archive
/
yt-dlp
mirror of https://github.com/yt-dlp/yt-dlp
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[extractor/europarl] Rewrite extractor (#7114)

Browse Source 
Authored by: HobbyistDev
Closes #6396
pull/7171/head
HobbyistDev 1 year ago committed by GitHub
parent dc3c44f349
commit 03789976d3
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 40 additions and 40 deletions
Show Stats Download Patch File Download Diff File

80
yt_dlp/extractor/europa.py
Unescape Escape View File

@ -6,6 +6,7 @@ from ..utils import (
parse_iso8601, parse_iso8601,
parse_qs, parse_qs,
qualities, qualities,
traverse_obj,
unified_strdate, unified_strdate,
xpath_text xpath_text
) )
@ -92,82 +93,81 @@ class EuropaIE(InfoExtractor):
class EuroParlWebstreamIE(InfoExtractor): class EuroParlWebstreamIE(InfoExtractor):
_VALID_URL = r'''(?x) _VALID_URL = r'''(?x)
https?://(?:multimedia|webstreaming)\.europarl\.europa\.eu/[^/#?]+/ https?://multimedia\.europarl\.europa\.eu/[^/#?]+/
(?:embed/embed\.html\?event=|(?!video)[^/#?]+/[\w-]+_)(?P<id>[\w-]+) (?:(?!video)[^/#?]+/[\w-]+_)(?P<id>[\w-]+)
''' '''
_TESTS = [{ _TESTS = [{
'url': 'https://multimedia.europarl.europa.eu/pl/webstreaming/plenary-session_20220914-0900-PLENARY', 'url': 'https://multimedia.europarl.europa.eu/pl/webstreaming/plenary-session_20220914-0900-PLENARY',
'info_dict': { 'info_dict': {
'id': 'bcaa1db4-76ef-7e06-8da7-839bd0ad1dbe', 'id': '62388b15-d85b-4add-99aa-ba12ccf64f0d',
'ext': 'mp4', 'ext': 'mp4',
'release_timestamp': 1663137900,
'title': 'Plenary session', 'title': 'Plenary session',
'release_timestamp': 1663139069,
'release_date': '20220914', 'release_date': '20220914',
}, },
'params': { 'params': {
'skip_download': True, 'skip_download': True,
} }
}, { }, {
'url': 'https://multimedia.europarl.europa.eu/pl/webstreaming/eu-cop27-un-climate-change-conference-in-sharm-el-sheikh-egypt-ep-delegation-meets-with-ngo-represen_20221114-1600-SPECIAL-OTHER', # live webstream
'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/euroscola_20221115-1000-SPECIAL-EUROSCOLA',
'info_dict': { 'info_dict': {
'id': 'a8428de8-b9cd-6a2e-11e4-3805d9c9ff5c',
'ext': 'mp4', 'ext': 'mp4',
'release_timestamp': 1668434400, 'id': '510eda7f-ba72-161b-7ee7-0e836cd2e715',
'release_date': '20221114', 'release_timestamp': 1668502800,
'title': 'md5:d3550280c33cc70e0678652e3d52c028', 'title': 'Euroscola 2022-11-15 19:21',
'release_date': '20221115',
'live_status': 'is_live',
}, },
'params': { 'skip': 'not live anymore'
'skip_download': True,
}
}, { }, {
# embed webpage 'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/committee-on-culture-and-education_20230301-1130-COMMITTEE-CULT',
'url': 'https://webstreaming.europarl.europa.eu/ep/embed/embed.html?event=20220914-0900-PLENARY&language=en&autoplay=true&logo=true',
'info_dict': { 'info_dict': {
'id': 'bcaa1db4-76ef-7e06-8da7-839bd0ad1dbe', 'id': '7355662c-8eac-445e-4bb9-08db14b0ddd7',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Plenary session', 'release_date': '20230301',
'release_date': '20220914', 'title': 'Committee on Culture and Education',
'release_timestamp': 1663137900, 'release_timestamp': 1677666641,
},
'params': {
'skip_download': True,
} }
}, { }, {
# live webstream # live stream
'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/euroscola_20221115-1000-SPECIAL-EUROSCOLA', 'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/committee-on-environment-public-health-and-food-safety_20230524-0900-COMMITTEE-ENVI',
'info_dict': { 'info_dict': {
'id': 'e4255f56-10aa-4b3c-6530-08db56d5b0d9',
'ext': 'mp4', 'ext': 'mp4',
'id': '510eda7f-ba72-161b-7ee7-0e836cd2e715', 'release_date': '20230524',
'release_timestamp': 1668502800, 'title': r're:Committee on Environment, Public Health and Food Safety \d{4}-\d{2}-\d{2}\s\d{2}:\d{2}',
'title': 'Euroscola 2022-11-15 19:21', 'release_timestamp': 1684911541,
'release_date': '20221115',
'live_status': 'is_live', 'live_status': 'is_live',
}, },
'skip': 'not live anymore' 'skip': 'Not live anymore'
}] }]
def _real_extract(self, url): def _real_extract(self, url):
display_id = self._match_id(url) display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
webpage_nextjs = self._search_nextjs_data(webpage, display_id)['props']['pageProps']
json_info = self._download_json( json_info = self._download_json(
'https://vis-api.vuplay.co.uk/event/external', display_id, 'https://acs-api.europarl.connectedviews.eu/api/FullMeeting', display_id,
query={ query={
'player_key': 'europarl|718f822c-a48c-4841-9947-c9cb9bb1743c', 'api-version': 1.0,
'external_id': display_id, 'tenantId': 'bae646ca-1fc8-4363-80ba-2c04f06b4968',
'externalReference': display_id
}) })
formats, subtitles = self._extract_mpd_formats_and_subtitles(json_info['streaming_url'], display_id) formats, subtitles = [], {}
fmts, subs = self._extract_m3u8_formats_and_subtitles( for hls_url in traverse_obj(json_info, ((('meetingVideo'), ('meetingVideos', ...)), 'hlsUrl')):
json_info['streaming_url'].replace('.mpd', '.m3u8'), display_id) fmt, subs = self._extract_m3u8_formats_and_subtitles(hls_url, display_id)
formats.extend(fmt)
formats.extend(fmts) self._merge_subtitles(subs, target=subtitles)
self._merge_subtitles(subs, target=subtitles)
return { return {
'id': json_info['id'], 'id': json_info['id'],
'title': json_info.get('title'), 'title': traverse_obj(webpage_nextjs, (('mediaItem', 'title'), ('title', )), get_all=False),
'formats': formats, 'formats': formats,
'subtitles': subtitles, 'subtitles': subtitles,
'release_timestamp': parse_iso8601(json_info.get('published_start')), 'release_timestamp': parse_iso8601(json_info.get('startDateTime')),
'is_live': 'LIVE' in json_info.get('state', '') 'is_live': traverse_obj(webpage_nextjs, ('mediaItem', 'mediaSubType')) == 'Live'
} }

Write Preview
Loading…
Cancel
Save