mirrors/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2025-07-04 04:18:31 +00:00
Code Issues Projects Releases Wiki Activity

Apply suggestions

Browse Source
This commit is contained in:
doe1080 2025-05-19 12:30:19 +09:00
parent a27624343b
commit 39e2da0a29
1 changed files with 27 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
yt_dlp/extractor/telemb.py
View File

@ -1,5 +1,8 @@
import itertools
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import ( from ..utils import (
ExtractorError,
clean_html, clean_html,
extract_attributes, extract_attributes,
float_or_none, float_or_none,
@ -24,15 +27,13 @@ class TeleMBIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Frameries - Un concours pour conducteurs d\'engins de chantier', 'title': 'Frameries - Un concours pour conducteurs d\'engins de chantier',
'creators': ['Sabine Dupont'], 'creators': ['Sabine Dupont'],
'description': 'md5:bfb8fdff559b64684bb005ce4901af12', 'description': 'md5:1dc04a3aa56c5228503071baa8b4cc97',
'display_id': 'frameries-un-concours-pour-conducteurs-dengins-de-chantier', 'display_id': 'frameries-un-concours-pour-conducteurs-dengins-de-chantier',
'duration': 144.6, 'duration': 144.6,
'location': 'Frameries', 'location': 'Frameries',
'release_date': '20250515', 'release_date': '20250515',
'release_timestamp': 1747319520, 'release_timestamp': 1747319520,
'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)', 'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)',
'timestamp': 1747319229,
'upload_date': '20250515',
}, },
}, { }, {
'url': 'https://www.telemb.be/sports/karate-cinq-karatekas-du-bushikai-wasmuel-la-coupe-du-monde-tokyo/37849', 'url': 'https://www.telemb.be/sports/karate-cinq-karatekas-du-bushikai-wasmuel-la-coupe-du-monde-tokyo/37849',
@ -41,15 +42,13 @@ class TeleMBIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Karaté : Cinq karatékas du Bushikai Wasmuel à la Coupe du Monde à Tokyo', 'title': 'Karaté : Cinq karatékas du Bushikai Wasmuel à la Coupe du Monde à Tokyo',
'creators': ['Jacob Hemptinne'], 'creators': ['Jacob Hemptinne'],
'description': 'md5:82ebfa7a4ddd359c9e05b5a8c8ab04c5', 'description': 'md5:17f2d55a1533a69079cc21eadd14725f',
'display_id': 'karate-cinq-karatekas-du-bushikai-wasmuel-la-coupe-du-monde-tokyo', 'display_id': 'karate-cinq-karatekas-du-bushikai-wasmuel-la-coupe-du-monde-tokyo',
'duration': 211.6, 'duration': 211.6,
'location': 'Quaregnon', 'location': 'Quaregnon',
'release_date': '20250512', 'release_date': '20250512',
'release_timestamp': 1747066800, 'release_timestamp': 1747066800,
'thumbnail': r're:https?://www\.telemb\.be/cdn/.+\.(?:jpe?g|png)', 'thumbnail': r're:https?://www\.telemb\.be/cdn/.+\.(?:jpe?g|png)',
'timestamp': 1746987989,
'upload_date': '20250511',
}, },
}, { }, {
'url': 'https://www.telemb.be/replay/emission/les-infos/les-infos-16052025/36502', 'url': 'https://www.telemb.be/replay/emission/les-infos/les-infos-16052025/36502',
@ -57,14 +56,13 @@ class TeleMBIE(InfoExtractor):
'id': '36502', 'id': '36502',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Les Infos - 16/05/2025', 'title': 'Les Infos - 16/05/2025',
'description': 'md5:d41d8cd98f00b204e9800998ecf8427e', 'creators': ['Télé MB'],
'description': 'md5:dff75a3a51c769696c23454e932ff720',
'display_id': 'les-infos-16052025', 'display_id': 'les-infos-16052025',
'duration': 1144.32, 'duration': 1144.32,
'release_date': '20250516', 'release_date': '20250516',
'release_timestamp': 1747412520, 'release_timestamp': 1747412520,
'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)', 'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)',
'timestamp': 1747408485,
'upload_date': '20250516',
}, },
}, { }, {
'url': 'https://www.telemb.be/actu/linvite-des-infos/le-cma-de-jemappes-fetera-ses-20-ans-ce-week-end/36711', 'url': 'https://www.telemb.be/actu/linvite-des-infos/le-cma-de-jemappes-fetera-ses-20-ans-ce-week-end/36711',
@ -73,15 +71,13 @@ class TeleMBIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Le CMA de Jemappes fêtera ses 20 ans ce week-end', 'title': 'Le CMA de Jemappes fêtera ses 20 ans ce week-end',
'creators': ['Loélia Chais'], 'creators': ['Loélia Chais'],
'description': 'md5:d41d8cd98f00b204e9800998ecf8427e', 'description': 'md5:f244efcce22f217df044d755e116ddef',
'display_id': 'le-cma-de-jemappes-fetera-ses-20-ans-ce-week-end', 'display_id': 'le-cma-de-jemappes-fetera-ses-20-ans-ce-week-end',
'duration': 316.08, 'duration': 316.08,
'location': 'Mons', 'location': 'Mons',
'release_date': '20241128', 'release_date': '20241128',
'release_timestamp': 1732787226, 'release_timestamp': 1732787226,
'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)', 'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)',
'timestamp': 1732727988,
'upload_date': '20241127',
}, },
}] }]
@ -89,16 +85,15 @@ def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
display_id = self._match_valid_url(url).group('alt_id').split('/')[-1] display_id = self._match_valid_url(url).group('alt_id').split('/')[-1]
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
json_ld = next(itertools.islice(self._yield_json_ld(webpage, video_id), 1, 2), {})
player_info = traverse_obj(webpage, ( data_video_id = traverse_obj(webpage, (
{find_element(cls='freecaster-player', html=True)}, {extract_attributes}, { {find_element(cls='freecaster-player', html=True)},
'id': ('data-video-id', {str_or_none}), {extract_attributes}, 'data-video-id', {str_or_none}))
'thumbnail': ('data-poster', {lambda x: self._proto_relative_url(x)}, {url_or_none}), if not (video_info := traverse_obj(self._download_json(
}, f'https://tvlocales-player-v12.freecaster.com/embed/{data_video_id}.json', video_id,
)) ), ('video', {dict}), default={})):
video_info = self._download_json( raise ExtractorError('Failed to fetch video information')
f'https://tvlocales-player-v12.freecaster.com/embed/{player_info.pop("id")}.json', video_id,
).get('video')
qualities = { qualities = {
'3': (640, 360), '3': (640, 360),
@ -108,12 +103,13 @@ def _real_extract(self, url):
} }
formats = [] formats = []
for src in traverse_obj(video_info, ('src', lambda _, v: v['src'])): for src in traverse_obj(video_info, ('src', lambda _, v: v['src'])):
src_url = src['src'] if not (src_url := url_or_none(src['src'])):
continue
ext = mimetype2ext(src.get('type')) ext = mimetype2ext(src.get('type'))
if ext == 'mp4': if ext == 'mp4':
quality = src_url.rpartition('_')[2].removesuffix('.mp4') quality = src_url.rpartition('_')[2].removesuffix('.mp4')
width, height = qualities.get(quality) width, height = qualities.get(quality, (None, None))
formats.append({ formats.append({
'acodec': 'mp4a.40.2', 'acodec': 'mp4a.40.2',
'ext': ext, 'ext': ext,
@ -146,19 +142,15 @@ def _real_extract(self, url):
return { return {
'id': video_id, 'id': video_id,
'display_id': display_id, 'display_id': display_id,
'duration': traverse_obj(video_info, ('duration', {float_or_none})),
'formats': formats, 'formats': formats,
**player_info, 'location': traverse_obj(webpage, (
**traverse_obj(webpage, { {find_element(cls='content-location')}, {clean_html})),
'creator': ({find_element(cls='content-author')}, {clean_html}), **traverse_obj(json_ld, {
'location': ({find_element(cls='content-location')}, {clean_html}), 'title': ('headline', {clean_html}),
}), 'creator': ('author', 'name', {clean_html}),
**traverse_obj(self._search_json_ld(webpage, video_id), {
'title': ('title', {clean_html}),
'description': ('description', {clean_html}), 'description': ('description', {clean_html}),
'release_timestamp': ('timestamp', {int_or_none}), 'release_timestamp': ('datePublished', {parse_iso8601}),
}), 'thumbnail': ('image', ..., {lambda x: self._proto_relative_url(x)}, {url_or_none}, any),
**traverse_obj(video_info, {
'duration': ('duration', {float_or_none}),
'timestamp': ('published_at', {parse_iso8601}),
}), }),
} }