From 3783b5f1d13380f9472bcbdca192aff349c01b17 Mon Sep 17 00:00:00 2001 From: ajj8 <35781586+ajj8@users.noreply.github.com> Date: Thu, 28 Oct 2021 11:57:09 +0100 Subject: [PATCH] [itv] Add support for ITV News (#1456) Authored by: ajj8 --- yt_dlp/extractor/itv.py | 34 ++++++++++++++++++++-------------- 1 file changed, 20 insertions(+), 14 deletions(-) diff --git a/yt_dlp/extractor/itv.py b/yt_dlp/extractor/itv.py index d69782b782..6e6a3673cd 100644 --- a/yt_dlp/extractor/itv.py +++ b/yt_dlp/extractor/itv.py @@ -220,16 +220,23 @@ class ITVIE(InfoExtractor): class ITVBTCCIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?itv\.com/btcc/(?:[^/]+/)*(?P[^/?#&]+)' - _TEST = { + _VALID_URL = r'https?://(?:www\.)?itv\.com/(?:news|btcc)/(?:[^/]+/)*(?P[^/?#&]+)' + _TESTS = [{ 'url': 'https://www.itv.com/btcc/articles/btcc-2019-brands-hatch-gp-race-action', 'info_dict': { 'id': 'btcc-2019-brands-hatch-gp-race-action', 'title': 'BTCC 2019: Brands Hatch GP race action', }, 'playlist_count': 12, - } - BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1582188683001/HkiHLnNRx_default/index.html?videoId=%s' + }, { + 'url': 'https://www.itv.com/news/2021-10-27/i-have-to-protect-the-country-says-rishi-sunak-as-uk-faces-interest-rate-hike', + 'info_dict': { + 'id': 'i-have-to-protect-the-country-says-rishi-sunak-as-uk-faces-interest-rate-hike', + 'title': 'md5:6ef054dd9f069330db3dcc66cb772d32' + }, + 'playlist_count': 4 + }] + BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s' def _real_extract(self, url): playlist_id = self._match_id(url) @@ -240,15 +247,15 @@ class ITVBTCCIE(InfoExtractor): '(?s)]+id=[\'"]__NEXT_DATA__[^>]*>([^<]+)', webpage, 'json_map'), playlist_id), lambda x: x['props']['pageProps']['article']['body']['content']) or [] - # Discard empty objects - video_ids = [] + entries = [] for video in json_map: - if video['data'].get('id'): - video_ids.append(video['data']['id']) - - entries = [ - self.url_result( - smuggle_url(self.BRIGHTCOVE_URL_TEMPLATE % video_id, { + if not any(video['data'].get(attr) == 'Brightcove' for attr in ('name', 'type')): + continue + video_id = video['data']['id'] + account_id = video['data']['accountId'] + player_id = video['data']['playerId'] + entries.append(self.url_result( + smuggle_url(self.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, video_id), { # ITV does not like some GB IP ranges, so here are some # IP blocks it accepts 'geo_ip_blocks': [ @@ -256,8 +263,7 @@ class ITVBTCCIE(InfoExtractor): ], 'referrer': url, }), - ie=BrightcoveNewIE.ie_key(), video_id=video_id) - for video_id in video_ids] + ie=BrightcoveNewIE.ie_key(), video_id=video_id)) title = self._og_search_title(webpage, fatal=False)