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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     parse_duration,
   8     parse_iso8601,
   9 )
  10
  11
  12 class MLBIE(InfoExtractor):
  13     _VALID_URL = r'''(?x)
  14                     https?://
  15                         (?:[\da-z_-]+\.)*mlb\.com/
  16                         (?:
  17                             (?:
  18                                 (?:.*?/)?video/(?:topic/[\da-z_-]+/)?v|
  19                                 (?:
  20                                     shared/video/embed/(?:embed|m-internal-embed)\.html|
  21                                     (?:[^/]+/)+(?:play|index)\.jsp|
  22                                 )\?.*?\bcontent_id=
  23                             )
  24                             (?P<id>n?\d+)|
  25                             (?:[^/]+/)*(?P<path>[^/]+)
  26                         )
  27                     '''
  28     _TESTS = [
  29         {
  30             'url': 'http://m.mlb.com/sea/video/topic/51231442/v34698933/nymsea-ackley-robs-a-home-run-with-an-amazing-catch/?c_id=sea',
  31             'md5': 'ff56a598c2cf411a9a38a69709e97079',
  32             'info_dict': {
  33                 'id': '34698933',
  34                 'ext': 'mp4',
  35                 'title': "Ackley's spectacular catch",
  36                 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
  37                 'duration': 66,
  38                 'timestamp': 1405980600,
  39                 'upload_date': '20140721',
  40                 'thumbnail': r're:^https?://.*\.jpg$',
  41             },
  42         },
  43         {
  44             'url': 'http://m.mlb.com/video/topic/81536970/v34496663/mianym-stanton-practices-for-the-home-run-derby',
  45             'md5': 'd9c022c10d21f849f49c05ae12a8a7e9',
  46             'info_dict': {
  47                 'id': '34496663',
  48                 'ext': 'mp4',
  49                 'title': 'Stanton prepares for Derby',
  50                 'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
  51                 'duration': 46,
  52                 'timestamp': 1405105800,
  53                 'upload_date': '20140711',
  54                 'thumbnail': r're:^https?://.*\.jpg$',
  55             },
  56         },
  57         {
  58             'url': 'http://m.mlb.com/video/topic/vtp_hrd_sponsor/v34578115/hrd-cespedes-wins-2014-gillette-home-run-derby',
  59             'md5': '0e6e73d509321e142409b695eadd541f',
  60             'info_dict': {
  61                 'id': '34578115',
  62                 'ext': 'mp4',
  63                 'title': 'Cespedes repeats as Derby champ',
  64                 'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
  65                 'duration': 488,
  66                 'timestamp': 1405399936,
  67                 'upload_date': '20140715',
  68                 'thumbnail': r're:^https?://.*\.jpg$',
  69             },
  70         },
  71         {
  72             'url': 'http://m.mlb.com/video/v34577915/bautista-on-derby-captaining-duties-his-performance',
  73             'md5': 'b8fd237347b844365d74ea61d4245967',
  74             'info_dict': {
  75                 'id': '34577915',
  76                 'ext': 'mp4',
  77                 'title': 'Bautista on Home Run Derby',
  78                 'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
  79                 'duration': 52,
  80                 'timestamp': 1405390722,
  81                 'upload_date': '20140715',
  82                 'thumbnail': r're:^https?://.*\.jpg$',
  83             },
  84         },
  85         {
  86             'url': 'http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer',
  87             'md5': 'b190e70141fb9a1552a85426b4da1b5d',
  88             'info_dict': {
  89                 'id': '75609783',
  90                 'ext': 'mp4',
  91                 'title': 'Must C: Pillar climbs for catch',
  92                 'description': '4/15/15: Blue Jays outfielder Kevin Pillar continues his defensive dominance by climbing the wall in left to rob Tim Beckham of a home run',
  93                 'timestamp': 1429124820,
  94                 'upload_date': '20150415',
  95             }
  96         },
  97         {
  98             'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
  99             'only_matching': True,
 100         },
 101         {
 102             'url': 'http://mlb.mlb.com/shared/video/embed/embed.html?content_id=36599553',
 103             'only_matching': True,
 104         },
 105         {
 106             'url': 'http://mlb.mlb.com/es/video/play.jsp?content_id=36599553',
 107             'only_matching': True,
 108         },
 109         {
 110             'url': 'http://m.cardinals.mlb.com/stl/video/v51175783/atlstl-piscotty-makes-great-sliding-catch-on-line/?partnerId=as_mlb_20150321_42500876&adbid=579409712979910656&adbpl=tw&adbpr=52847728',
 111             'only_matching': True,
 112         },
 113         {
 114             # From http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer
 115             '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',
 116             'only_matching': True,
 117         },
 118         {
 119             'url': 'http://washington.nationals.mlb.com/mlb/gameday/index.jsp?c_id=was&gid=2015_05_09_atlmlb_wasmlb_1&lang=en&content_id=108309983&mode=video#',
 120             'only_matching': True,
 121         }
 122     ]
 123
 124     def _real_extract(self, url):
 125         mobj = re.match(self._VALID_URL, url)
 126         video_id = mobj.group('id')
 127
 128         if not video_id:
 129             video_path = mobj.group('path')
 130             webpage = self._download_webpage(url, video_path)
 131             video_id = self._search_regex(
 132                 [r'data-video-?id="(\d+)"', r'content_id=(\d+)'], webpage, 'video id')
 133
 134         detail = self._download_xml(
 135             'http://m.mlb.com/gen/multimedia/detail/%s/%s/%s/%s.xml'
 136             % (video_id[-3], video_id[-2], video_id[-1], video_id), video_id)
 137
 138         title = detail.find('./headline').text
 139         description = detail.find('./big-blurb').text
 140         duration = parse_duration(detail.find('./duration').text)
 141         timestamp = parse_iso8601(detail.attrib['date'][:-5])
 142
 143         thumbnails = [{
 144             'url': thumbnail.text,
 145         } for thumbnail in detail.findall('./thumbnailScenarios/thumbnailScenario')]
 146
 147         formats = []
 148         for media_url in detail.findall('./url'):
 149             playback_scenario = media_url.attrib['playback_scenario']
 150             fmt = {
 151                 'url': media_url.text,
 152                 'format_id': playback_scenario,
 153             }
 154             m = re.search(r'(?P<vbr>\d+)K_(?P<width>\d+)X(?P<height>\d+)', playback_scenario)
 155             if m:
 156                 fmt.update({
 157                     'vbr': int(m.group('vbr')) * 1000,
 158                     'width': int(m.group('width')),
 159                     'height': int(m.group('height')),
 160                 })
 161             formats.append(fmt)
 162
 163         self._sort_formats(formats)
 164
 165         return {
 166             'id': video_id,
 167             'title': title,
 168             'description': description,
 169             'duration': duration,
 170             'timestamp': timestamp,
 171             'formats': formats,
 172             'thumbnails': thumbnails,
 173         }