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