mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2025-06-28 01:18:30 +00:00
Merge ce2fb298f3 into 73bf102116
This commit is contained in:
doe1080
GitHub
commit
6479552c6a
@ -1,75 +1,156 @@
|
|||||||
import re
|
import itertools
|
||||||
|
|
||||||
from .common import InfoExtractor
|
from .common import InfoExtractor
|
||||||
from ..utils import remove_start
|
from ..utils import (
|
||||||
|
ExtractorError,
|
||||||
|
clean_html,
|
||||||
|
extract_attributes,
|
||||||
|
float_or_none,
|
||||||
|
int_or_none,
|
||||||
|
mimetype2ext,
|
||||||
|
parse_iso8601,
|
||||||
|
str_or_none,
|
||||||
|
url_or_none,
|
||||||
|
)
|
||||||
|
from ..utils.traversal import find_element, traverse_obj
|
||||||
|
|
||||||
|
|
||||||
class TeleMBIE(InfoExtractor):
|
class TeleMBIE(InfoExtractor):
|
||||||
_WORKING = False
|
IE_NAME = 'telemb'
|
||||||
_VALID_URL = r'https?://(?:www\.)?telemb\.be/(?P<display_id>.+?)_d_(?P<id>\d+)\.html'
|
IE_DESC = 'Télé MB'
|
||||||
_TESTS = [
|
|
||||||
{
|
_VALID_URL = r'https?://(?:www\.)?telemb\.be(?:/replay)?/(?:actu|emission|sports)(?P<alt_id>(?:/[\w-]+)+)/(?P<id>\d+)'
|
||||||
'url': 'http://www.telemb.be/mons-cook-with-danielle-des-cours-de-cuisine-en-anglais-_d_13466.html',
|
_TESTS = [{
|
||||||
'md5': 'f45ea69878516ba039835794e0f8f783',
|
'url': 'https://www.telemb.be/actu/frameries-un-concours-pour-conducteurs-dengins-de-chantier/37879',
|
||||||
'info_dict': {
|
'info_dict': {
|
||||||
'id': '13466',
|
'id': '37879',
|
||||||
'display_id': 'mons-cook-with-danielle-des-cours-de-cuisine-en-anglais-',
|
|
||||||
'ext': 'mp4',
|
'ext': 'mp4',
|
||||||
'title': 'Mons - Cook with Danielle : des cours de cuisine en anglais ! - Les reportages',
|
'title': 'Frameries - Un concours pour conducteurs d\'engins de chantier',
|
||||||
'description': 'md5:bc5225f47b17c309761c856ad4776265',
|
'creators': ['Sabine Dupont'],
|
||||||
'thumbnail': r're:^http://.*\.(?:jpg|png)$',
|
'description': 'md5:1dc04a3aa56c5228503071baa8b4cc97',
|
||||||
|
'display_id': 'frameries-un-concours-pour-conducteurs-dengins-de-chantier',
|
||||||
|
'duration': 144.6,
|
||||||
|
'location': 'Frameries',
|
||||||
|
'release_date': '20250515',
|
||||||
|
'release_timestamp': 1747319520,
|
||||||
|
'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)',
|
||||||
},
|
},
|
||||||
},
|
}, {
|
||||||
{
|
'url': 'https://www.telemb.be/sports/karate-cinq-karatekas-du-bushikai-wasmuel-la-coupe-du-monde-tokyo/37849',
|
||||||
# non-ASCII characters in download URL
|
|
||||||
'url': 'http://telemb.be/les-reportages-havre-incendie-mortel_d_13514.html',
|
|
||||||
'md5': '6e9682736e5ccd4eab7f21e855350733',
|
|
||||||
'info_dict': {
|
'info_dict': {
|
||||||
'id': '13514',
|
'id': '37849',
|
||||||
'display_id': 'les-reportages-havre-incendie-mortel',
|
|
||||||
'ext': 'mp4',
|
'ext': 'mp4',
|
||||||
'title': 'Havré - Incendie mortel - Les reportages',
|
'title': 'Karaté : Cinq karatékas du Bushikai Wasmuel à la Coupe du Monde à Tokyo',
|
||||||
'description': 'md5:5e54cb449acb029c2b7734e2d946bd4a',
|
'creators': ['Jacob Hemptinne'],
|
||||||
'thumbnail': r're:^http://.*\.(?:jpg|png)$',
|
'description': 'md5:17f2d55a1533a69079cc21eadd14725f',
|
||||||
|
'display_id': 'karate-cinq-karatekas-du-bushikai-wasmuel-la-coupe-du-monde-tokyo',
|
||||||
|
'duration': 211.6,
|
||||||
|
'location': 'Quaregnon',
|
||||||
|
'release_date': '20250512',
|
||||||
|
'release_timestamp': 1747066800,
|
||||||
|
'thumbnail': r're:https?://www\.telemb\.be/cdn/.+\.(?:jpe?g|png)',
|
||||||
},
|
},
|
||||||
|
}, {
|
||||||
|
'url': 'https://www.telemb.be/replay/emission/les-infos/les-infos-16052025/36502',
|
||||||
|
'info_dict': {
|
||||||
|
'id': '36502',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': 'Les Infos - 16/05/2025',
|
||||||
|
'creators': ['Télé MB'],
|
||||||
|
'description': 'md5:dff75a3a51c769696c23454e932ff720',
|
||||||
|
'display_id': 'les-infos-16052025',
|
||||||
|
'duration': 1144.32,
|
||||||
|
'release_date': '20250516',
|
||||||
|
'release_timestamp': 1747412520,
|
||||||
|
'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)',
|
||||||
},
|
},
|
||||||
]
|
}, {
|
||||||
|
'url': 'https://www.telemb.be/actu/linvite-des-infos/le-cma-de-jemappes-fetera-ses-20-ans-ce-week-end/36711',
|
||||||
|
'info_dict': {
|
||||||
|
'id': '36711',
|
||||||
|
'ext': 'mp4',
|
||||||
|
'title': 'Le CMA de Jemappes fêtera ses 20 ans ce week-end',
|
||||||
|
'creators': ['Loélia Chais'],
|
||||||
|
'description': 'md5:f244efcce22f217df044d755e116ddef',
|
||||||
|
'display_id': 'le-cma-de-jemappes-fetera-ses-20-ans-ce-week-end',
|
||||||
|
'duration': 316.08,
|
||||||
|
'location': 'Mons',
|
||||||
|
'release_date': '20241128',
|
||||||
|
'release_timestamp': 1732787226,
|
||||||
|
'thumbnail': r're:https?://www\.telemb\.be/cdn.+\.(?:jpe?g|png)',
|
||||||
|
},
|
||||||
|
}]
|
||||||
|
|
||||||
def _real_extract(self, url):
|
def _real_extract(self, url):
|
||||||
mobj = self._match_valid_url(url)
|
video_id = self._match_id(url)
|
||||||
video_id = mobj.group('id')
|
display_id = self._match_valid_url(url).group('alt_id').split('/')[-1]
|
||||||
display_id = mobj.group('display_id')
|
webpage = self._download_webpage(url, video_id)
|
||||||
|
json_ld = next(itertools.islice(self._yield_json_ld(webpage, video_id), 1, 2), {})
|
||||||
|
|
||||||
webpage = self._download_webpage(url, display_id)
|
data_video_id = traverse_obj(webpage, (
|
||||||
|
{find_element(cls='freecaster-player', html=True)},
|
||||||
|
{extract_attributes}, 'data-video-id', {str_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={})):
|
||||||
|
raise ExtractorError('Failed to fetch video information')
|
||||||
|
|
||||||
formats = []
|
qualities = {
|
||||||
for video_url in re.findall(r'file\s*:\s*"([^"]+)"', webpage):
|
'3': (640, 360),
|
||||||
fmt = {
|
'5': (960, 540),
|
||||||
'url': video_url,
|
'9': (1280, 720),
|
||||||
'format_id': video_url.split(':')[0],
|
'11': (1920, 1080),
|
||||||
}
|
}
|
||||||
rtmp = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>.+))/(?P<playpath>mp4:.+)$', video_url)
|
formats = []
|
||||||
if rtmp:
|
for src in traverse_obj(video_info, ('src', lambda _, v: v['src'])):
|
||||||
fmt.update({
|
if not (src_url := url_or_none(src['src'])):
|
||||||
'play_path': rtmp.group('playpath'),
|
continue
|
||||||
'app': rtmp.group('app'),
|
ext = mimetype2ext(src.get('type'))
|
||||||
'player_url': 'http://p.jwpcdn.com/6/10/jwplayer.flash.swf',
|
|
||||||
'page_url': 'http://www.telemb.be',
|
|
||||||
'preference': -10,
|
|
||||||
})
|
|
||||||
formats.append(fmt)
|
|
||||||
|
|
||||||
title = remove_start(self._og_search_title(webpage), 'TéléMB : ')
|
if ext == 'mp4':
|
||||||
description = self._html_search_regex(
|
quality = src_url.rpartition('_')[2].removesuffix('.mp4')
|
||||||
r'<meta property="og:description" content="(.+?)" />',
|
width, height = qualities.get(quality, (None, None))
|
||||||
webpage, 'description', fatal=False)
|
formats.append({
|
||||||
thumbnail = self._og_search_thumbnail(webpage)
|
'acodec': 'mp4a.40.2',
|
||||||
|
'ext': ext,
|
||||||
|
'format_id': f'mp4-{quality}',
|
||||||
|
'height': height,
|
||||||
|
'url': src_url,
|
||||||
|
'vcodec': 'avc1',
|
||||||
|
'width': width,
|
||||||
|
})
|
||||||
|
continue
|
||||||
|
elif ext == 'm3u8':
|
||||||
|
fmts = self._extract_m3u8_formats(
|
||||||
|
src_url, video_id, 'mp4', m3u8_id='hls', fatal=False)
|
||||||
|
for fmt in fmts:
|
||||||
|
if fmt.get('vcodec') == 'none':
|
||||||
|
fmt.update({
|
||||||
|
'abr': int_or_none(self._search_regex(
|
||||||
|
r'-(\d+)-', fmt['format_id'], 'adaptive bit rate')),
|
||||||
|
'acodec': 'mp4a.40.2',
|
||||||
|
'ext': 'm4a',
|
||||||
|
})
|
||||||
|
elif ext == 'mpd':
|
||||||
|
fmts = self._extract_mpd_formats(
|
||||||
|
src_url, video_id, mpd_id='dash', fatal=False)
|
||||||
|
else:
|
||||||
|
self.report_warning(f'Unsupported stream type: {ext}')
|
||||||
|
continue
|
||||||
|
formats.extend(fmts)
|
||||||
|
|
||||||
return {
|
return {
|
||||||
'id': video_id,
|
'id': video_id,
|
||||||
'display_id': display_id,
|
'display_id': display_id,
|
||||||
'title': title,
|
'duration': traverse_obj(video_info, ('duration', {float_or_none})),
|
||||||
'description': description,
|
|
||||||
'thumbnail': thumbnail,
|
|
||||||
'formats': formats,
|
'formats': formats,
|
||||||
|
'location': traverse_obj(webpage, (
|
||||||
|
{find_element(cls='content-location')}, {clean_html})),
|
||||||
|
**traverse_obj(json_ld, {
|
||||||
|
'title': ('headline', {clean_html}),
|
||||||
|
'creator': ('author', 'name', {clean_html}),
|
||||||
|
'description': ('description', {clean_html}),
|
||||||
|
'release_timestamp': ('datePublished', {parse_iso8601}),
|
||||||
|
'thumbnail': ('image', ..., {lambda x: self._proto_relative_url(x)}, {url_or_none}, any),
|
||||||
|
}),
|
||||||
}
|
}
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user