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