yt_dlp/extractor/likee.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     int_or_none,
   6     js_to_json,
   7     parse_iso8601,
   8     str_or_none,
   9     traverse_obj,
  10 )
  11
  12
  13 class LikeeIE(InfoExtractor):
  14     IE_NAME = 'likee'
  15     _VALID_URL = r'(?x)https?://(www\.)?likee\.video/(?:(?P<channel_name>[^/]+)/video/|v/)(?P<id>\w+)'
  16     _TESTS = [{
  17         'url': 'https://likee.video/@huynh_hong_quan_/video/7093444807096327263',
  18         'info_dict': {
  19             'id': '7093444807096327263',
  20             'ext': 'mp4',
  21             'title': '🤴🤴🤴',
  22             'description': 'md5:9a7ebe816f0e78722ee5ed76f75983b4',
  23             'thumbnail': r're:^https?://.+\.jpg',
  24             'uploader': 'Huỳnh Hồng Qu&acirc;n ',
  25             'play_count': int,
  26             'download_count': int,
  27             'artist': 'Huỳnh Hồng Qu&acirc;n ',
  28             'timestamp': 1651571320,
  29             'upload_date': '20220503',
  30             'view_count': int,
  31             'uploader_id': 'huynh_hong_quan_',
  32             'duration': 12374,
  33             'comment_count': int,
  34             'like_count': int,
  35         },
  36     }, {
  37         'url': 'https://likee.video/@649222262/video/7093167848050058862',
  38         'info_dict': {
  39             'id': '7093167848050058862',
  40             'ext': 'mp4',
  41             'title': 'likee video #7093167848050058862',
  42             'description': 'md5:3f971c8c6ee8a216f2b1a9094c5de99f',
  43             'thumbnail': r're:^https?://.+\.jpg',
  44             'comment_count': int,
  45             'like_count': int,
  46             'uploader': 'Vương Phước Nhi',
  47             'download_count': int,
  48             'timestamp': 1651506835,
  49             'upload_date': '20220502',
  50             'duration': 60024,
  51             'play_count': int,
  52             'artist': 'Vương Phước Nhi',
  53             'uploader_id': '649222262',
  54             'view_count': int,
  55         },
  56     }, {
  57         'url': 'https://likee.video/@fernanda_rivasg/video/6932224568407629502',
  58         'info_dict': {
  59             'id': '6932224568407629502',
  60             'ext': 'mp4',
  61             'title': 'Un trend viejito🔥 #LIKEE #Ferlovers #trend ',
  62             'description': 'md5:c42b903a72a99d6d8b73e3d1126fbcef',
  63             'thumbnail': r're:^https?://.+\.jpg',
  64             'comment_count': int,
  65             'duration': 9684,
  66             'uploader_id': 'fernanda_rivasg',
  67             'view_count': int,
  68             'play_count': int,
  69             'artist': 'La Cami La✨',
  70             'download_count': int,
  71             'like_count': int,
  72             'uploader': 'Fernanda Rivas🎶',
  73             'timestamp': 1614034308,
  74             'upload_date': '20210222',
  75         },
  76     }, {
  77         'url': 'https://likee.video/v/k6QcOp',
  78         'info_dict': {
  79             'id': 'k6QcOp',
  80             'ext': 'mp4',
  81             'title': '#AguaChallenge t&uacute; ya lo intentaste?😱🤩',
  82             'description': 'md5:b0cc462689d4ff2b624daa4dba7640d9',
  83             'thumbnail': r're:^https?://.+\.jpg',
  84             'comment_count': int,
  85             'duration': 18014,
  86             'play_count': int,
  87             'view_count': int,
  88             'timestamp': 1611694774,
  89             'like_count': int,
  90             'uploader': 'Fernanda Rivas🎶',
  91             'uploader_id': 'fernanda_rivasg',
  92             'download_count': int,
  93             'artist': 'ʟᴇʀɪᴋ_ᴜɴɪᴄᴏʀɴ♡︎',
  94             'upload_date': '20210126',
  95         },
  96     }, {
  97         'url': 'https://www.likee.video/@649222262/video/7093167848050058862',
  98         'only_matching': True,
  99     }, {
 100         'url': 'https://www.likee.video/v/k6QcOp',
 101         'only_matching': True,
 102     }]
 103
 104     def _real_extract(self, url):
 105         video_id = self._match_id(url)
 106         webpage = self._download_webpage(url, video_id)
 107         info = self._parse_json(
 108             self._search_regex(r'window\.data\s=\s({.+?});', webpage, 'video info'),
 109             video_id, transform_source=js_to_json)
 110         video_url = traverse_obj(info, 'video_url', ('originVideoInfo', 'video_url'))
 111         if not video_url:
 112             self.raise_no_formats('Video was deleted', expected=True)
 113         formats = [{
 114             'format_id': 'mp4-with-watermark',
 115             'url': video_url,
 116             'height': info.get('video_height'),
 117             'width': info.get('video_width'),
 118         }, {
 119             'format_id': 'mp4-without-watermark',
 120             'url': video_url.replace('_4', ''),
 121             'height': info.get('video_height'),
 122             'width': info.get('video_width'),
 123             'quality': 1,
 124         }]
 125         return {
 126             'id': video_id,
 127             'title': info.get('msgText'),
 128             'description': info.get('share_desc'),
 129             'view_count': int_or_none(info.get('video_count')),
 130             'like_count': int_or_none(info.get('likeCount')),
 131             'play_count': int_or_none(info.get('play_count')),
 132             'download_count': int_or_none(info.get('download_count')),
 133             'comment_count': int_or_none(info.get('comment_count')),
 134             'uploader': str_or_none(info.get('nick_name')),
 135             'uploader_id': str_or_none(info.get('likeeId')),
 136             'artist': str_or_none(traverse_obj(info, ('sound', 'owner_name'))),
 137             'timestamp': parse_iso8601(info.get('uploadDate')),
 138             'thumbnail': info.get('coverUrl'),
 139             'duration': int_or_none(traverse_obj(info, ('option_data', 'dur'))),
 140             'formats': formats,
 141         }
 142
 143
 144 class LikeeUserIE(InfoExtractor):
 145     IE_NAME = 'likee:user'
 146     _VALID_URL = r'https?://(www\.)?likee\.video/(?P<id>[^/]+)/?$'
 147     _TESTS = [{
 148         'url': 'https://likee.video/@fernanda_rivasg',
 149         'info_dict': {
 150             'id': '925638334',
 151             'title': 'fernanda_rivasg',
 152         },
 153         'playlist_mincount': 500,
 154     }, {
 155         'url': 'https://likee.video/@may_hmoob',
 156         'info_dict': {
 157             'id': '2943949041',
 158             'title': 'may_hmoob',
 159         },
 160         'playlist_mincount': 80,
 161     }]
 162     _PAGE_SIZE = 50
 163     _API_GET_USER_VIDEO = 'https://api.like-video.com/likee-activity-flow-micro/videoApi/getUserVideo'
 164
 165     def _entries(self, user_name, user_id):
 166         last_post_id = ''
 167         while True:
 168             user_videos = self._download_json(
 169                 self._API_GET_USER_VIDEO, user_name,
 170                 data=json.dumps({
 171                     'uid': user_id,
 172                     'count': self._PAGE_SIZE,
 173                     'lastPostId': last_post_id,
 174                     'tabType': 0,
 175                 }).encode('utf-8'),
 176                 headers={'content-type': 'application/json'},
 177                 note=f'Get user info with lastPostId #{last_post_id}')
 178             items = traverse_obj(user_videos, ('data', 'videoList'))
 179             if not items:
 180                 break
 181             for item in items:
 182                 last_post_id = item['postId']
 183                 yield self.url_result(f'https://likee.video/{user_name}/video/{last_post_id}')
 184
 185     def _real_extract(self, url):
 186         user_name = self._match_id(url)
 187         webpage = self._download_webpage(url, user_name)
 188         info = self._parse_json(
 189             self._search_regex(r'window\.data\s*=\s*({.+?});', webpage, 'user info'),
 190             user_name, transform_source=js_to_json)
 191         user_id = traverse_obj(info, ('userinfo', 'uid'))
 192         return self.playlist_result(self._entries(user_name, user_id), user_id, traverse_obj(info, ('userinfo', 'user_name')))