youtube_dl/extractor/nytimes.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import hmac
   5 import hashlib
   6 import base64
   7
   8 from .common import InfoExtractor
   9 from ..utils import (
  10     determine_ext,
  11     float_or_none,
  12     int_or_none,
  13     js_to_json,
  14     mimetype2ext,
  15     parse_iso8601,
  16     remove_start,
  17 )
  18
  19
  20 class NYTimesBaseIE(InfoExtractor):
  21     _SECRET = b'pX(2MbU2);4N{7J8)>YwKRJ+/pQ3JkiU2Q^V>mFYv6g6gYvt6v'
  22
  23     def _extract_video_from_id(self, video_id):
  24         # Authorization generation algorithm is reverse engineered from `signer` in
  25         # http://graphics8.nytimes.com/video/vhs/vhs-2.x.min.js
  26         path = '/svc/video/api/v3/video/' + video_id
  27         hm = hmac.new(self._SECRET, (path + ':vhs').encode(), hashlib.sha512).hexdigest()
  28         video_data = self._download_json('http://www.nytimes.com' + path, video_id, 'Downloading video JSON', headers={
  29             'Authorization': 'NYTV ' + base64.b64encode(hm.encode()).decode(),
  30             'X-NYTV': 'vhs',
  31         }, fatal=False)
  32         if not video_data:
  33             video_data = self._download_json(
  34                 'http://www.nytimes.com/svc/video/api/v2/video/' + video_id,
  35                 video_id, 'Downloading video JSON')
  36
  37         title = video_data['headline']
  38
  39         def get_file_size(file_size):
  40             if isinstance(file_size, int):
  41                 return file_size
  42             elif isinstance(file_size, dict):
  43                 return int(file_size.get('value', 0))
  44             else:
  45                 return None
  46
  47         urls = []
  48         formats = []
  49         for video in video_data.get('renditions', []):
  50             video_url = video.get('url')
  51             format_id = video.get('type')
  52             if not video_url or format_id == 'thumbs' or video_url in urls:
  53                 continue
  54             urls.append(video_url)
  55             ext = mimetype2ext(video.get('mimetype')) or determine_ext(video_url)
  56             if ext == 'm3u8':
  57                 formats.extend(self._extract_m3u8_formats(
  58                     video_url, video_id, 'mp4', 'm3u8_native',
  59                     m3u8_id=format_id or 'hls', fatal=False))
  60             elif ext == 'mpd':
  61                 continue
  62             #     formats.extend(self._extract_mpd_formats(
  63             #         video_url, video_id, format_id or 'dash', fatal=False))
  64             else:
  65                 formats.append({
  66                     'url': video_url,
  67                     'format_id': format_id,
  68                     'vcodec': video.get('videoencoding') or video.get('video_codec'),
  69                     'width': int_or_none(video.get('width')),
  70                     'height': int_or_none(video.get('height')),
  71                     'filesize': get_file_size(video.get('file_size') or video.get('fileSize')),
  72                     'tbr': int_or_none(video.get('bitrate'), 1000) or None,
  73                     'ext': ext,
  74                 })
  75         self._sort_formats(formats, ('height', 'width', 'filesize', 'tbr', 'fps', 'format_id'))
  76
  77         thumbnails = []
  78         for image in video_data.get('images', []):
  79             image_url = image.get('url')
  80             if not image_url:
  81                 continue
  82             thumbnails.append({
  83                 'url': 'http://www.nytimes.com/' + image_url,
  84                 'width': int_or_none(image.get('width')),
  85                 'height': int_or_none(image.get('height')),
  86             })
  87
  88         publication_date = video_data.get('publication_date')
  89         timestamp = parse_iso8601(publication_date[:-8]) if publication_date else None
  90
  91         return {
  92             'id': video_id,
  93             'title': title,
  94             'description': video_data.get('summary'),
  95             'timestamp': timestamp,
  96             'uploader': video_data.get('byline'),
  97             'duration': float_or_none(video_data.get('duration'), 1000),
  98             'formats': formats,
  99             'thumbnails': thumbnails,
 100         }
 101
 102
 103 class NYTimesIE(NYTimesBaseIE):
 104     _VALID_URL = r'https?://(?:(?:www\.)?nytimes\.com/video/(?:[^/]+/)+?|graphics8\.nytimes\.com/bcvideo/\d+(?:\.\d+)?/iframe/embed\.html\?videoId=)(?P<id>\d+)'
 105
 106     _TESTS = [{
 107         'url': 'http://www.nytimes.com/video/opinion/100000002847155/verbatim-what-is-a-photocopier.html?playlistId=100000001150263',
 108         'md5': 'd665342765db043f7e225cff19df0f2d',
 109         'info_dict': {
 110             'id': '100000002847155',
 111             'ext': 'mov',
 112             'title': 'Verbatim: What Is a Photocopier?',
 113             'description': 'md5:93603dada88ddbda9395632fdc5da260',
 114             'timestamp': 1398631707,
 115             'upload_date': '20140427',
 116             'uploader': 'Brett Weiner',
 117             'duration': 419,
 118         }
 119     }, {
 120         'url': 'http://www.nytimes.com/video/travel/100000003550828/36-hours-in-dubai.html',
 121         'only_matching': True,
 122     }]
 123
 124     def _real_extract(self, url):
 125         video_id = self._match_id(url)
 126
 127         return self._extract_video_from_id(video_id)
 128
 129
 130 class NYTimesArticleIE(NYTimesBaseIE):
 131     _VALID_URL = r'https?://(?:www\.)?nytimes\.com/(.(?<!video))*?/(?:[^/]+/)*(?P<id>[^.]+)(?:\.html)?'
 132     _TESTS = [{
 133         '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',
 134         'md5': 'e2076d58b4da18e6a001d53fd56db3c9',
 135         'info_dict': {
 136             'id': '100000003628438',
 137             'ext': 'mov',
 138             'title': 'New Minimum Wage: $70,000 a Year',
 139             'description': 'Dan Price, C.E.O. of Gravity Payments, surprised his 120-person staff by announcing that he planned over the next three years to raise the salary of every employee to $70,000 a year.',
 140             'timestamp': 1429033037,
 141             'upload_date': '20150414',
 142             'uploader': 'Matthew Williams',
 143         }
 144     }, {
 145         'url': 'http://www.nytimes.com/2016/10/14/podcasts/revelations-from-the-final-weeks.html',
 146         'md5': 'e0d52040cafb07662acf3c9132db3575',
 147         'info_dict': {
 148             'id': '100000004709062',
 149             'title': 'The Run-Up: ‘He Was Like an Octopus’',
 150             'ext': 'mp3',
 151             'description': 'md5:fb5c6b93b12efc51649b4847fe066ee4',
 152             'series': 'The Run-Up',
 153             'episode': '‘He Was Like an Octopus’',
 154             'episode_number': 20,
 155             'duration': 2130,
 156         }
 157     }, {
 158         'url': 'http://www.nytimes.com/2016/10/16/books/review/inside-the-new-york-times-book-review-the-rise-of-hitler.html',
 159         'info_dict': {
 160             'id': '100000004709479',
 161             'title': 'The Rise of Hitler',
 162             'ext': 'mp3',
 163             'description': 'md5:bce877fd9e3444990cb141875fab0028',
 164             'creator': 'Pamela Paul',
 165             'duration': 3475,
 166         },
 167         'params': {
 168             'skip_download': True,
 169         },
 170     }, {
 171         'url': 'http://www.nytimes.com/news/minute/2014/03/17/times-minute-whats-next-in-crimea/?_php=true&_type=blogs&_php=true&_type=blogs&_r=1',
 172         'only_matching': True,
 173     }]
 174
 175     def _extract_podcast_from_json(self, json, page_id, webpage):
 176         podcast_audio = self._parse_json(
 177             json, page_id, transform_source=js_to_json)
 178
 179         audio_data = podcast_audio['data']
 180         track = audio_data['track']
 181
 182         episode_title = track['title']
 183         video_url = track['source']
 184
 185         description = track.get('description') or self._html_search_meta(
 186             ['og:description', 'twitter:description'], webpage)
 187
 188         podcast_title = audio_data.get('podcast', {}).get('title')
 189         title = ('%s: %s' % (podcast_title, episode_title)
 190                  if podcast_title else episode_title)
 191
 192         episode = audio_data.get('podcast', {}).get('episode') or ''
 193         episode_number = int_or_none(self._search_regex(
 194             r'[Ee]pisode\s+(\d+)', episode, 'episode number', default=None))
 195
 196         return {
 197             'id': remove_start(podcast_audio.get('target'), 'FT') or page_id,
 198             'url': video_url,
 199             'title': title,
 200             'description': description,
 201             'creator': track.get('credit'),
 202             'series': podcast_title,
 203             'episode': episode_title,
 204             'episode_number': episode_number,
 205             'duration': int_or_none(track.get('duration')),
 206         }
 207
 208     def _real_extract(self, url):
 209         page_id = self._match_id(url)
 210
 211         webpage = self._download_webpage(url, page_id)
 212
 213         video_id = self._search_regex(
 214             r'data-videoid=["\'](\d+)', webpage, 'video id',
 215             default=None, fatal=False)
 216         if video_id is not None:
 217             return self._extract_video_from_id(video_id)
 218
 219         podcast_data = self._search_regex(
 220             (r'NYTD\.FlexTypes\.push\s*\(\s*({.+?})\s*\)\s*;\s*</script',
 221              r'NYTD\.FlexTypes\.push\s*\(\s*({.+})\s*\)\s*;'),
 222             webpage, 'podcast data')
 223         return self._extract_podcast_from_json(podcast_data, page_id, webpage)