yt_dlp/extractor/myvideoge.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     MONTH_NAMES,
   6     clean_html,
   7     get_element_by_class,
   8     get_element_by_id,
   9     int_or_none,
  10     js_to_json,
  11     qualities,
  12     unified_strdate,
  13 )
  14
  15
  16 class MyVideoGeIE(InfoExtractor):
  17     _VALID_URL = r'https?://(?:www\.)?myvideo\.ge/v/(?P<id>[0-9]+)'
  18     _TEST = {
  19         'url': 'https://www.myvideo.ge/v/3941048',
  20         'md5': '8c192a7d2b15454ba4f29dc9c9a52ea9',
  21         'info_dict': {
  22             'id': '3941048',
  23             'ext': 'mp4',
  24             'title': 'The best prikol',
  25             'upload_date': '20200611',
  26             'thumbnail': r're:^https?://.*\.jpg$',
  27             'uploader': 'chixa33',
  28             'description': 'md5:5b067801318e33c2e6eea4ab90b1fdd3',
  29         },
  30     }
  31     _MONTH_NAMES_KA = ['იანვარი', 'თებერვალი', 'მარტი', 'აპრილი', 'მაისი', 'ივნისი', 'ივლისი', 'აგვისტო', 'სექტემბერი', 'ოქტომბერი', 'ნოემბერი', 'დეკემბერი']
  32
  33     _quality = staticmethod(qualities(('SD', 'HD')))
  34
  35     def _real_extract(self, url):
  36         video_id = self._match_id(url)
  37         webpage = self._download_webpage(url, video_id)
  38
  39         title = (
  40             self._og_search_title(webpage, default=None)
  41             or clean_html(get_element_by_class('my_video_title', webpage))
  42             or self._html_search_regex(r'<title\b[^>]*>([^<]+)</title\b', webpage, 'title'))
  43
  44         jwplayer_sources = self._parse_json(
  45             self._search_regex(
  46                 r'''(?s)jwplayer\s*\(\s*['"]mvplayer['"]\s*\)\s*\.\s*setup\s*\(.*?\bsources\s*:\s*(\[.*?])\s*[,});]''', webpage, 'jwplayer sources', fatal=False)
  47             or '',
  48             video_id, transform_source=js_to_json, fatal=False)
  49
  50         formats = self._parse_jwplayer_formats(jwplayer_sources or [], video_id)
  51         for f in formats or []:
  52             f['quality'] = self._quality(f['format_id'])
  53
  54         description = (
  55             self._og_search_description(webpage)
  56             or get_element_by_id('long_desc_holder', webpage)
  57             or self._html_search_meta('description', webpage))
  58
  59         uploader = self._search_regex(r'<a[^>]+class="mv_user_name"[^>]*>([^<]+)<', webpage, 'uploader', fatal=False)
  60
  61         upload_date = get_element_by_class('mv_vid_upl_date', webpage)
  62         # as ka locale may not be present roll a local date conversion
  63         upload_date = (unified_strdate(
  64             # translate any ka month to an en one
  65             re.sub('|'.join(self._MONTH_NAMES_KA),
  66                    lambda m: MONTH_NAMES['en'][self._MONTH_NAMES_KA.index(m.group(0))],
  67                    upload_date, flags=re.I))
  68             if upload_date else None)
  69
  70         return {
  71             'id': video_id,
  72             'title': title,
  73             'description': description,
  74             'uploader': uploader,
  75             'formats': formats,
  76             'thumbnail': self._og_search_thumbnail(webpage),
  77             'upload_date': upload_date,
  78             'view_count': int_or_none(get_element_by_class('mv_vid_views', webpage)),
  79             'like_count': int_or_none(get_element_by_id('likes_count', webpage)),
  80             'dislike_count': int_or_none(get_element_by_id('dislikes_count', webpage)),
  81         }