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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_str
   6 from ..utils import (
   7     int_or_none,
   8     parse_age_limit,
   9     parse_iso8601,
  10     update_url_query,
  11 )
  12
  13
  14 class IndavideoEmbedIE(InfoExtractor):
  15     _VALID_URL = r'https?://(?:(?:embed\.)?indavideo\.hu/player/video/|assets\.indavideo\.hu/swf/player\.swf\?.*\b(?:v(?:ID|id))=)(?P<id>[\da-f]+)'
  16     _TESTS = [{
  17         'url': 'http://indavideo.hu/player/video/1bdc3c6d80/',
  18         'md5': 'f79b009c66194acacd40712a6778acfa',
  19         'info_dict': {
  20             'id': '1837039',
  21             'ext': 'mp4',
  22             'title': 'Cicatánc',
  23             'description': '',
  24             'thumbnail': r're:^https?://.*\.jpg$',
  25             'uploader': 'cukiajanlo',
  26             'uploader_id': '83729',
  27             'timestamp': 1439193826,
  28             'upload_date': '20150810',
  29             'duration': 72,
  30             'age_limit': 0,
  31             'tags': ['tánc', 'cica', 'cuki', 'cukiajanlo', 'newsroom'],
  32         },
  33     }, {
  34         'url': 'http://embed.indavideo.hu/player/video/1bdc3c6d80?autostart=1&hide=1',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://assets.indavideo.hu/swf/player.swf?v=fe25e500&vID=1bdc3c6d80&autostart=1&hide=1&i=1',
  38         'only_matching': True,
  39     }]
  40
  41     def _real_extract(self, url):
  42         video_id = self._match_id(url)
  43
  44         video = self._download_json(
  45             'http://amfphp.indavideo.hu/SYm0json.php/player.playerHandler.getVideoData/%s' % video_id,
  46             video_id)['data']
  47
  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         filesh = video.get('filesh')
  64         formats = [
  65             self.video_url_to_format(video_url, filesh)
  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') or []]
  79
  80         return {
  81             'id': video.get('id') or 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     def video_url_to_format(self, video_url, filesh):
  95         height = int_or_none(self._search_regex(
  96             r'\.(\d{3,4})\.mp4(?:\?|$)', video_url, 'height', default=None))
  97         if height and filesh:
  98             token = filesh.get(compat_str(height))
  99             if token is not None:
 100                 video_url = update_url_query(video_url, {'token': token})
 101         return {
 102             'url': video_url,
 103             'height': height,
 104         }
 105
 106
 107 class IndavideoIE(InfoExtractor):
 108     _VALID_URL = r'https?://(?:.+?\.)?indavideo\.hu/video/(?P<id>[^/#?]+)'
 109     _TESTS = [{
 110         'url': 'http://indavideo.hu/video/Vicces_cica_1',
 111         'md5': '8c82244ba85d2a2310275b318eb51eac',
 112         'info_dict': {
 113             'id': '1335611',
 114             'display_id': 'Vicces_cica_1',
 115             'ext': 'mp4',
 116             'title': 'Vicces cica',
 117             'description': 'Játszik a tablettel. :D',
 118             'thumbnail': r're:^https?://.*\.jpg$',
 119             'uploader': 'Jet_Pack',
 120             'uploader_id': '491217',
 121             'timestamp': 1390821212,
 122             'upload_date': '20140127',
 123             'duration': 7,
 124             'age_limit': 0,
 125             'tags': ['vicces', 'macska', 'cica', 'ügyes', 'nevetés', 'játszik', 'Cukiság', 'Jet_Pack'],
 126         },
 127     }, {
 128         'url': 'http://index.indavideo.hu/video/2015_0728_beregszasz',
 129         'only_matching': True,
 130     }, {
 131         'url': 'http://auto.indavideo.hu/video/Sajat_utanfutoban_a_kis_tacsko',
 132         'only_matching': True,
 133     }, {
 134         'url': 'http://erotika.indavideo.hu/video/Amator_tini_punci',
 135         'only_matching': True,
 136     }, {
 137         'url': 'http://film.indavideo.hu/video/f_hrom_nagymamm_volt',
 138         'only_matching': True,
 139     }, {
 140         'url': 'http://palyazat.indavideo.hu/video/Embertelen_dal_Dodgem_egyuttes',
 141         'only_matching': True,
 142     }]
 143
 144     def _real_extract(self, url):
 145         display_id = self._match_id(url)
 146
 147         webpage = self._download_webpage(url, display_id)
 148         embed_url = self._search_regex(
 149             r'<link[^>]+rel="video_src"[^>]+href="(.+?)"', webpage, 'embed url')
 150
 151         return {
 152             '_type': 'url_transparent',
 153             'ie_key': 'IndavideoEmbed',
 154             'url': embed_url,
 155             'display_id': display_id,
 156         }