yt_dlp/extractor/ctvnews.py

   1 import json
   2 import re
   3 import urllib.parse
   4
   5 from .common import InfoExtractor
   6 from .ninecninemedia import NineCNineMediaIE
   7 from ..utils import extract_attributes, orderedSet
   8 from ..utils.traversal import find_element, traverse_obj
   9
  10
  11 class CTVNewsIE(InfoExtractor):
  12     _BASE_REGEX = r'https?://(?:[^.]+\.)?ctvnews\.ca/'
  13     _VIDEO_ID_RE = r'(?P<id>\d{5,})'
  14     _PLAYLIST_ID_RE = r'(?P<id>\d\.\d{5,})'
  15     _VALID_URL = [
  16         rf'{_BASE_REGEX}video/c{_VIDEO_ID_RE}',
  17         rf'{_BASE_REGEX}video(?:-gallery)?/?\?clipId={_VIDEO_ID_RE}',
  18         rf'{_BASE_REGEX}video/?\?(?:playlist|bin)Id={_PLAYLIST_ID_RE}',
  19         rf'{_BASE_REGEX}(?!video/)[^?#]*?{_PLAYLIST_ID_RE}/?(?:$|[?#])',
  20         rf'{_BASE_REGEX}(?!video/)[^?#]+\?binId={_PLAYLIST_ID_RE}',
  21     ]
  22     _TESTS = [{
  23         'url': 'http://www.ctvnews.ca/video?clipId=901995',
  24         'md5': 'b608f466c7fa24b9666c6439d766ab7e',
  25         'info_dict': {
  26             'id': '901995',
  27             'ext': 'flv',
  28             'title': 'Extended: \'That person cannot be me\' Johnson says',
  29             'description': 'md5:958dd3b4f5bbbf0ed4d045c790d89285',
  30             'timestamp': 1467286284,
  31             'upload_date': '20160630',
  32             'categories': [],
  33             'season_number': 0,
  34             'season': 'Season 0',
  35             'tags': [],
  36             'series': 'CTV News National | Archive | Stories 2',
  37             'season_id': '57981',
  38             'thumbnail': r're:https?://.*\.jpg$',
  39             'duration': 764.631,
  40         },
  41     }, {
  42         'url': 'https://barrie.ctvnews.ca/video/c3030933-here_s-what_s-making-news-for-nov--15?binId=1272429',
  43         'md5': '8b8c2b33c5c1803e3c26bc74ff8694d5',
  44         'info_dict': {
  45             'id': '3030933',
  46             'ext': 'flv',
  47             'title': 'Here’s what’s making news for Nov. 15',
  48             'description': 'Here are the top stories we’re working on for CTV News at 11 for Nov. 15',
  49             'thumbnail': 'http://images2.9c9media.com/image_asset/2021_2_22_a602e68e-1514-410e-a67a-e1f7cccbacab_png_2000x1125.jpg',
  50             'season_id': '58104',
  51             'season_number': 0,
  52             'tags': [],
  53             'season': 'Season 0',
  54             'categories': [],
  55             'series': 'CTV News Barrie',
  56             'upload_date': '20241116',
  57             'duration': 42.943,
  58             'timestamp': 1731722452,
  59         },
  60     }, {
  61         'url': 'http://www.ctvnews.ca/video?playlistId=1.2966224',
  62         'info_dict':
  63         {
  64             'id': '1.2966224',
  65         },
  66         'playlist_mincount': 19,
  67     }, {
  68         'url': 'http://www.ctvnews.ca/video?binId=1.2876780',
  69         'info_dict':
  70         {
  71             'id': '1.2876780',
  72         },
  73         'playlist_mincount': 100,
  74     }, {
  75         'url': 'https://www.ctvnews.ca/it-s-been-23-years-since-toronto-called-in-the-army-after-a-major-snowstorm-1.5736957',
  76         'info_dict':
  77         {
  78             'id': '1.5736957',
  79         },
  80         'playlist_mincount': 6,
  81     }, {
  82         'url': 'https://www.ctvnews.ca/business/respondents-to-bank-of-canada-questionnaire-largely-oppose-creating-a-digital-loonie-1.6665797',
  83         'md5': '24bc4b88cdc17d8c3fc01dfc228ab72c',
  84         'info_dict': {
  85             'id': '2695026',
  86             'ext': 'flv',
  87             'season_id': '89852',
  88             'series': 'From CTV News Channel',
  89             'description': 'md5:796a985a23cacc7e1e2fafefd94afd0a',
  90             'season': '2023',
  91             'title': 'Bank of Canada asks public about digital currency',
  92             'categories': [],
  93             'tags': [],
  94             'upload_date': '20230526',
  95             'season_number': 2023,
  96             'thumbnail': 'http://images2.9c9media.com/image_asset/2019_3_28_35f5afc3-10f6-4d92-b194-8b9a86f55c6a_png_1920x1080.jpg',
  97             'timestamp': 1685105157,
  98             'duration': 253.553,
  99         },
 100     }, {
 101         'url': 'https://stox.ctvnews.ca/video-gallery?clipId=582589',
 102         'md5': '135cc592df607d29dddc931f1b756ae2',
 103         'info_dict': {
 104             'id': '582589',
 105             'ext': 'flv',
 106             'categories': [],
 107             'timestamp': 1427906183,
 108             'season_number': 0,
 109             'duration': 125.559,
 110             'thumbnail': 'http://images2.9c9media.com/image_asset/2019_3_28_35f5afc3-10f6-4d92-b194-8b9a86f55c6a_png_1920x1080.jpg',
 111             'series': 'CTV News Stox',
 112             'description': 'CTV original footage of the rise and fall of the Berlin Wall.',
 113             'title': 'Berlin Wall',
 114             'season_id': '63817',
 115             'season': 'Season 0',
 116             'tags': [],
 117             'upload_date': '20150401',
 118         },
 119     }, {
 120         'url': 'https://ottawa.ctvnews.ca/features/regional-contact/regional-contact-archive?binId=1.1164587#3023759',
 121         'md5': 'a14c0603557decc6531260791c23cc5e',
 122         'info_dict': {
 123             'id': '3023759',
 124             'ext': 'flv',
 125             'season_number': 2024,
 126             'timestamp': 1731798000,
 127             'season': '2024',
 128             'episode': 'Episode 125',
 129             'description': 'CTV News Ottawa at Six',
 130             'duration': 2712.076,
 131             'episode_number': 125,
 132             'upload_date': '20241116',
 133             'title': 'CTV News Ottawa at Six for Saturday, November 16, 2024',
 134             'thumbnail': 'http://images2.9c9media.com/image_asset/2019_3_28_35f5afc3-10f6-4d92-b194-8b9a86f55c6a_png_1920x1080.jpg',
 135             'categories': [],
 136             'tags': [],
 137             'series': 'CTV News Ottawa at Six',
 138             'season_id': '92667',
 139         },
 140     }, {
 141         'url': 'http://www.ctvnews.ca/1.810401',
 142         'only_matching': True,
 143     }, {
 144         'url': 'http://www.ctvnews.ca/canadiens-send-p-k-subban-to-nashville-in-blockbuster-trade-1.2967231',
 145         'only_matching': True,
 146     }, {
 147         'url': 'http://vancouverisland.ctvnews.ca/video?clipId=761241',
 148         'only_matching': True,
 149     }]
 150
 151     def _ninecninemedia_url_result(self, clip_id):
 152         return self.url_result(f'9c9media:ctvnews_web:{clip_id}', NineCNineMediaIE, clip_id)
 153
 154     def _real_extract(self, url):
 155         page_id = self._match_id(url)
 156
 157         if mobj := re.fullmatch(self._VIDEO_ID_RE, urllib.parse.urlparse(url).fragment):
 158             page_id = mobj.group('id')
 159
 160         if re.fullmatch(self._VIDEO_ID_RE, page_id):
 161             return self._ninecninemedia_url_result(page_id)
 162
 163         webpage = self._download_webpage(f'https://www.ctvnews.ca/{page_id}', page_id, query={
 164             'ot': 'example.AjaxPageLayout.ot',
 165             'maxItemsPerPage': 1000000,
 166         })
 167         entries = [self._ninecninemedia_url_result(clip_id)
 168                    for clip_id in orderedSet(re.findall(r'clip\.id\s*=\s*(\d+);', webpage))]
 169         if not entries:
 170             webpage = self._download_webpage(url, page_id)
 171             if 'getAuthStates("' in webpage:
 172                 entries = [self._ninecninemedia_url_result(clip_id) for clip_id in
 173                            self._search_regex(r'getAuthStates\("([\d+,]+)"', webpage, 'clip ids').split(',')]
 174             else:
 175                 entries = [
 176                     self._ninecninemedia_url_result(clip_id) for clip_id in
 177                     traverse_obj(webpage, (
 178                         {find_element(tag='jasper-player-container', html=True)},
 179                         {extract_attributes}, 'axis-ids', {json.loads}, ..., 'axisId', {str}))
 180                 ]
 181
 182         return self.playlist_result(entries, page_id)