_ 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': 'c8a507a1c7410685f83a06eaeeaafeab',
  19         'info_dict': {
  20             'id': '1837039',
  21             'ext': 'mp4',
  22             'title': 'Cicatánc',
  23             'description': '',
  24             'thumbnail': r're:^https?://.*\.jpg$',
  25             'uploader': 'cukiajanlo',
  26             'uploader_id': '83729',
  27             'timestamp': 1439193826,
  28             'upload_date': '20150810',
  29             'duration': 72,
  30             'age_limit': 0,
  31             'tags': ['tánc', 'cica', 'cuki', 'cukiajanlo', 'newsroom'],
  32         },
  33     }, {
  34         'url': 'http://embed.indavideo.hu/player/video/1bdc3c6d80?autostart=1&hide=1',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://assets.indavideo.hu/swf/player.swf?v=fe25e500&vID=1bdc3c6d80&autostart=1&hide=1&i=1',
  38         'only_matching': True,
  39     }]
  40
  41     def _real_extract(self, url):
  42         video_id = self._match_id(url)
  43
  44         video = self._download_json(
  45             'http://amfphp.indavideo.hu/SYm0json.php/player.playerHandler.getVideoData/%s' % video_id,
  46             video_id)['data']
  47
  48         title = video['title']
  49
  50         video_urls = []
  51
  52         video_files = video.get('video_files')
  53         if isinstance(video_files, list):
  54             video_urls.extend(video_files)
  55         elif isinstance(video_files, dict):
  56             video_urls.extend(video_files.values())
  57
  58         video_file = video.get('video_file')
  59         if video:
  60             video_urls.append(video_file)
  61         video_urls = list(set(video_urls))
  62
  63         video_prefix = video_urls[0].rsplit('/', 1)[0]
  64
  65         for flv_file in video.get('flv_files', []):
  66             flv_url = '%s/%s' % (video_prefix, flv_file)
  67             if flv_url not in video_urls:
  68                 video_urls.append(flv_url)
  69
  70         filesh = video.get('filesh')
  71
  72         formats = []
  73         for video_url in video_urls:
  74             height = int_or_none(self._search_regex(
  75                 r'\.(\d{3,4})\.mp4(?:\?|$)', video_url, 'height', default=None))
  76             if filesh:
  77                 if not height:
  78                     continue
  79                 token = filesh.get(compat_str(height))
  80                 if token is None:
  81                     continue
  82                 video_url = update_url_query(video_url, {'token': token})
  83             formats.append({
  84                 'url': video_url,
  85                 'height': height,
  86             })
  87         self._sort_formats(formats)
  88
  89         timestamp = video.get('date')
  90         if timestamp:
  91             # upload date is in CEST
  92             timestamp = parse_iso8601(timestamp + ' +0200', ' ')
  93
  94         thumbnails = [{
  95             'url': self._proto_relative_url(thumbnail)
  96         } for thumbnail in video.get('thumbnails', [])]
  97
  98         tags = [tag['title'] for tag in video.get('tags') or []]
  99
 100         return {
 101             'id': video.get('id') or video_id,
 102             'title': title,
 103             'description': video.get('description'),
 104             'thumbnails': thumbnails,
 105             'uploader': video.get('user_name'),
 106             'uploader_id': video.get('user_id'),
 107             'timestamp': timestamp,
 108             'duration': int_or_none(video.get('length')),
 109             'age_limit': parse_age_limit(video.get('age_limit')),
 110             'tags': tags,
 111             'formats': formats,
 112         }
 113
 114
 115 class IndavideoIE(InfoExtractor):
 116     _VALID_URL = r'https?://(?:.+?\.)?indavideo\.hu/video/(?P<id>[^/#?]+)'
 117     _TESTS = [{
 118         'url': 'http://indavideo.hu/video/Vicces_cica_1',
 119         'md5': '8c82244ba85d2a2310275b318eb51eac',
 120         'info_dict': {
 121             'id': '1335611',
 122             'display_id': 'Vicces_cica_1',
 123             'ext': 'mp4',
 124             'title': 'Vicces cica',
 125             'description': 'Játszik a tablettel. :D',
 126             'thumbnail': r're:^https?://.*\.jpg$',
 127             'uploader': 'Jet_Pack',
 128             'uploader_id': '491217',
 129             'timestamp': 1390821212,
 130             'upload_date': '20140127',
 131             'duration': 7,
 132             'age_limit': 0,
 133             'tags': list,
 134         },
 135     }, {
 136         'url': 'http://index.indavideo.hu/video/2015_0728_beregszasz',
 137         'only_matching': True,
 138     }, {
 139         'url': 'http://auto.indavideo.hu/video/Sajat_utanfutoban_a_kis_tacsko',
 140         'only_matching': True,
 141     }, {
 142         'url': 'http://erotika.indavideo.hu/video/Amator_tini_punci',
 143         'only_matching': True,
 144     }, {
 145         'url': 'http://film.indavideo.hu/video/f_hrom_nagymamm_volt',
 146         'only_matching': True,
 147     }, {
 148         'url': 'http://palyazat.indavideo.hu/video/Embertelen_dal_Dodgem_egyuttes',
 149         'only_matching': True,
 150     }]
 151
 152     def _real_extract(self, url):
 153         display_id = self._match_id(url)
 154
 155         webpage = self._download_webpage(url, display_id)
 156         embed_url = self._search_regex(
 157             (r'<iframe[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//embed\.indavideo\.hu/player/video/.+?)\1',
 158              r'<link[^>]+rel="video_src"[^>]+href="(?P<url>.+?)"'),
 159             webpage, 'embed url', group='url')
 160
 161         return {
 162             '_type': 'url_transparent',
 163             'ie_key': 'IndavideoEmbed',
 164             'url': embed_url,
 165             'display_id': display_id,
 166         }