yt_dlp/extractor/varzesh3.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     clean_html,
   7     parse_qs,
   8     remove_start,
   9 )
  10
  11
  12 class Varzesh3IE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?'
  14     _TESTS = [{
  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': {
  18             'id': '76337',
  19             'ext': 'mp4',
  20             'title': '۵ واکنش برتر دروازه‌بانان؛هفته ۲۶ بوندسلیگا',
  21             'description': 'فصل ۲۰۱۵-۲۰۱۴',
  22             'thumbnail': r're:^https?://.*\.jpg$',
  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',
  27         'md5': '841b7cd3afbc76e61708d94e53a4a4e7',
  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     }]
  36
  37     def _real_extract(self, url):
  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')
  44
  45         title = remove_start(self._html_extract_title(webpage), 'ویدیو ورزش 3 | ')
  46
  47         description = self._html_search_regex(
  48             r'(?s)<div class="matn">(.+?)</div>',
  49             webpage, 'description', default=None)
  50         if description is None:
  51             description = clean_html(self._html_search_meta('description', webpage))
  52
  53         thumbnail = self._og_search_thumbnail(webpage, default=None)
  54         if thumbnail is None:
  55             fb_sharer_url = self._search_regex(
  56                 r'<a[^>]+href="(https?://www\.facebook\.com/sharer/sharer\.php?[^"]+)"',
  57                 webpage, 'facebook sharer URL', fatal=False)
  58             sharer_params = parse_qs(fb_sharer_url)
  59             thumbnail = sharer_params.get('p[images][0]', [None])[0]
  60
  61         video_id = self._search_regex(
  62             r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'",
  63             webpage, display_id, default=None)
  64         if video_id is None:
  65             video_id = self._search_regex(
  66                 r'var\s+VideoId\s*=\s*(\d+);', webpage, 'video id',
  67                 default=display_id)
  68
  69         return {
  70             'url': video_url,
  71             'id': video_id,
  72             'title': title,
  73             'description': description,
  74             'thumbnail': thumbnail,
  75         }