yt_dlp/extractor/nekohacker.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     determine_ext,
   7     extract_attributes,
   8     get_element_by_class,
   9     parse_duration,
  10     url_or_none,
  11 )
  12 from ..utils.traversal import find_element, traverse_obj
  13
  14
  15 class NekoHackerIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?nekohacker\.com/(?P<id>(?!free-dl)[\w-]+)'
  17     _TESTS = [{
  18         'url': 'https://nekohacker.com/nekoverse/',
  19         'info_dict': {
  20             'id': 'nekoverse',
  21             'title': 'Nekoverse',
  22         },
  23         'playlist': [
  24             {
  25                 'url': 'https://nekohacker.com/wp-content/uploads/2022/11/01-Spaceship.mp3',
  26                 'md5': '44223701ebedba0467ebda4cc07fb3aa',
  27                 'info_dict': {
  28                     'id': '1712',
  29                     'ext': 'mp3',
  30                     'title': 'Spaceship',
  31                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
  32                     'vcodec': 'none',
  33                     'acodec': 'mp3',
  34                     'release_date': '20221101',
  35                     'album': 'Nekoverse',
  36                     'artists': ['Neko Hacker'],
  37                     'track': 'Spaceship',
  38                     'track_number': 1,
  39                     'duration': 195.0,
  40                 },
  41             },
  42             {
  43                 'url': 'https://nekohacker.com/wp-content/uploads/2022/11/02-City-Runner.mp3',
  44                 'md5': '8f853c71719389d32bbbd3f1a87b3f08',
  45                 'info_dict': {
  46                     'id': '1713',
  47                     'ext': 'mp3',
  48                     'title': 'City Runner',
  49                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
  50                     'vcodec': 'none',
  51                     'acodec': 'mp3',
  52                     'release_date': '20221101',
  53                     'album': 'Nekoverse',
  54                     'artists': ['Neko Hacker'],
  55                     'track': 'City Runner',
  56                     'track_number': 2,
  57                     'duration': 148.0,
  58                 },
  59             },
  60             {
  61                 'url': 'https://nekohacker.com/wp-content/uploads/2022/11/03-Nature-Talk.mp3',
  62                 'md5': '5a8a8ae852720cee4c0ac95c7d1a7450',
  63                 'info_dict': {
  64                     'id': '1714',
  65                     'ext': 'mp3',
  66                     'title': 'Nature Talk',
  67                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
  68                     'vcodec': 'none',
  69                     'acodec': 'mp3',
  70                     'release_date': '20221101',
  71                     'album': 'Nekoverse',
  72                     'artists': ['Neko Hacker'],
  73                     'track': 'Nature Talk',
  74                     'track_number': 3,
  75                     'duration': 174.0,
  76                 },
  77             },
  78             {
  79                 'url': 'https://nekohacker.com/wp-content/uploads/2022/11/04-Crystal-World.mp3',
  80                 'md5': 'd8e59a48061764e50d92386a294abd50',
  81                 'info_dict': {
  82                     'id': '1715',
  83                     'ext': 'mp3',
  84                     'title': 'Crystal World',
  85                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
  86                     'vcodec': 'none',
  87                     'acodec': 'mp3',
  88                     'release_date': '20221101',
  89                     'album': 'Nekoverse',
  90                     'artists': ['Neko Hacker'],
  91                     'track': 'Crystal World',
  92                     'track_number': 4,
  93                     'duration': 199.0,
  94                 },
  95             },
  96         ],
  97     }, {
  98         'url': 'https://nekohacker.com/susume/',
  99         'info_dict': {
 100             'id': 'susume',
 101             'title': '進め！むじなカンパニー',
 102         },
 103         'playlist': [
 104             {
 105                 'url': 'https://nekohacker.com/wp-content/uploads/2021/01/進め！むじなカンパニー-feat.-六科なじむ-CV_-日高里菜-割戶真友-CV_-金元寿子-軽井沢ユキ-CV_-上坂すみれ-出稼ぎガルシア-CV_-金子彩花-.mp3',
 106                 'md5': 'fb13f008aa81f26ba48f91fd2d6186ce',
 107                 'info_dict': {
 108                     'id': '711',
 109                     'ext': 'mp3',
 110                     'title': 'md5:1a5fcbc96ca3c3265b1c6f9f79f30fd0',
 111                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
 112                     'vcodec': 'none',
 113                     'acodec': 'mp3',
 114                     'release_date': '20210115',
 115                     'album': '進め！むじなカンパニー',
 116                     'artists': ['Neko Hacker'],
 117                     'track': 'md5:1a5fcbc96ca3c3265b1c6f9f79f30fd0',
 118                     'track_number': 1,
 119                 },
 120             },
 121             {
 122                 'url': 'https://nekohacker.com/wp-content/uploads/2021/01/むじな-de-なじむ-feat.-六科なじむ-CV_-日高里菜-.mp3',
 123                 'md5': '028803f70241df512b7764e73396fdd1',
 124                 'info_dict': {
 125                     'id': '709',
 126                     'ext': 'mp3',
 127                     'title': 'むじな de なじむ feat. 六科なじむ (CV: 日高里菜 )',
 128                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
 129                     'vcodec': 'none',
 130                     'acodec': 'mp3',
 131                     'release_date': '20210115',
 132                     'album': '進め！むじなカンパニー',
 133                     'artists': ['Neko Hacker'],
 134                     'track': 'むじな de なじむ feat. 六科なじむ (CV: 日高里菜 )',
 135                     'track_number': 2,
 136                 },
 137             },
 138             {
 139                 'url': 'https://nekohacker.com/wp-content/uploads/2021/01/進め！むじなカンパニー-instrumental.mp3',
 140                 'md5': 'adde9e9a16e1da5e602b579c247d0fb9',
 141                 'info_dict': {
 142                     'id': '710',
 143                     'ext': 'mp3',
 144                     'title': '進め！むじなカンパニー (instrumental)',
 145                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
 146                     'vcodec': 'none',
 147                     'acodec': 'mp3',
 148                     'release_date': '20210115',
 149                     'album': '進め！むじなカンパニー',
 150                     'artists': ['Neko Hacker'],
 151                     'track': '進め！むじなカンパニー (instrumental)',
 152                     'track_number': 3,
 153                 },
 154             },
 155             {
 156                 'url': 'https://nekohacker.com/wp-content/uploads/2021/01/むじな-de-なじむ-instrumental.mp3',
 157                 'md5': 'ebb0443039cf5f9ff7fd557ed9b23599',
 158                 'info_dict': {
 159                     'id': '712',
 160                     'ext': 'mp3',
 161                     'title': 'むじな de なじむ (instrumental)',
 162                     'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
 163                     'vcodec': 'none',
 164                     'acodec': 'mp3',
 165                     'release_date': '20210115',
 166                     'album': '進め！むじなカンパニー',
 167                     'artists': ['Neko Hacker'],
 168                     'track': 'むじな de なじむ (instrumental)',
 169                     'track_number': 4,
 170                 },
 171             },
 172         ],
 173     }]
 174
 175     def _real_extract(self, url):
 176         playlist_id = self._match_id(url)
 177
 178         webpage = self._download_webpage(url, playlist_id)
 179         playlist = get_element_by_class('playlist', webpage)
 180
 181         if not playlist:
 182             iframe_src = traverse_obj(webpage, (
 183                 {find_element(tag='iframe', html=True)}, {extract_attributes}, 'src', {url_or_none}))
 184             if not iframe_src:
 185                 raise ExtractorError('No playlist or embed found in webpage')
 186             elif re.match(r'https?://(?:\w+\.)?spotify\.com/', iframe_src):
 187                 raise ExtractorError('Spotify embeds are not supported', expected=True)
 188             return self.url_result(url, 'Generic')
 189
 190         player_params = self._search_json(
 191             r'var srp_player_params_[\da-f]+\s*=', webpage, 'player params', playlist_id, default={})
 192
 193         entries = []
 194         for track_number, track in enumerate(re.findall(r'(<li[^>]+data-audiopath[^>]+>)', playlist), 1):
 195             entry = traverse_obj(extract_attributes(track), {
 196                 'url': ('data-audiopath', {url_or_none}),
 197                 'ext': ('data-audiopath', {determine_ext}),
 198                 'id': 'data-trackid',
 199                 'title': 'data-tracktitle',
 200                 'track': 'data-tracktitle',
 201                 'album': 'data-albumtitle',
 202                 'duration': ('data-tracktime', {parse_duration}),
 203                 'release_date': ('data-releasedate', {lambda x: re.match(r'\d{8}', x.replace('.', ''))}, 0),
 204             })
 205             entries.append({
 206                 **entry,
 207                 'thumbnail': url_or_none(player_params.get('artwork')),
 208                 'track_number': track_number,
 209                 'artists': ['Neko Hacker'],
 210                 'vcodec': 'none',
 211                 'acodec': 'mp3' if entry['ext'] == 'mp3' else None,
 212             })
 213
 214         return self.playlist_result(entries, playlist_id, traverse_obj(entries, (0, 'album')))