2 from __future__ import unicode_literals
4 from .common import InfoExtractor
6 compat_urllib_parse_urlparse,
15 class Varzesh3IE(InfoExtractor):
16 _VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?'
18 '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/',
19 'md5': '2a933874cb7dce4366075281eb49e855',
23 'title': '۵ واکنش برتر دروازهبانان؛هفته ۲۶ بوندسلیگا',
24 'description': 'فصل ۲۰۱۵-۲۰۱۴',
25 'thumbnail': 're:^https?://.*\.jpg$',
27 'skip': 'HTTP 404 Error',
29 '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',
33 'title': 'دله علی؛ ستاره نوظهور لیگ برتر جزیره',
34 'description': 'فوتبال 120',
36 'expected_warnings': ['description'],
39 def _real_extract(self, url):
40 display_id = self._match_id(url)
42 webpage = self._download_webpage(url, display_id)
44 video_url = self._search_regex(
45 r'<source[^>]+src="([^"]+)"', webpage, 'video url')
47 title = remove_start(self._html_search_regex(
48 r'<title>([^<]+)</title>', webpage, 'title'), 'ویدیو ورزش 3 | ')
50 description = self._html_search_regex(
51 r'(?s)<div class="matn">(.+?)</div>',
52 webpage, 'description', default=None)
53 if description is None:
54 description = clean_html(self._html_search_meta('description', webpage))
56 thumbnail = self._og_search_thumbnail(webpage, default=None)
58 fb_sharer_url = self._search_regex(
59 r'<a[^>]+href="(https?://www\.facebook\.com/sharer/sharer\.php?[^"]+)"',
60 webpage, 'facebook sharer URL', fatal=False)
61 sharer_params = compat_parse_qs(compat_urllib_parse_urlparse(fb_sharer_url).query)
62 thumbnail = sharer_params.get('p[images][0]', [None])[0]
64 video_id = self._search_regex(
65 r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'",
66 webpage, display_id, default=None)
68 video_id = self._search_regex(
69 'var\s+VideoId\s*=\s*(\d+);', webpage, 'video id',
76 'description': description,
77 'thumbnail': thumbnail,