yt_dlp/extractor/mlb.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     determine_ext,
   6     int_or_none,
   7     parse_duration,
   8     parse_iso8601,
   9     try_get,
  10 )
  11
  12
  13 class MLBBaseIE(InfoExtractor):
  14     def _real_extract(self, url):
  15         display_id = self._match_id(url)
  16         video = self._download_video_data(display_id)
  17         video_id = video['id']
  18         title = video['title']
  19         feed = self._get_feed(video)
  20
  21         formats = []
  22         for playback in (feed.get('playbacks') or []):
  23             playback_url = playback.get('url')
  24             if not playback_url:
  25                 continue
  26             name = playback.get('name')
  27             ext = determine_ext(playback_url)
  28             if ext == 'm3u8':
  29                 formats.extend(self._extract_m3u8_formats(
  30                     playback_url, video_id, 'mp4',
  31                     'm3u8_native', m3u8_id=name, fatal=False))
  32             else:
  33                 f = {
  34                     'format_id': name,
  35                     'url': playback_url,
  36                 }
  37                 mobj = re.search(r'_(\d+)K_(\d+)X(\d+)', name)
  38                 if mobj:
  39                     f.update({
  40                         'height': int(mobj.group(3)),
  41                         'tbr': int(mobj.group(1)),
  42                         'width': int(mobj.group(2)),
  43                     })
  44                 mobj = re.search(r'_(\d+)x(\d+)_(\d+)_(\d+)K\.mp4', playback_url)
  45                 if mobj:
  46                     f.update({
  47                         'fps': int(mobj.group(3)),
  48                         'height': int(mobj.group(2)),
  49                         'tbr': int(mobj.group(4)),
  50                         'width': int(mobj.group(1)),
  51                     })
  52                 formats.append(f)
  53         self._sort_formats(formats)
  54
  55         thumbnails = []
  56         for cut in (try_get(feed, lambda x: x['image']['cuts'], list) or []):
  57             src = cut.get('src')
  58             if not src:
  59                 continue
  60             thumbnails.append({
  61                 'height': int_or_none(cut.get('height')),
  62                 'url': src,
  63                 'width': int_or_none(cut.get('width')),
  64             })
  65
  66         language = (video.get('language') or 'EN').lower()
  67
  68         return {
  69             'id': video_id,
  70             'title': title,
  71             'formats': formats,
  72             'description': video.get('description'),
  73             'duration': parse_duration(feed.get('duration')),
  74             'thumbnails': thumbnails,
  75             'timestamp': parse_iso8601(video.get(self._TIMESTAMP_KEY)),
  76             'subtitles': self._extract_mlb_subtitles(feed, language),
  77         }
  78
  79
  80 class MLBIE(MLBBaseIE):
  81     _VALID_URL = r'''(?x)
  82                     https?://
  83                         (?:[\da-z_-]+\.)*mlb\.com/
  84                         (?:
  85                             (?:
  86                                 (?:[^/]+/)*video/[^/]+/c-|
  87                                 (?:
  88                                     shared/video/embed/(?:embed|m-internal-embed)\.html|
  89                                     (?:[^/]+/)+(?:play|index)\.jsp|
  90                                 )\?.*?\bcontent_id=
  91                             )
  92                             (?P<id>\d+)
  93                         )
  94                     '''
  95     _TESTS = [
  96         {
  97             'url': 'https://www.mlb.com/mariners/video/ackleys-spectacular-catch/c-34698933',
  98             'md5': '632358dacfceec06bad823b83d21df2d',
  99             'info_dict': {
 100                 'id': '34698933',
 101                 'ext': 'mp4',
 102                 'title': "Ackley's spectacular catch",
 103                 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
 104                 'duration': 66,
 105                 'timestamp': 1405995000,
 106                 'upload_date': '20140722',
 107                 'thumbnail': r're:^https?://.*\.jpg$',
 108             },
 109         },
 110         {
 111             'url': 'https://www.mlb.com/video/stanton-prepares-for-derby/c-34496663',
 112             'md5': 'bf2619bf9cacc0a564fc35e6aeb9219f',
 113             'info_dict': {
 114                 'id': '34496663',
 115                 'ext': 'mp4',
 116                 'title': 'Stanton prepares for Derby',
 117                 'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
 118                 'duration': 46,
 119                 'timestamp': 1405120200,
 120                 'upload_date': '20140711',
 121                 'thumbnail': r're:^https?://.*\.jpg$',
 122             },
 123         },
 124         {
 125             'url': 'https://www.mlb.com/video/cespedes-repeats-as-derby-champ/c-34578115',
 126             'md5': '99bb9176531adc600b90880fb8be9328',
 127             'info_dict': {
 128                 'id': '34578115',
 129                 'ext': 'mp4',
 130                 'title': 'Cespedes repeats as Derby champ',
 131                 'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
 132                 'duration': 488,
 133                 'timestamp': 1405414336,
 134                 'upload_date': '20140715',
 135                 'thumbnail': r're:^https?://.*\.jpg$',
 136             },
 137         },
 138         {
 139             'url': 'https://www.mlb.com/video/bautista-on-home-run-derby/c-34577915',
 140             'md5': 'da8b57a12b060e7663ee1eebd6f330ec',
 141             'info_dict': {
 142                 'id': '34577915',
 143                 'ext': 'mp4',
 144                 'title': 'Bautista on Home Run Derby',
 145                 'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
 146                 'duration': 52,
 147                 'timestamp': 1405405122,
 148                 'upload_date': '20140715',
 149                 'thumbnail': r're:^https?://.*\.jpg$',
 150             },
 151         },
 152         {
 153             'url': 'https://www.mlb.com/video/hargrove-homers-off-caldwell/c-1352023483?tid=67793694',
 154             'only_matching': True,
 155         },
 156         {
 157             'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
 158             'only_matching': True,
 159         },
 160         {
 161             'url': 'http://mlb.mlb.com/shared/video/embed/embed.html?content_id=36599553',
 162             'only_matching': True,
 163         },
 164         {
 165             'url': 'http://mlb.mlb.com/es/video/play.jsp?content_id=36599553',
 166             'only_matching': True,
 167         },
 168         {
 169             'url': 'https://www.mlb.com/cardinals/video/piscottys-great-sliding-catch/c-51175783',
 170             'only_matching': True,
 171         },
 172         {
 173             # From http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer
 174             'url': 'http://mlb.mlb.com/shared/video/embed/m-internal-embed.html?content_id=75609783&property=mlb&autoplay=true&hashmode=false&siteSection=mlb/multimedia/article_118550098/article_embed&club=mlb',
 175             'only_matching': True,
 176         },
 177     ]
 178     _TIMESTAMP_KEY = 'date'
 179
 180     @staticmethod
 181     def _get_feed(video):
 182         return video
 183
 184     @staticmethod
 185     def _extract_mlb_subtitles(feed, language):
 186         subtitles = {}
 187         for keyword in (feed.get('keywordsAll') or []):
 188             keyword_type = keyword.get('type')
 189             if keyword_type and keyword_type.startswith('closed_captions_location_'):
 190                 cc_location = keyword.get('value')
 191                 if cc_location:
 192                     subtitles.setdefault(language, []).append({
 193                         'url': cc_location,
 194                     })
 195         return subtitles
 196
 197     def _download_video_data(self, display_id):
 198         return self._download_json(
 199             'http://content.mlb.com/mlb/item/id/v1/%s/details/web-v1.json' % display_id,
 200             display_id)
 201
 202
 203 class MLBVideoIE(MLBBaseIE):
 204     _VALID_URL = r'https?://(?:www\.)?mlb\.com/(?:[^/]+/)*video/(?P<id>[^/?&#]+)'
 205     _TEST = {
 206         'url': 'https://www.mlb.com/mariners/video/ackley-s-spectacular-catch-c34698933',
 207         'md5': '632358dacfceec06bad823b83d21df2d',
 208         'info_dict': {
 209             'id': 'c04a8863-f569-42e6-9f87-992393657614',
 210             'ext': 'mp4',
 211             'title': "Ackley's spectacular catch",
 212             'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
 213             'duration': 66,
 214             'timestamp': 1405995000,
 215             'upload_date': '20140722',
 216             'thumbnail': r're:^https?://.+',
 217         },
 218     }
 219     _TIMESTAMP_KEY = 'timestamp'
 220
 221     @classmethod
 222     def suitable(cls, url):
 223         return False if MLBIE.suitable(url) else super(MLBVideoIE, cls).suitable(url)
 224
 225     @staticmethod
 226     def _get_feed(video):
 227         return video['feeds'][0]
 228
 229     @staticmethod
 230     def _extract_mlb_subtitles(feed, language):
 231         subtitles = {}
 232         for cc_location in (feed.get('closedCaptions') or []):
 233             subtitles.setdefault(language, []).append({
 234                 'url': cc_location,
 235             })
 236
 237     def _download_video_data(self, display_id):
 238         # https://www.mlb.com/data-service/en/videos/[SLUG]
 239         return self._download_json(
 240             'https://fastball-gateway.mlb.com/graphql',
 241             display_id, query={
 242                 'query': '''{
 243   mediaPlayback(ids: "%s") {
 244     description
 245     feeds(types: CMS) {
 246       closedCaptions
 247       duration
 248       image {
 249         cuts {
 250           width
 251           height
 252           src
 253         }
 254       }
 255       playbacks {
 256         name
 257         url
 258       }
 259     }
 260     id
 261     timestamp
 262     title
 263   }
 264 }''' % display_id,
 265             })['data']['mediaPlayback'][0]