Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/nytimes.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     float_or_none,
   6     int_or_none,
   7     parse_iso8601,
   8 )
   9
  10
  11 class NYTimesBaseIE(InfoExtractor):
  12     def _extract_video_from_id(self, video_id):
  13         video_data = self._download_json(
  14             'http://www.nytimes.com/svc/video/api/v2/video/%s' % video_id,
  15             video_id, 'Downloading video JSON')
  16
  17         title = video_data['headline']
  18         description = video_data.get('summary')
  19         duration = float_or_none(video_data.get('duration'), 1000)
  20
  21         uploader = video_data.get('byline')
  22         publication_date = video_data.get('publication_date')
  23         timestamp = parse_iso8601(publication_date[:-8]) if publication_date else None
  24
  25         def get_file_size(file_size):
  26             if isinstance(file_size, int):
  27                 return file_size
  28             elif isinstance(file_size, dict):
  29                 return int(file_size.get('value', 0))
  30             else:
  31                 return 0
  32
  33         formats = [
  34             {
  35                 'url': video['url'],
  36                 'format_id': video.get('type'),
  37                 'vcodec': video.get('video_codec'),
  38                 'width': int_or_none(video.get('width')),
  39                 'height': int_or_none(video.get('height')),
  40                 'filesize': get_file_size(video.get('fileSize')),
  41             } for video in video_data['renditions'] if video.get('url')
  42         ]
  43         self._sort_formats(formats)
  44
  45         thumbnails = [
  46             {
  47                 'url': 'http://www.nytimes.com/%s' % image['url'],
  48                 'width': int_or_none(image.get('width')),
  49                 'height': int_or_none(image.get('height')),
  50             } for image in video_data.get('images', []) if image.get('url')
  51         ]
  52
  53         return {
  54             'id': video_id,
  55             'title': title,
  56             'description': description,
  57             'timestamp': timestamp,
  58             'uploader': uploader,
  59             'duration': duration,
  60             'formats': formats,
  61             'thumbnails': thumbnails,
  62         }
  63
  64
  65 class NYTimesIE(NYTimesBaseIE):
  66     _VALID_URL = r'https?://(?:(?:www\.)?nytimes\.com/video/(?:[^/]+/)+?|graphics8\.nytimes\.com/bcvideo/\d+(?:\.\d+)?/iframe/embed\.html\?videoId=)(?P<id>\d+)'
  67
  68     _TESTS = [{
  69         'url': 'http://www.nytimes.com/video/opinion/100000002847155/verbatim-what-is-a-photocopier.html?playlistId=100000001150263',
  70         'md5': '18a525a510f942ada2720db5f31644c0',
  71         'info_dict': {
  72             'id': '100000002847155',
  73             'ext': 'mov',
  74             'title': 'Verbatim: What Is a Photocopier?',
  75             'description': 'md5:93603dada88ddbda9395632fdc5da260',
  76             'timestamp': 1398631707,
  77             'upload_date': '20140427',
  78             'uploader': 'Brett Weiner',
  79             'duration': 419,
  80         }
  81     }, {
  82         'url': 'http://www.nytimes.com/video/travel/100000003550828/36-hours-in-dubai.html',
  83         'only_matching': True,
  84     }]
  85
  86     def _real_extract(self, url):
  87         video_id = self._match_id(url)
  88
  89         return self._extract_video_from_id(video_id)
  90
  91
  92 class NYTimesArticleIE(NYTimesBaseIE):
  93     _VALID_URL = r'https?://(?:www\.)?nytimes\.com/(.(?<!video))*?/(?:[^/]+/)*(?P<id>[^.]+)(?:\.html)?'
  94     _TESTS = [{
  95         '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',
  96         'md5': 'e2076d58b4da18e6a001d53fd56db3c9',
  97         'info_dict': {
  98             'id': '100000003628438',
  99             'ext': 'mov',
 100             'title': 'New Minimum Wage: $70,000 a Year',
 101             '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.',
 102             'timestamp': 1429033037,
 103             'upload_date': '20150414',
 104             'uploader': 'Matthew Williams',
 105         }
 106     }, {
 107         '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',
 108         'only_matching': True,
 109     }]
 110
 111     def _real_extract(self, url):
 112         video_id = self._match_id(url)
 113
 114         webpage = self._download_webpage(url, video_id)
 115
 116         video_id = self._html_search_regex(r'data-videoid="(\d+)"', webpage, 'video id')
 117
 118         return self._extract_video_from_id(video_id)