yt_dlp/extractor/varzesh3.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     clean_html,
   4     parse_qs,
   5     remove_start,
   6 )
   7
   8
   9 class Varzesh3IE(InfoExtractor):
  10     _WORKING = False
  11     _VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?'
  12     _TESTS = [{
  13         'url': 'http://video.varzesh3.com/germany/bundesliga/5-%D9%88%D8%A7%DA%A9%D9%86%D8%B4-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AF%D8%B1%D9%88%D8%A7%D8%B2%D9%87%E2%80%8C%D8%A8%D8%A7%D9%86%D8%A7%D9%86%D8%9B%D9%87%D9%81%D8%AA%D9%87-26-%D8%A8%D9%88%D9%86%D8%AF%D8%B3/',
  14         'md5': '2a933874cb7dce4366075281eb49e855',
  15         'info_dict': {
  16             'id': '76337',
  17             'ext': 'mp4',
  18             'title': '۵ واکنش برتر دروازه‌بانان؛هفته ۲۶ بوندسلیگا',
  19             'description': 'فصل ۲۰۱۵-۲۰۱۴',
  20             'thumbnail': r're:^https?://.*\.jpg$',
  21         },
  22         'skip': 'HTTP 404 Error',
  23     }, {
  24         'url': 'http://video.varzesh3.com/video/112785/%D8%AF%D9%84%D9%87-%D8%B9%D9%84%DB%8C%D8%9B-%D8%B3%D8%AA%D8%A7%D8%B1%D9%87-%D9%86%D9%88%D8%B8%D9%87%D9%88%D8%B1-%D9%84%DB%8C%DA%AF-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AC%D8%B2%DB%8C%D8%B1%D9%87',
  25         'md5': '841b7cd3afbc76e61708d94e53a4a4e7',
  26         'info_dict': {
  27             'id': '112785',
  28             'ext': 'mp4',
  29             'title': 'دله علی؛ ستاره نوظهور لیگ برتر جزیره',
  30             'description': 'فوتبال 120',
  31         },
  32         'expected_warnings': ['description'],
  33     }]
  34
  35     def _real_extract(self, url):
  36         display_id = self._match_id(url)
  37
  38         webpage = self._download_webpage(url, display_id)
  39
  40         video_url = self._search_regex(
  41             r'<source[^>]+src="([^"]+)"', webpage, 'video url')
  42
  43         title = remove_start(self._html_extract_title(webpage), 'ویدیو ورزش 3 | ')
  44
  45         description = self._html_search_regex(
  46             r'(?s)<div class="matn">(.+?)</div>',
  47             webpage, 'description', default=None)
  48         if description is None:
  49             description = clean_html(self._html_search_meta('description', webpage))
  50
  51         thumbnail = self._og_search_thumbnail(webpage, default=None)
  52         if thumbnail is None:
  53             fb_sharer_url = self._search_regex(
  54                 r'<a[^>]+href="(https?://www\.facebook\.com/sharer/sharer\.php?[^"]+)"',
  55                 webpage, 'facebook sharer URL', fatal=False)
  56             sharer_params = parse_qs(fb_sharer_url)
  57             thumbnail = sharer_params.get('p[images][0]', [None])[0]
  58
  59         video_id = self._search_regex(
  60             r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'",
  61             webpage, display_id, default=None)
  62         if video_id is None:
  63             video_id = self._search_regex(
  64                 r'var\s+VideoId\s*=\s*(\d+);', webpage, 'video id',
  65                 default=display_id)
  66
  67         return {
  68             'url': video_url,
  69             'id': video_id,
  70             'title': title,
  71             'description': description,
  72             'thumbnail': thumbnail,
  73         }