yt_dlp/extractor/nytimes.py

   1 import json
   2 import uuid
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     ExtractorError,
   7     clean_html,
   8     determine_ext,
   9     extract_attributes,
  10     float_or_none,
  11     get_elements_html_by_class,
  12     int_or_none,
  13     merge_dicts,
  14     mimetype2ext,
  15     parse_iso8601,
  16     remove_end,
  17     remove_start,
  18     str_or_none,
  19     traverse_obj,
  20     url_or_none,
  21 )
  22
  23
  24 class NYTimesBaseIE(InfoExtractor):
  25     _DNS_NAMESPACE = uuid.UUID('36dd619a-56dc-595b-9e09-37f4152c7b5d')
  26     _TOKEN = 'MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAuNIzKBOFB77aT/jN/FQ+/QVKWq5V1ka1AYmCR9hstz1pGNPH5ajOU9gAqta0T89iPnhjwla+3oec/Z3kGjxbpv6miQXufHFq3u2RC6HyU458cLat5kVPSOQCe3VVB5NRpOlRuwKHqn0txfxnwSSj8mqzstR997d3gKB//RO9zE16y3PoWlDQXkASngNJEWvL19iob/xwAkfEWCjyRILWFY0JYX3AvLMSbq7wsqOCE5srJpo7rRU32zsByhsp1D5W9OYqqwDmflsgCEQy2vqTsJjrJohuNg+urMXNNZ7Y3naMoqttsGDrWVxtPBafKMI8pM2ReNZBbGQsQXRzQNo7+QIDAQAB'
  27     _GRAPHQL_API = 'https://samizdat-graphql.nytimes.com/graphql/v2'
  28     _GRAPHQL_QUERY = '''query VideoQuery($id: String!) {
  29   video(id: $id) {
  30     ... on Video {
  31       bylines {
  32         renderedRepresentation
  33       }
  34       duration
  35       firstPublished
  36       promotionalHeadline
  37       promotionalMedia {
  38         ... on Image {
  39           crops {
  40             name
  41             renditions {
  42               name
  43               width
  44               height
  45               url
  46             }
  47           }
  48         }
  49       }
  50       renditions {
  51         type
  52         width
  53         height
  54         url
  55         bitrate
  56       }
  57       summary
  58     }
  59   }
  60 }'''
  61
  62     def _call_api(self, media_id):
  63         # reference: `id-to-uri.js`
  64         video_uuid = uuid.uuid5(self._DNS_NAMESPACE, 'video')
  65         media_uuid = uuid.uuid5(video_uuid, media_id)
  66
  67         return traverse_obj(self._download_json(
  68             self._GRAPHQL_API, media_id, 'Downloading JSON from GraphQL API', data=json.dumps({
  69                 'query': self._GRAPHQL_QUERY,
  70                 'variables': {'id': f'nyt://video/{media_uuid}'},
  71             }, separators=(',', ':')).encode(), headers={
  72                 'Content-Type': 'application/json',
  73                 'Nyt-App-Type': 'vhs',
  74                 'Nyt-App-Version': 'v3.52.21',
  75                 'Nyt-Token': self._TOKEN,
  76                 'Origin': 'https://nytimes.com',
  77             }, fatal=False), ('data', 'video', {dict})) or {}
  78
  79     def _extract_thumbnails(self, thumbs):
  80         return traverse_obj(thumbs, (lambda _, v: url_or_none(v['url']), {
  81             'url': 'url',
  82             'width': ('width', {int_or_none}),
  83             'height': ('height', {int_or_none}),
  84         }), default=None)
  85
  86     def _extract_formats_and_subtitles(self, video_id, content_media_json):
  87         urls = []
  88         formats = []
  89         subtitles = {}
  90         for video in traverse_obj(content_media_json, ('renditions', ..., {dict})):
  91             video_url = video.get('url')
  92             format_id = video.get('type')
  93             if not video_url or format_id == 'thumbs' or video_url in urls:
  94                 continue
  95             urls.append(video_url)
  96             ext = mimetype2ext(video.get('mimetype')) or determine_ext(video_url)
  97             if ext == 'm3u8':
  98                 m3u8_fmts, m3u8_subs = self._extract_m3u8_formats_and_subtitles(
  99                     video_url, video_id, 'mp4', 'm3u8_native',
 100                     m3u8_id=format_id or 'hls', fatal=False)
 101                 formats.extend(m3u8_fmts)
 102                 self._merge_subtitles(m3u8_subs, target=subtitles)
 103             elif ext == 'mpd':
 104                 continue  # all mpd urls give 404 errors
 105             else:
 106                 formats.append({
 107                     'url': video_url,
 108                     'format_id': format_id,
 109                     'vcodec': video.get('videoencoding') or video.get('video_codec'),
 110                     'width': int_or_none(video.get('width')),
 111                     'height': int_or_none(video.get('height')),
 112                     'filesize': traverse_obj(video, (
 113                         ('file_size', 'fileSize'), (None, ('value')), {int_or_none}), get_all=False),
 114                     'tbr': int_or_none(video.get('bitrate'), 1000) or None,
 115                     'ext': ext,
 116                 })
 117
 118         return formats, subtitles
 119
 120     def _extract_video(self, media_id):
 121         data = self._call_api(media_id)
 122         formats, subtitles = self._extract_formats_and_subtitles(media_id, data)
 123
 124         return {
 125             'id': media_id,
 126             'title': data.get('promotionalHeadline'),
 127             'description': data.get('summary'),
 128             'timestamp': parse_iso8601(data.get('firstPublished')),
 129             'duration': float_or_none(data.get('duration'), scale=1000),
 130             'creator': ', '.join(traverse_obj(data, (  # TODO: change to 'creators'
 131                 'bylines', ..., 'renderedRepresentation', {lambda x: remove_start(x, 'By ')}))),
 132             'formats': formats,
 133             'subtitles': subtitles,
 134             'thumbnails': self._extract_thumbnails(
 135                 traverse_obj(data, ('promotionalMedia', 'crops', ..., 'renditions', ...))),
 136         }
 137
 138
 139 class NYTimesIE(NYTimesBaseIE):
 140     _VALID_URL = r'https?://(?:(?:www\.)?nytimes\.com/video/(?:[^/]+/)+?|graphics8\.nytimes\.com/bcvideo/\d+(?:\.\d+)?/iframe/embed\.html\?videoId=)(?P<id>\d+)'
 141     _EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//graphics8\.nytimes\.com/bcvideo/[^/]+/iframe/embed\.html.+?)\1>']
 142     _TESTS = [{
 143         'url': 'http://www.nytimes.com/video/opinion/100000002847155/verbatim-what-is-a-photocopier.html?playlistId=100000001150263',
 144         'md5': 'a553aa344014e3723d33893d89d4defc',
 145         'info_dict': {
 146             'id': '100000002847155',
 147             'ext': 'mp4',
 148             'title': 'Verbatim: What Is a Photocopier?',
 149             'description': 'md5:93603dada88ddbda9395632fdc5da260',
 150             'timestamp': 1398646132,
 151             'upload_date': '20140428',
 152             'creator': 'Brett Weiner',
 153             'thumbnail': r're:https?://\w+\.nyt.com/images/.+\.jpg',
 154             'duration': 419,
 155         },
 156     }, {
 157         'url': 'http://www.nytimes.com/video/travel/100000003550828/36-hours-in-dubai.html',
 158         'only_matching': True,
 159     }]
 160
 161     def _real_extract(self, url):
 162         video_id = self._match_id(url)
 163
 164         return self._extract_video(video_id)
 165
 166
 167 class NYTimesArticleIE(NYTimesBaseIE):
 168     _VALID_URL = r'https?://(?:www\.)?nytimes\.com/\d{4}/\d{2}/\d{2}/(?!books|podcasts)[^/?#]+/(?:\w+/)?(?P<id>[^./?#]+)(?:\.html)?'
 169     _TESTS = [{
 170         'url': 'http://www.nytimes.com/2015/04/14/business/owner-of-gravity-payments-a-credit-card-processor-is-setting-a-new-minimum-wage-70000-a-year.html?_r=0',
 171         'md5': '3eb5ddb1d6f86254fe4f233826778737',
 172         'info_dict': {
 173             'id': '100000003628438',
 174             'ext': 'mp4',
 175             'title': 'One Company’s New Minimum Wage: $70,000 a Year',
 176             'description': 'md5:89ba9ab67ca767bb92bf823d1f138433',
 177             'timestamp': 1429047468,
 178             'upload_date': '20150414',
 179             'uploader': 'Matthew Williams',
 180             'creator': 'Patricia Cohen',
 181             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 182             'duration': 119.0,
 183         },
 184     }, {
 185         # article with audio and no video
 186         'url': 'https://www.nytimes.com/2023/09/29/health/mosquitoes-genetic-engineering.html',
 187         'md5': '2365b3555c8aa7f4dd34ca735ad02e6a',
 188         'info_dict': {
 189             'id': '100000009110381',
 190             'ext': 'mp3',
 191             'title': 'The Gamble: Can Genetically Modified Mosquitoes End Disease?',
 192             'description': 'md5:9ff8b47acbaf7f3ca8c732f5c815be2e',
 193             'timestamp': 1695960700,
 194             'upload_date': '20230929',
 195             'creator': 'Stephanie Nolen, Natalija Gormalova',
 196             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 197             'duration': 1322,
 198         },
 199     }, {
 200         'url': 'https://www.nytimes.com/2023/11/29/business/dealbook/kamala-harris-biden-voters.html',
 201         'md5': '3eb5ddb1d6f86254fe4f233826778737',
 202         'info_dict': {
 203             'id': '100000009202270',
 204             'ext': 'mp4',
 205             'title': 'Kamala Harris Defends Biden Policies, but Says ‘More Work’ Needed to Reach Voters',
 206             'description': 'md5:de4212a7e19bb89e4fb14210ca915f1f',
 207             'timestamp': 1701290997,
 208             'upload_date': '20231129',
 209             'uploader': 'By The New York Times',
 210             'creator': 'Katie Rogers',
 211             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 212             'duration': 97.631,
 213         },
 214         'params': {
 215             'skip_download': 'm3u8',
 216         },
 217     }, {
 218         # multiple videos in the same article
 219         'url': 'https://www.nytimes.com/2023/12/02/business/air-traffic-controllers-safety.html',
 220         'info_dict': {
 221             'id': 'air-traffic-controllers-safety',
 222             'title': 'Drunk and Asleep on the Job: Air Traffic Controllers Pushed to the Brink',
 223             'description': 'md5:549e5a5e935bf7d048be53ba3d2c863d',
 224             'upload_date': '20231202',
 225             'creator': 'Emily Steel, Sydney Ember',
 226             'timestamp': 1701511264,
 227         },
 228         'playlist_count': 3,
 229     }, {
 230         'url': 'https://www.nytimes.com/2023/12/02/business/media/netflix-squid-game-challenge.html',
 231         'only_matching': True,
 232     }]
 233
 234     def _extract_content_from_block(self, block):
 235         details = traverse_obj(block, {
 236             'id': ('sourceId', {str}),
 237             'uploader': ('bylines', ..., 'renderedRepresentation', {str}),
 238             'duration': (None, (('duration', {float_or_none(scale=1000)}), ('length', {int_or_none}))),
 239             'timestamp': ('firstPublished', {parse_iso8601}),
 240             'series': ('podcastSeries', {str}),
 241         }, get_all=False)
 242
 243         formats, subtitles = self._extract_formats_and_subtitles(details.get('id'), block)
 244         # audio articles will have an url and no formats
 245         url = traverse_obj(block, ('fileUrl', {url_or_none}))
 246         if not formats and url:
 247             formats.append({'url': url, 'vcodec': 'none'})
 248
 249         return {
 250             **details,
 251             'thumbnails': self._extract_thumbnails(traverse_obj(
 252                 block, ('promotionalMedia', 'crops', ..., 'renditions', ...))),
 253             'formats': formats,
 254             'subtitles': subtitles,
 255         }
 256
 257     def _real_extract(self, url):
 258         page_id = self._match_id(url)
 259         webpage = self._download_webpage(url, page_id)
 260         art_json = self._search_json(
 261             r'window\.__preloadedData\s*=', webpage, 'media details', page_id,
 262             transform_source=lambda x: x.replace('undefined', 'null'))['initialData']['data']['article']
 263
 264         blocks = traverse_obj(art_json, (
 265             'sprinkledBody', 'content', ..., ('ledeMedia', None),
 266             lambda _, v: v['__typename'] in ('Video', 'Audio')))
 267         if not blocks:
 268             raise ExtractorError('Unable to extract any media blocks from webpage')
 269
 270         common_info = {
 271             'title': remove_end(self._html_extract_title(webpage), ' - The New York Times'),
 272             'description': traverse_obj(art_json, (
 273                 'sprinkledBody', 'content', ..., 'summary', 'content', ..., 'text', {str}),
 274                 get_all=False) or self._html_search_meta(['og:description', 'twitter:description'], webpage),
 275             'timestamp': traverse_obj(art_json, ('firstPublished', {parse_iso8601})),
 276             'creator': ', '.join(
 277                 traverse_obj(art_json, ('bylines', ..., 'creators', ..., 'displayName'))),  # TODO: change to 'creators' (list)
 278             'thumbnails': self._extract_thumbnails(traverse_obj(
 279                 art_json, ('promotionalMedia', 'assetCrops', ..., 'renditions', ...))),
 280         }
 281
 282         entries = []
 283         for block in blocks:
 284             entries.append(merge_dicts(self._extract_content_from_block(block), common_info))
 285
 286         if len(entries) > 1:
 287             return self.playlist_result(entries, page_id, **common_info)
 288
 289         return {
 290             'id': page_id,
 291             **entries[0],
 292         }
 293
 294
 295 class NYTimesCookingIE(NYTimesBaseIE):
 296     IE_NAME = 'NYTimesCookingGuide'
 297     _VALID_URL = r'https?://cooking\.nytimes\.com/guides/(?P<id>[\w-]+)'
 298     _TESTS = [{
 299         'url': 'https://cooking.nytimes.com/guides/13-how-to-cook-a-turkey',
 300         'info_dict': {
 301             'id': '13-how-to-cook-a-turkey',
 302             'title': 'How to Cook a Turkey',
 303             'description': 'md5:726cfd3f9b161bdf5c279879e8050ca0',
 304         },
 305         'playlist_count': 2,
 306     }, {
 307         # single video example
 308         'url': 'https://cooking.nytimes.com/guides/50-how-to-make-mac-and-cheese',
 309         'md5': '64415805fe0b8640fce6b0b9def5989a',
 310         'info_dict': {
 311             'id': '100000005835845',
 312             'ext': 'mp4',
 313             'title': 'How to Make Mac and Cheese',
 314             'description': 'md5:b8f2f33ec1fb7523b21367147c9594f1',
 315             'timestamp': 1522950315,
 316             'upload_date': '20180405',
 317             'duration': 9.51,
 318             'creator': 'Alison Roman',
 319             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 320         },
 321     }, {
 322         'url': 'https://cooking.nytimes.com/guides/20-how-to-frost-a-cake',
 323         'md5': '64415805fe0b8640fce6b0b9def5989a',
 324         'info_dict': {
 325             'id': '20-how-to-frost-a-cake',
 326             'title': 'How to Frost a Cake',
 327             'description': 'md5:a31fe3b98a8ce7b98aae097730c269cd',
 328         },
 329         'playlist_count': 8,
 330     }]
 331
 332     def _real_extract(self, url):
 333         page_id = self._match_id(url)
 334         webpage = self._download_webpage(url, page_id)
 335         title = self._html_search_meta(['og:title', 'twitter:title'], webpage)
 336         description = self._html_search_meta(['og:description', 'twitter:description'], webpage)
 337
 338         lead_video_id = self._search_regex(
 339             r'data-video-player-id="(\d+)"></div>', webpage, 'lead video')
 340         media_ids = traverse_obj(
 341             get_elements_html_by_class('video-item', webpage), (..., {extract_attributes}, 'data-video-id'))
 342
 343         if media_ids:
 344             media_ids.append(lead_video_id)
 345             return self.playlist_result(
 346                 [self._extract_video(media_id) for media_id in media_ids], page_id, title, description)
 347
 348         return {
 349             **self._extract_video(lead_video_id),
 350             'title': title,
 351             'description': description,
 352             'creator': self._search_regex(  # TODO: change to 'creators'
 353                 r'<span itemprop="author">([^<]+)</span></p>', webpage, 'author', default=None),
 354         }
 355
 356
 357 class NYTimesCookingRecipeIE(InfoExtractor):
 358     _VALID_URL = r'https?://cooking\.nytimes\.com/recipes/(?P<id>\d+)'
 359     _TESTS = [{
 360         'url': 'https://cooking.nytimes.com/recipes/1017817-cranberry-curd-tart',
 361         'md5': '579e83bbe8e61e9de67f80edba8a78a8',
 362         'info_dict': {
 363             'id': '1017817',
 364             'ext': 'mp4',
 365             'title': 'Cranberry Curd Tart',
 366             'description': 'md5:ad77a3fc321db636256d4343c5742152',
 367             'timestamp': 1447804800,
 368             'upload_date': '20151118',
 369             'creator': 'David Tanis',
 370             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 371         },
 372     }, {
 373         'url': 'https://cooking.nytimes.com/recipes/1024781-neapolitan-checkerboard-cookies',
 374         'md5': '58df35998241dcf0620e99e646331b42',
 375         'info_dict': {
 376             'id': '1024781',
 377             'ext': 'mp4',
 378             'title': 'Neapolitan Checkerboard Cookies',
 379             'description': 'md5:ba12394c585ababea951cb6d2fcc6631',
 380             'timestamp': 1701302400,
 381             'upload_date': '20231130',
 382             'creator': 'Sue Li',
 383             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 384         },
 385     }, {
 386         'url': 'https://cooking.nytimes.com/recipes/1019516-overnight-oats',
 387         'md5': '2fe7965a3adc899913b8e25ada360823',
 388         'info_dict': {
 389             'id': '1019516',
 390             'ext': 'mp4',
 391             'timestamp': 1546387200,
 392             'description': 'md5:8856ce10239161bd2596ac335b9f9bfb',
 393             'upload_date': '20190102',
 394             'title': 'Overnight Oats',
 395             'creator': 'Genevieve Ko',
 396             'thumbnail': r're:https?://\w+\.nyt.com/images/.*\.jpg',
 397         },
 398     }]
 399
 400     def _real_extract(self, url):
 401         page_id = self._match_id(url)
 402         webpage = self._download_webpage(url, page_id)
 403         recipe_data = self._search_nextjs_data(webpage, page_id)['props']['pageProps']['recipe']
 404
 405         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 406             recipe_data['videoSrc'], page_id, 'mp4', m3u8_id='hls')
 407
 408         return {
 409             **traverse_obj(recipe_data, {
 410                 'id': ('id', {str_or_none}),
 411                 'title': ('title', {str}),
 412                 'description': ('topnote', {clean_html}),
 413                 'timestamp': ('publishedAt', {int_or_none}),
 414                 'creator': ('contentAttribution', 'cardByline', {str}),
 415             }),
 416             'formats': formats,
 417             'subtitles': subtitles,
 418             'thumbnails': [{'url': thumb_url} for thumb_url in traverse_obj(
 419                 recipe_data, ('image', 'crops', 'recipe', ..., {url_or_none}))],
 420         }