb790fdc00df',
'timestamp': 1636048859,
'upload_date': '20211104',
'display_id': 'epeisodio-65-12',
- 'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/11/16-1-1.jpg',
+ 'thumbnail': r're:https?://www\.megatv\.com/wp-content/uploads/.+\.jpg',
},
}]
@@ -104,8 +102,8 @@ class MegaTVComEmbedIE(MegaTVComBaseIE):
IE_DESC = 'megatv.com embedded videos'
_VALID_URL = r'(?:https?:)?//(?:www\.)?megatv\.com/embed/?\?p=(?P\d+)'
_EMBED_REGEX = [rf'''