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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import int_or_none
   8
   9
  10 class FiveTVIE(InfoExtractor):
  11     _VALID_URL = r'''(?x)
  12                     http://
  13                         (?:www\.)?5-tv\.ru/
  14                         (?:
  15                             (?:[^/]+/)+(?P<id>\d+)|
  16                             (?P<path>[^/?#]+)(?:[/?#])?
  17                         )
  18                     '''
  19
  20     _TESTS = [{
  21         'url': 'http://5-tv.ru/news/96814/',
  22         'md5': 'bbff554ad415ecf5416a2f48c22d9283',
  23         'info_dict': {
  24             'id': '96814',
  25             'ext': 'mp4',
  26             'title': 'Россияне выбрали имя для общенациональной платежной системы',
  27             'description': 'md5:a8aa13e2b7ad36789e9f77a74b6de660',
  28             'thumbnail': r're:^https?://.*\.jpg$',
  29             'duration': 180,
  30         },
  31     }, {
  32         'url': 'http://5-tv.ru/video/1021729/',
  33         'info_dict': {
  34             'id': '1021729',
  35             'ext': 'mp4',
  36             'title': '3D принтер',
  37             'description': 'md5:d76c736d29ef7ec5c0cf7d7c65ffcb41',
  38             'thumbnail': r're:^https?://.*\.jpg$',
  39             'duration': 180,
  40         },
  41     }, {
  42         'url': 'http://www.5-tv.ru/glavnoe/#itemDetails',
  43         'info_dict': {
  44             'id': 'glavnoe',
  45             'ext': 'mp4',
  46             'title': r're:^Итоги недели с \d+ по \d+ \w+ \d{4} года$',
  47             'thumbnail': r're:^https?://.*\.jpg$',
  48         },
  49     }, {
  50         'url': 'http://www.5-tv.ru/glavnoe/broadcasts/508645/',
  51         'only_matching': True,
  52     }, {
  53         'url': 'http://5-tv.ru/films/1507502/',
  54         'only_matching': True,
  55     }, {
  56         'url': 'http://5-tv.ru/programs/broadcast/508713/',
  57         'only_matching': True,
  58     }, {
  59         'url': 'http://5-tv.ru/angel/',
  60         'only_matching': True,
  61     }, {
  62         'url': 'http://www.5-tv.ru/schedule/?iframe=true&width=900&height=450',
  63         'only_matching': True,
  64     }]
  65
  66     def _real_extract(self, url):
  67         mobj = re.match(self._VALID_URL, url)
  68         video_id = mobj.group('id') or mobj.group('path')
  69
  70         webpage = self._download_webpage(url, video_id)
  71
  72         video_url = self._search_regex(
  73             [r'<div[^>]+?class="flowplayer[^>]+?data-href="([^"]+)"',
  74              r'<a[^>]+?href="([^"]+)"[^>]+?class="videoplayer"'],
  75             webpage, 'video url')
  76
  77         title = self._og_search_title(webpage, default=None) or self._search_regex(
  78             r'<title>([^<]+)</title>', webpage, 'title')
  79         duration = int_or_none(self._og_search_property(
  80             'video:duration', webpage, 'duration', default=None))
  81
  82         return {
  83             'id': video_id,
  84             'url': video_url,
  85             'title': title,
  86             'description': self._og_search_description(webpage, default=None),
  87             'thumbnail': self._og_search_thumbnail(webpage, default=None),
  88             'duration': duration,
  89         }