_ Git - youtube-dl/blob - youtube_dl/extractor/indavideo.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .. import utils
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     int_or_none,
   8     parse_age_limit,
   9     parse_iso8601,
  10 )
  11
  12
  13 class IndavideoEmbedIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:(?:embed\.)?indavideo\.hu/player/video/|assets\.indavideo\.hu/swf/player\.swf\?.*\b(?:v(?:ID|id))=)(?P<id>[\da-f]+)'
  15     _TESTS = [{
  16         'url': 'http://indavideo.hu/player/video/1bdc3c6d80/',
  17         'md5': 'f79b009c66194acacd40712a6778acfa',
  18         'info_dict': {
  19             'id': '1837039',
  20             'ext': 'mp4',
  21             'title': 'Cicatánc',
  22             'description': '',
  23             'thumbnail': 're:^https?://.*\.jpg$',
  24             'uploader': 'cukiajanlo',
  25             'uploader_id': '83729',
  26             'timestamp': 1439193826,
  27             'upload_date': '20150810',
  28             'duration': 72,
  29             'age_limit': 0,
  30             'tags': ['tánc', 'cica', 'cuki', 'cukiajanlo', 'newsroom'],
  31         },
  32     }, {
  33         'url': 'http://embed.indavideo.hu/player/video/1bdc3c6d80?autostart=1&hide=1',
  34         'only_matching': True,
  35     }, {
  36         'url': 'http://assets.indavideo.hu/swf/player.swf?v=fe25e500&vID=1bdc3c6d80&autostart=1&hide=1&i=1',
  37         'only_matching': True,
  38     }]
  39
  40     def _real_extract(self, url):
  41         video_id = self._match_id(url)
  42
  43         video = self._download_json(
  44             'http://amfphp.indavideo.hu/SYm0json.php/player.playerHandler.getVideoData/%s' % video_id,
  45             video_id)['data']
  46
  47         video_id = video['id']
  48         title = video['title']
  49
  50         video_urls = video.get('video_files', [])
  51         video_file = video.get('video_file')
  52         if video:
  53             video_urls.append(video_file)
  54         video_urls = list(set(video_urls))
  55
  56         video_prefix = video_urls[0].rsplit('/', 1)[0]
  57
  58         for flv_file in video.get('flv_files', []):
  59             flv_url = '%s/%s' % (video_prefix, flv_file)
  60             if flv_url not in video_urls:
  61                 video_urls.append(flv_url)
  62
  63         formats = [{
  64             'url': video_url,
  65             'height': self._search_regex(r'\.(\d{3,4})\.mp4$', video_url, 'height', default=None),
  66         } for video_url in video_urls]
  67         self._sort_formats(formats)
  68
  69         timestamp = video.get('date')
  70         if timestamp:
  71             # upload date is in CEST
  72             timestamp = parse_iso8601(timestamp + ' +0200', ' ')
  73
  74         thumbnails = [{
  75             'url': self._proto_relative_url(thumbnail)
  76         } for thumbnail in video.get('thumbnails', [])]
  77
  78         tags = [tag['title'] for tag in video.get('tags', [])]
  79
  80         return {
  81             'id': video_id,
  82             'title': title,
  83             'description': video.get('description'),
  84             'thumbnails': thumbnails,
  85             'uploader': video.get('user_name'),
  86             'uploader_id': video.get('user_id'),
  87             'timestamp': timestamp,
  88             'duration': int_or_none(video.get('length')),
  89             'age_limit': parse_age_limit(video.get('age_limit')),
  90             'tags': tags,
  91             'formats': formats,
  92         }
  93
  94
  95 class IndavideoIE(InfoExtractor):
  96     _VALID_URL = r'https?://(?:www\.)?indavideo\.hu/video/(?P<id>[^/#?]+)'
  97     _TEST = {
  98         'url': 'http://indavideo.hu/video/Vicces_cica_1',
  99         'md5': '8c82244ba85d2a2310275b318eb51eac',
 100         'info_dict': {
 101             'id': '1335611',
 102             'display_id': 'Vicces_cica_1',
 103             'ext': 'mp4',
 104             'title': 'Vicces cica',
 105             'description': 'Játszik a tablettel. :D',
 106             'thumbnail': 're:^https?://.*\.jpg$',
 107             'uploader': 'Jet_Pack',
 108             'uploader_id': '491217',
 109             'timestamp': 1390821212,
 110             'upload_date': '20140127',
 111             'duration': 7,
 112             'age_limit': 0,
 113             'tags': ['vicces', 'macska', 'cica', 'ügyes', 'nevetés', 'játszik', 'Cukiság', 'Jet_Pack'],
 114         },
 115     }
 116
 117     def _real_extract(self, url):
 118         display_id = self._match_id(url)
 119
 120         webpage = self._download_webpage(url, display_id)
 121         embed_url = self._search_regex(
 122             r'<link[^>]+rel="video_src"[^>]+href="(.+?)"', webpage, 'embed url')
 123
 124         return {
 125             '_type': 'url_transparent',
 126             'ie_key': 'IndavideoEmbed',
 127             'url': embed_url,
 128             'display_id': display_id,
 129         }