yt_dlp/extractor/mlb.py

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