From 593e43c0306aeb70ff8849ce30079f18bb088f00 Mon Sep 17 00:00:00 2001 From: Ashish Gupta <39122144+Ashish0804@users.noreply.github.com> Date: Fri, 21 Jan 2022 11:32:31 +0530 Subject: [PATCH] [LnkIE] Add extractor (#2408) Closes: #2268 Authored by: Ashish0804 --- yt_dlp/extractor/extractors.py | 5 +- yt_dlp/extractor/lnkgo.py | 84 ++++++++++++++++++++++++++++++++++ 2 files changed, 88 insertions(+), 1 deletion(-) diff --git a/yt_dlp/extractor/extractors.py b/yt_dlp/extractor/extractors.py index d659d7a8e..4bab736e5 100644 --- a/yt_dlp/extractor/extractors.py +++ b/yt_dlp/extractor/extractors.py @@ -756,7 +756,10 @@ from .livestream import ( LivestreamOriginalIE, LivestreamShortenerIE, ) -from .lnkgo import LnkGoIE +from .lnkgo import ( + LnkGoIE, + LnkIE, +) from .localnews8 import LocalNews8IE from .lovehomeporn import LoveHomePornIE from .lrt import LRTIE diff --git a/yt_dlp/extractor/lnkgo.py b/yt_dlp/extractor/lnkgo.py index 14675968e..7da0b4284 100644 --- a/yt_dlp/extractor/lnkgo.py +++ b/yt_dlp/extractor/lnkgo.py @@ -6,8 +6,10 @@ from .common import InfoExtractor from ..utils import ( clean_html, compat_str, + format_field, int_or_none, parse_iso8601, + unified_strdate, ) @@ -85,3 +87,85 @@ class LnkGoIE(InfoExtractor): 'timestamp': parse_iso8601(video_info.get('airDate')), 'view_count': int_or_none(video_info.get('viewsCount')), } + + +class LnkIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?lnk\.lt/[^/]+/(?P\d+)' + + _TESTS = [{ + 'url': 'https://lnk.lt/zinios/79791', + 'info_dict': { + 'id': '79791', + 'ext': 'mp4', + 'title': 'LNK.lt: Viešintų gyventojai sukilo prieš radijo bangų siųstuvą', + 'description': 'Svarbiausios naujienos trumpai, LNK žinios ir Info dienos pokalbiai.', + 'view_count': int, + 'duration': 233, + 'upload_date': '20191123', + 'thumbnail': r're:^https?://.*\.jpg$', + 'episode_number': 13431, + 'series': 'Naujausi žinių reportažai', + 'episode': 'Episode 13431' + }, + 'params': {'skip_download': True} + }, { + 'url': 'https://lnk.lt/istorijos-trumpai/152546', + 'info_dict': { + 'id': '152546', + 'ext': 'mp4', + 'title': 'Radžio koncertas gaisre ', + 'description': 'md5:0666b5b85cb9fc7c1238dec96f71faba', + 'view_count': int, + 'duration': 54, + 'upload_date': '20220105', + 'thumbnail': r're:^https?://.*\.jpg$', + 'episode_number': 1036, + 'series': 'Istorijos trumpai', + 'episode': 'Episode 1036' + }, + 'params': {'skip_download': True} + }, { + 'url': 'https://lnk.lt/gyvunu-pasaulis/151549', + 'info_dict': { + 'id': '151549', + 'ext': 'mp4', + 'title': 'Gyvūnų pasaulis', + 'description': '', + 'view_count': int, + 'duration': 1264, + 'upload_date': '20220108', + 'thumbnail': r're:^https?://.*\.jpg$', + 'episode_number': 16, + 'series': 'Gyvūnų pasaulis', + 'episode': 'Episode 16' + }, + 'params': {'skip_download': True} + }] + + def _real_extract(self, url): + id = self._match_id(url) + video_json = self._download_json(f'https://lnk.lt/api/video/video-config/{id}', id)['videoInfo'] + formats, subtitles = [], {} + if video_json.get('videoUrl'): + fmts, subs = self._extract_m3u8_formats_and_subtitles(video_json['videoUrl'], id) + formats.extend(fmts) + subtitles = self._merge_subtitles(subtitles, subs) + if video_json.get('videoFairplayUrl') and not video_json.get('drm'): + fmts, subs = self._extract_m3u8_formats_and_subtitles(video_json['videoFairplayUrl'], id) + formats.extend(fmts) + subtitles = self._merge_subtitles(subtitles, subs) + + self._sort_formats(formats) + return { + 'id': id, + 'title': video_json.get('title'), + 'description': video_json.get('description'), + 'view_count': video_json.get('viewsCount'), + 'duration': video_json.get('duration'), + 'upload_date': unified_strdate(video_json.get('airDate')), + 'thumbnail': format_field(video_json, 'posterImage', 'https://lnk.lt/all-images/%s'), + 'episode_number': int_or_none(video_json.get('episodeNumber')), + 'series': video_json.get('programTitle'), + 'formats': formats, + 'subtitles': subtitles, + }