]>
Commit | Line | Data |
---|---|---|
db40364b MTP |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
55cde6ef | 3 | |
db40364b | 4 | from .common import InfoExtractor |
ee94e7e6 YCH |
5 | from ..utils import ( |
6 | clean_html, | |
4dfbf869 | 7 | parse_qs, |
ee94e7e6 YCH |
8 | remove_start, |
9 | ) | |
db40364b MTP |
10 | |
11 | ||
12 | class Varzesh3IE(InfoExtractor): | |
55cde6ef | 13 | _VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?' |
ee94e7e6 | 14 | _TESTS = [{ |
db40364b MTP |
15 | '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/', |
16 | 'md5': '2a933874cb7dce4366075281eb49e855', | |
17 | 'info_dict': { | |
db40364b MTP |
18 | 'id': '76337', |
19 | 'ext': 'mp4', | |
8896b614 | 20 | 'title': '۵ واکنش برتر دروازهبانان؛هفته ۲۶ بوندسلیگا', |
2315fb5e | 21 | 'description': 'فصل ۲۰۱۵-۲۰۱۴', |
ec85ded8 | 22 | 'thumbnail': r're:^https?://.*\.jpg$', |
ee94e7e6 YCH |
23 | }, |
24 | 'skip': 'HTTP 404 Error', | |
25 | }, { | |
26 | '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', | |
00b426d6 | 27 | 'md5': '841b7cd3afbc76e61708d94e53a4a4e7', |
ee94e7e6 YCH |
28 | 'info_dict': { |
29 | 'id': '112785', | |
30 | 'ext': 'mp4', | |
31 | 'title': 'دله علی؛ ستاره نوظهور لیگ برتر جزیره', | |
32 | 'description': 'فوتبال 120', | |
33 | }, | |
34 | 'expected_warnings': ['description'], | |
35 | }] | |
db40364b MTP |
36 | |
37 | def _real_extract(self, url): | |
55cde6ef S |
38 | display_id = self._match_id(url) |
39 | ||
40 | webpage = self._download_webpage(url, display_id) | |
41 | ||
42 | video_url = self._search_regex( | |
43 | r'<source[^>]+src="([^"]+)"', webpage, 'video url') | |
db40364b | 44 | |
ee94e7e6 YCH |
45 | title = remove_start(self._html_search_regex( |
46 | r'<title>([^<]+)</title>', webpage, 'title'), 'ویدیو ورزش 3 | ') | |
47 | ||
55cde6ef S |
48 | description = self._html_search_regex( |
49 | r'(?s)<div class="matn">(.+?)</div>', | |
ee94e7e6 YCH |
50 | webpage, 'description', default=None) |
51 | if description is None: | |
52 | description = clean_html(self._html_search_meta('description', webpage)) | |
53 | ||
54 | thumbnail = self._og_search_thumbnail(webpage, default=None) | |
55 | if thumbnail is None: | |
56 | fb_sharer_url = self._search_regex( | |
57 | r'<a[^>]+href="(https?://www\.facebook\.com/sharer/sharer\.php?[^"]+)"', | |
58 | webpage, 'facebook sharer URL', fatal=False) | |
4dfbf869 | 59 | sharer_params = parse_qs(fb_sharer_url) |
ee94e7e6 | 60 | thumbnail = sharer_params.get('p[images][0]', [None])[0] |
db40364b | 61 | |
55cde6ef S |
62 | video_id = self._search_regex( |
63 | r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'", | |
ee94e7e6 YCH |
64 | webpage, display_id, default=None) |
65 | if video_id is None: | |
66 | video_id = self._search_regex( | |
ec85ded8 | 67 | r'var\s+VideoId\s*=\s*(\d+);', webpage, 'video id', |
ee94e7e6 | 68 | default=display_id) |
db40364b MTP |
69 | |
70 | return { | |
55cde6ef S |
71 | 'url': video_url, |
72 | 'id': video_id, | |
db40364b | 73 | 'title': title, |
db40364b MTP |
74 | 'description': description, |
75 | 'thumbnail': thumbnail, | |
76 | } |