youtube_dl/extractor/nova.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7
   8
   9 class NovaIE(InfoExtractor):
  10     IE_DESC = 'TN.cz, Prásk.tv, Nova.cz, Novaplus.cz, FANDA.tv, Krásná.cz and Doma.cz'
  11     _VALID_URL = 'http://(?:[^.]+\.)?(?P<site>tv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P<id>[^/]+)(?:\.html|/?)'
  12     _TESTS = [{
  13         'url': 'http://tvnoviny.nova.cz/clanek/novinky/co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou.html',
  14         'info_dict': {
  15             'id': '1608920',
  16             'display_id': 'co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou',
  17             'ext': 'flv',
  18             'title': 'Duel: Michal Hrdlička a Petr Suchoň',
  19             'description': 'md5:d0cc509858eee1b1374111c588c6f5d5',
  20             'thumbnail': 're:^https?://.*\.(?:jpg)',
  21         },
  22         'params': {
  23             # rtmp download
  24             'skip_download': True,
  25         }
  26     }, {
  27         'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html',
  28         'md5': '1dd7b9d5ea27bc361f110cd855a19bd3',
  29         'info_dict': {
  30             'id': '1757139',
  31             'display_id': 'tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci',
  32             'ext': 'mp4',
  33             'title': 'Podzemní nemocnice v pražské Krči',
  34             'description': 'md5:f0a42dd239c26f61c28f19e62d20ef53',
  35             'thumbnail': 're:^https?://.*\.(?:jpg)',
  36         }
  37     }, {
  38         'url': 'http://novaplus.nova.cz/porad/policie-modrava/video/5591-policie-modrava-15-dil-blondynka-na-hrbitove/',
  39         'info_dict': {
  40             'id': '1756825',
  41             'display_id': '5591-policie-modrava-15-dil-blondynka-na-hrbitove',
  42             'ext': 'flv',
  43             'title': 'Policie Modrava - 15. díl - Blondýnka na hřbitově',
  44             'description': 'md5:d804ba6b30bc7da2705b1fea961bddfe',
  45             'thumbnail': 're:^https?://.*\.(?:jpg)',
  46         },
  47         'params': {
  48             # rtmp download
  49             'skip_download': True,
  50         }
  51     }, {
  52         'url': 'http://sport.tn.nova.cz/clanek/sport/hokej/nhl/zivot-jde-dal-hodnotil-po-vyrazeni-z-playoff-jiri-sekac.html',
  53         'only_matching': True,
  54     }, {
  55         'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
  56         'only_matching': True,
  57     }, {
  58         'url': 'http://doma.nova.cz/clanek/zdravi/prijdte-se-zapsat-do-registru-kostni-drene-jiz-ve-stredu-3-cervna.html',
  59         'only_matching': True,
  60     }, {
  61         'url': 'http://prask.nova.cz/clanek/novinky/co-si-na-sobe-nase-hvezdy-nechaly-pojistit.html',
  62         'only_matching': True,
  63     }, {
  64         'url': 'http://tv.nova.cz/clanek/novinky/zivot-je-zivot-bondovsky-trailer.html',
  65         'only_matching': True,
  66     }]
  67
  68     def _real_extract(self, url):
  69         mobj = re.match(self._VALID_URL, url)
  70         display_id = mobj.group('id')
  71         site = mobj.group('site')
  72
  73         webpage = self._download_webpage(url, display_id)
  74
  75         video_id = self._search_regex(
  76             [r"(?:media|video_id)\s*:\s*'(\d+)'",
  77              r'media=(\d+)',
  78              r'id="article_video_(\d+)"',
  79              r'id="player_(\d+)"'],
  80             webpage, 'video id')
  81
  82         config_url = self._search_regex(
  83             r'src="(http://tn\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"',
  84             webpage, 'config url', default=None)
  85
  86         if not config_url:
  87             DEFAULT_SITE_ID = '23000'
  88             SITES = {
  89                 'tvnoviny': DEFAULT_SITE_ID,
  90                 'novaplus': DEFAULT_SITE_ID,
  91                 'vymena': DEFAULT_SITE_ID,
  92                 'krasna': DEFAULT_SITE_ID,
  93                 'fanda': '30',
  94                 'tn': '30',
  95                 'doma': '30',
  96             }
  97
  98             site_id = self._search_regex(
  99                 r'site=(\d+)', webpage, 'site id', default=None) or SITES.get(site, DEFAULT_SITE_ID)
 100
 101             config_url = ('http://tn.nova.cz/bin/player/videojs/config.php?site=%s&media=%s&jsVar=vjsconfig'
 102                           % (site_id, video_id))
 103
 104         config = self._download_json(
 105             config_url, display_id,
 106             'Downloading config JSON',
 107             transform_source=lambda s: re.sub(r'var\s+[\da-zA-Z_]+\s*=\s*({.+?});', r'\1', s))
 108
 109         mediafile = config['mediafile']
 110         video_url = mediafile['src']
 111
 112         m = re.search(r'^(?P<url>rtmpe?://[^/]+/(?P<app>[^/]+?))/&*(?P<playpath>.+)$', video_url)
 113         if m:
 114             formats = [{
 115                 'url': m.group('url'),
 116                 'app': m.group('app'),
 117                 'play_path': m.group('playpath'),
 118                 'player_path': 'http://tvnoviny.nova.cz/static/shared/app/videojs/video-js.swf',
 119                 'ext': 'flv',
 120             }]
 121         else:
 122             formats = [{
 123                 'url': video_url,
 124             }]
 125         self._sort_formats(formats)
 126
 127         title = mediafile.get('meta', {}).get('title') or self._og_search_title(webpage)
 128         description = self._og_search_description(webpage)
 129         thumbnail = config.get('poster')
 130
 131         return {
 132             'id': video_id,
 133             'display_id': display_id,
 134             'title': title,
 135             'description': description,
 136             'thumbnail': thumbnail,
 137             'formats': formats,
 138         }