yt_dlp/extractor/mx3.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..networking import HEADRequest
   5 from ..utils import (
   6     get_element_by_class,
   7     int_or_none,
   8     try_call,
   9     url_or_none,
  10     urlhandle_detect_ext,
  11 )
  12 from ..utils.traversal import traverse_obj
  13
  14
  15 class Mx3BaseIE(InfoExtractor):
  16     _VALID_URL_TMPL = r'https?://(?:www\.)?%s/t/(?P<id>\w+)'
  17     _FORMATS = [{
  18         'url': 'player_asset',
  19         'format_id': 'default',
  20         'quality': 0,
  21     }, {
  22         'url': 'player_asset?quality=hd',
  23         'format_id': 'hd',
  24         'quality': 1,
  25     }, {
  26         'url': 'download',
  27         'format_id': 'download',
  28         'quality': 2,
  29     }, {
  30         'url': 'player_asset?quality=source',
  31         'format_id': 'source',
  32         'quality': 2,
  33     }]
  34
  35     def _extract_formats(self, track_id):
  36         formats = []
  37         for fmt in self._FORMATS:
  38             format_url = f'https://{self._DOMAIN}/tracks/{track_id}/{fmt["url"]}'
  39             urlh = self._request_webpage(
  40                 HEADRequest(format_url), track_id, fatal=False, expected_status=404,
  41                 note=f'Checking for format {fmt["format_id"]}')
  42             if urlh and urlh.status == 200:
  43                 formats.append({
  44                     **fmt,
  45                     'url': format_url,
  46                     'ext': urlhandle_detect_ext(urlh),
  47                     'filesize': int_or_none(urlh.headers.get('Content-Length')),
  48                 })
  49         return formats
  50
  51     def _real_extract(self, url):
  52         track_id = self._match_id(url)
  53         webpage = self._download_webpage(url, track_id)
  54         more_info = get_element_by_class('single-more-info', webpage)
  55         data = self._download_json(f'https://{self._DOMAIN}/t/{track_id}.json', track_id, fatal=False)
  56
  57         def get_info_field(name):
  58             return self._html_search_regex(
  59                 rf'<dt[^>]*>\s*{name}\s*</dt>\s*<dd[^>]*>(.*?)</dd>',
  60                 more_info, name, default=None, flags=re.DOTALL)
  61
  62         return {
  63             'id': track_id,
  64             'formats': self._extract_formats(track_id),
  65             'genre': self._html_search_regex(
  66                 r'<div\b[^>]+class="single-band-genre"[^>]*>([^<]+)</div>', webpage, 'genre', default=None),
  67             'release_year': int_or_none(get_info_field('Year of creation')),
  68             'description': get_info_field('Description'),
  69             'tags': try_call(lambda: get_info_field('Tag').split(', '), list),
  70             **traverse_obj(data, {
  71                 'title': ('title', {str}),
  72                 'artist': (('performer_name', 'artist'), {str}),
  73                 'album_artist': ('artist', {str}),
  74                 'composer': ('composer_name', {str}),
  75                 'thumbnail': (('picture_url_xlarge', 'picture_url'), {url_or_none}),
  76             }, get_all=False),
  77         }
  78
  79
  80 class Mx3IE(Mx3BaseIE):
  81     _DOMAIN = 'mx3.ch'
  82     _VALID_URL = Mx3BaseIE._VALID_URL_TMPL % re.escape(_DOMAIN)
  83     _TESTS = [{
  84         'url': 'https://mx3.ch/t/1Cru',
  85         'md5': '7ba09e9826b4447d4e1ce9d69e0e295f',
  86         'info_dict': {
  87             'id': '1Cru',
  88             'ext': 'wav',
  89             'artist': 'Godina',
  90             'album_artist': 'Tortue Tortue',
  91             'composer': 'Olivier Godinat',
  92             'genre': 'Rock',
  93             'thumbnail': 'https://mx3.ch/pictures/mx3/file/0101/4643/square_xlarge/1-s-envoler-1.jpg?1630272813',
  94             'title': "S'envoler",
  95             'release_year': 2021,
  96             'tags': [],
  97         },
  98     }, {
  99         'url': 'https://mx3.ch/t/1LIY',
 100         'md5': '48293cb908342547827f963a5a2e9118',
 101         'info_dict': {
 102             'id': '1LIY',
 103             'ext': 'mov',
 104             'artist': 'Tania Kimfumu',
 105             'album_artist': 'The Broots',
 106             'composer': 'Emmanuel Diserens',
 107             'genre': 'Electro',
 108             'thumbnail': 'https://mx3.ch/pictures/mx3/file/0110/0003/video_xlarge/frame_0000.png?1686963670',
 109             'title': 'The Broots-Larytta remix "Begging For Help"',
 110             'release_year': 2023,
 111             'tags': ['the broots', 'cassata records', 'larytta'],
 112             'description': '"Begging for Help" Larytta Remix Official Video\nRealized By Kali Donkilie in 2023',
 113         },
 114     }, {
 115         'url': 'https://mx3.ch/t/1C6E',
 116         'md5': '1afcd578493ddb8e5008e94bb6d97e25',
 117         'info_dict': {
 118             'id': '1C6E',
 119             'ext': 'wav',
 120             'artist': 'Alien Bubblegum',
 121             'album_artist': 'Alien Bubblegum',
 122             'composer': 'Alien Bubblegum',
 123             'genre': 'Punk',
 124             'thumbnail': 'https://mx3.ch/pictures/mx3/file/0101/1551/square_xlarge/pandora-s-box-cover-with-title.png?1627054733',
 125             'title': 'Wide Awake',
 126             'release_year': 2021,
 127             'tags': ['alien bubblegum', 'bubblegum', 'alien', 'pop punk', 'poppunk'],
 128         },
 129     }]
 130
 131
 132 class Mx3NeoIE(Mx3BaseIE):
 133     _DOMAIN = 'neo.mx3.ch'
 134     _VALID_URL = Mx3BaseIE._VALID_URL_TMPL % re.escape(_DOMAIN)
 135     _TESTS = [{
 136         'url': 'https://neo.mx3.ch/t/1hpd',
 137         'md5': '6d9986bbae5cac3296ec8813bf965eb2',
 138         'info_dict': {
 139             'id': '1hpd',
 140             'ext': 'wav',
 141             'artist': 'Baptiste Lopez',
 142             'album_artist': 'Kammerorchester Basel',
 143             'composer': 'Jannik Giger',
 144             'genre': 'Composition, Orchestra',
 145             'title': 'Troisième œil. Für Kammerorchester (2023)',
 146             'thumbnail': 'https://neo.mx3.ch/pictures/neo/file/0000/0241/square_xlarge/kammerorchester-basel-group-photo-2_c_-lukasz-rajchert.jpg?1560341252',
 147             'release_year': 2023,
 148             'tags': [],
 149         },
 150     }]
 151
 152
 153 class Mx3VolksmusikIE(Mx3BaseIE):
 154     _DOMAIN = 'volksmusik.mx3.ch'
 155     _VALID_URL = Mx3BaseIE._VALID_URL_TMPL % re.escape(_DOMAIN)
 156     _TESTS = [{
 157         'url': 'https://volksmusik.mx3.ch/t/Zx',
 158         'md5': 'dd967a7b0c1ef898f3e072cf9c2eae3c',
 159         'info_dict': {
 160             'id': 'Zx',
 161             'ext': 'mp3',
 162             'artist': 'Ländlerkapelle GrischArt',
 163             'album_artist': 'Ländlerkapelle GrischArt',
 164             'composer': 'Urs Glauser',
 165             'genre': 'Instrumental, Graubünden',
 166             'title': 'Chämilouf',
 167             'thumbnail': 'https://volksmusik.mx3.ch/pictures/vxm/file/0000/3815/square_xlarge/grischart1.jpg?1450530120',
 168             'release_year': 2012,
 169             'tags': [],
 170         },
 171     }]