From 4c7d7215cd069bd15239d5a43ecb764e60a49ce7 Mon Sep 17 00:00:00 2001 From: Remita Amine Date: Sat, 26 Dec 2020 09:37:40 +0100 Subject: [PATCH] [theweatherchannel] fix extraction (closes #25930)(closes #26051) --- youtube_dl/extractor/theweatherchannel.py | 43 +++++++++++++++++------ 1 file changed, 33 insertions(+), 10 deletions(-) diff --git a/youtube_dl/extractor/theweatherchannel.py b/youtube_dl/extractor/theweatherchannel.py index c34a49d03..b2a8c3797 100644 --- a/youtube_dl/extractor/theweatherchannel.py +++ b/youtube_dl/extractor/theweatherchannel.py @@ -1,18 +1,22 @@ # coding: utf-8 from __future__ import unicode_literals +import json +import re + from .theplatform import ThePlatformIE from ..utils import ( determine_ext, parse_duration, + parse_iso8601, ) class TheWeatherChannelIE(ThePlatformIE): - _VALID_URL = r'https?://(?:www\.)?weather\.com/(?:[^/]+/)*video/(?P[^/?#]+)' + _VALID_URL = r'https?://(?:www\.)?weather\.com(?P(?:/(?P[a-z]{2}-[A-Z]{2}))?/(?:[^/]+/)*video/(?P[^/?#]+))' _TESTS = [{ 'url': 'https://weather.com/series/great-outdoors/video/ice-climber-is-in-for-a-shock', - 'md5': 'ab924ac9574e79689c24c6b95e957def', + 'md5': 'c4cbe74c9c17c5676b704b950b73dd92', 'info_dict': { 'id': 'cc82397e-cc3f-4d11-9390-a785add090e8', 'ext': 'mp4', @@ -20,18 +24,33 @@ class TheWeatherChannelIE(ThePlatformIE): 'description': 'md5:55606ce1378d4c72e6545e160c9d9695', 'uploader': 'TWC - Digital (No Distro)', 'uploader_id': '6ccd5455-16bb-46f2-9c57-ff858bb9f62c', + 'upload_date': '20160720', + 'timestamp': 1469018835, } + }, { + 'url': 'https://weather.com/en-CA/international/videos/video/unidentified-object-falls-from-sky-in-india', + 'only_matching': True, }] def _real_extract(self, url): - display_id = self._match_id(url) - webpage = self._download_webpage(url, display_id) - drupal_settings = self._parse_json(self._search_regex( - r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);', - webpage, 'drupal settings'), display_id) - video_id = drupal_settings['twc']['contexts']['node']['uuid'] - video_data = self._download_json( - 'https://dsx.weather.com/cms/v4/asset-collection/en_US/' + video_id, video_id) + asset_name, locale, display_id = re.match(self._VALID_URL, url).groups() + if not locale: + locale = 'en-US' + video_data = list(self._download_json( + 'https://weather.com/api/v1/p/redux-dal', display_id, data=json.dumps([{ + 'name': 'getCMSAssetsUrlConfig', + 'params': { + 'language': locale.replace('-', '_'), + 'query': { + 'assetName': { + '$in': asset_name, + }, + }, + } + }]).encode(), headers={ + 'Content-Type': 'application/json', + })['dal']['getCMSAssetsUrlConfig'].values())[0]['data'][0] + video_id = video_data['id'] seo_meta = video_data.get('seometa', {}) title = video_data.get('title') or seo_meta['title'] @@ -66,6 +85,8 @@ class TheWeatherChannelIE(ThePlatformIE): }) self._sort_formats(formats) + cc_url = video_data.get('cc_url') + return { 'id': video_id, 'display_id': display_id, @@ -74,6 +95,8 @@ class TheWeatherChannelIE(ThePlatformIE): 'duration': parse_duration(video_data.get('duration')), 'uploader': video_data.get('providername'), 'uploader_id': video_data.get('providerid'), + 'timestamp': parse_iso8601(video_data.get('publishdate')), + 'subtitles': {locale[:2]: [{'url': cc_url}]} if cc_url else None, 'thumbnails': thumbnails, 'formats': formats, }