youtube_dl/extractor/nhl.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..compat import compat_str
   7 from ..utils import (
   8     determine_ext,
   9     int_or_none,
  10     parse_iso8601,
  11     parse_duration,
  12 )
  13
  14
  15 class NHLBaseIE(InfoExtractor):
  16     def _real_extract(self, url):
  17         site, tmp_id = re.match(self._VALID_URL, url).groups()
  18         video_data = self._download_json(
  19             'https://%s/%s/%sid/v1/%s/details/web-v1.json'
  20             % (self._CONTENT_DOMAIN, site[:3], 'item/' if site == 'mlb' else '', tmp_id), tmp_id)
  21         if video_data.get('type') != 'video':
  22             video_data = video_data['media']
  23             video = video_data.get('video')
  24             if video:
  25                 video_data = video
  26             else:
  27                 videos = video_data.get('videos')
  28                 if videos:
  29                     video_data = videos[0]
  30
  31         video_id = compat_str(video_data['id'])
  32         title = video_data['title']
  33
  34         formats = []
  35         for playback in video_data.get('playbacks', []):
  36             playback_url = playback.get('url')
  37             if not playback_url:
  38                 continue
  39             ext = determine_ext(playback_url)
  40             if ext == 'm3u8':
  41                 m3u8_formats = self._extract_m3u8_formats(
  42                     playback_url, video_id, 'mp4', 'm3u8_native',
  43                     m3u8_id=playback.get('name', 'hls'), fatal=False)
  44                 self._check_formats(m3u8_formats, video_id)
  45                 formats.extend(m3u8_formats)
  46             else:
  47                 height = int_or_none(playback.get('height'))
  48                 formats.append({
  49                     'format_id': playback.get('name', 'http' + ('-%dp' % height if height else '')),
  50                     'url': playback_url,
  51                     'width': int_or_none(playback.get('width')),
  52                     'height': height,
  53                     'tbr': int_or_none(self._search_regex(r'_(\d+)[kK]', playback_url, 'bitrate', default=None)),
  54                 })
  55         self._sort_formats(formats)
  56
  57         thumbnails = []
  58         cuts = video_data.get('image', {}).get('cuts') or []
  59         if isinstance(cuts, dict):
  60             cuts = cuts.values()
  61         for thumbnail_data in cuts:
  62             thumbnail_url = thumbnail_data.get('src')
  63             if not thumbnail_url:
  64                 continue
  65             thumbnails.append({
  66                 'url': thumbnail_url,
  67                 'width': int_or_none(thumbnail_data.get('width')),
  68                 'height': int_or_none(thumbnail_data.get('height')),
  69             })
  70
  71         return {
  72             'id': video_id,
  73             'title': title,
  74             'description': video_data.get('description'),
  75             'timestamp': parse_iso8601(video_data.get('date')),
  76             'duration': parse_duration(video_data.get('duration')),
  77             'thumbnails': thumbnails,
  78             'formats': formats,
  79         }
  80
  81
  82 class NHLIE(NHLBaseIE):
  83     IE_NAME = 'nhl.com'
  84     _VALID_URL = r'https?://(?:www\.)?(?P<site>nhl|wch2016)\.com/(?:[^/]+/)*c-(?P<id>\d+)'
  85     _CONTENT_DOMAIN = 'nhl.bamcontent.com'
  86     _TESTS = [{
  87         # type=video
  88         'url': 'https://www.nhl.com/video/anisimov-cleans-up-mess/t-277752844/c-43663503',
  89         'md5': '0f7b9a8f986fb4b4eeeece9a56416eaf',
  90         'info_dict': {
  91             'id': '43663503',
  92             'ext': 'mp4',
  93             'title': 'Anisimov cleans up mess',
  94             'description': 'md5:a02354acdfe900e940ce40706939ca63',
  95             'timestamp': 1461288600,
  96             'upload_date': '20160422',
  97         },
  98     }, {
  99         # type=article
 100         'url': 'https://www.nhl.com/news/dennis-wideman-suspended/c-278258934',
 101         'md5': '1f39f4ea74c1394dea110699a25b366c',
 102         'info_dict': {
 103             'id': '40784403',
 104             'ext': 'mp4',
 105             'title': 'Wideman suspended by NHL',
 106             'description': 'Flames defenseman Dennis Wideman was banned 20 games for violation of Rule 40 (Physical Abuse of Officials)',
 107             'upload_date': '20160204',
 108             'timestamp': 1454544904,
 109         },
 110     }, {
 111         # Some m3u8 URLs are invalid (https://github.com/ytdl-org/youtube-dl/issues/10713)
 112         'url': 'https://www.nhl.com/predators/video/poile-laviolette-on-subban-trade/t-277437416/c-44315003',
 113         'md5': '50b2bb47f405121484dda3ccbea25459',
 114         'info_dict': {
 115             'id': '44315003',
 116             'ext': 'mp4',
 117             'title': 'Poile, Laviolette on Subban trade',
 118             'description': 'General manager David Poile and head coach Peter Laviolette share their thoughts on acquiring P.K. Subban from Montreal (06/29/16)',
 119             'timestamp': 1467242866,
 120             'upload_date': '20160629',
 121         },
 122     }, {
 123         'url': 'https://www.wch2016.com/video/caneur-best-of-game-2-micd-up/t-281230378/c-44983703',
 124         'only_matching': True,
 125     }, {
 126         'url': 'https://www.wch2016.com/news/3-stars-team-europe-vs-team-canada/c-282195068',
 127         'only_matching': True,
 128     }]