yt_dlp/extractor/palcomp3.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4
   5 from .common import InfoExtractor
   6 from ..compat import compat_str
   7 from ..utils import (
   8     int_or_none,
   9     str_or_none,
  10     try_get,
  11 )
  12
  13
  14 class PalcoMP3BaseIE(InfoExtractor):
  15     _GQL_QUERY_TMPL = '''{
  16   artist(slug: "%s") {
  17     %s
  18   }
  19 }'''
  20     _ARTIST_FIELDS_TMPL = '''music(slug: "%%s") {
  21       %s
  22     }'''
  23     _MUSIC_FIELDS = '''duration
  24       hls
  25       mp3File
  26       musicID
  27       plays
  28       title'''
  29
  30     def _call_api(self, artist_slug, artist_fields):
  31         return self._download_json(
  32             'https://www.palcomp3.com.br/graphql/', artist_slug, query={
  33                 'query': self._GQL_QUERY_TMPL % (artist_slug, artist_fields),
  34             })['data']
  35
  36     def _parse_music(self, music):
  37         music_id = compat_str(music['musicID'])
  38         title = music['title']
  39
  40         formats = []
  41         hls_url = music.get('hls')
  42         if hls_url:
  43             formats.append({
  44                 'url': hls_url,
  45                 'protocol': 'm3u8_native',
  46                 'ext': 'mp4',
  47             })
  48         mp3_file = music.get('mp3File')
  49         if mp3_file:
  50             formats.append({
  51                 'url': mp3_file,
  52             })
  53
  54         return {
  55             'id': music_id,
  56             'title': title,
  57             'formats': formats,
  58             'duration': int_or_none(music.get('duration')),
  59             'view_count': int_or_none(music.get('plays')),
  60         }
  61
  62     def _real_initialize(self):
  63         self._ARTIST_FIELDS_TMPL = self._ARTIST_FIELDS_TMPL % self._MUSIC_FIELDS
  64
  65     def _real_extract(self, url):
  66         artist_slug, music_slug = self._match_valid_url(url).groups()
  67         artist_fields = self._ARTIST_FIELDS_TMPL % music_slug
  68         music = self._call_api(artist_slug, artist_fields)['artist']['music']
  69         return self._parse_music(music)
  70
  71
  72 class PalcoMP3IE(PalcoMP3BaseIE):
  73     IE_NAME = 'PalcoMP3:song'
  74     _VALID_URL = r'https?://(?:www\.)?palcomp3\.com(?:\.br)?/(?P<artist>[^/]+)/(?P<id>[^/?&#]+)'
  75     _TESTS = [{
  76         'url': 'https://www.palcomp3.com/maiaraemaraisaoficial/nossas-composicoes-cuida-bem-dela/',
  77         'md5': '99fd6405b2d8fd589670f6db1ba3b358',
  78         'info_dict': {
  79             'id': '3162927',
  80             'ext': 'mp3',
  81             'title': 'Nossas Composições - CUIDA BEM DELA',
  82             'duration': 210,
  83             'view_count': int,
  84         }
  85     }]
  86
  87     @classmethod
  88     def suitable(cls, url):
  89         return False if PalcoMP3VideoIE.suitable(url) else super(PalcoMP3IE, cls).suitable(url)
  90
  91
  92 class PalcoMP3ArtistIE(PalcoMP3BaseIE):
  93     IE_NAME = 'PalcoMP3:artist'
  94     _VALID_URL = r'https?://(?:www\.)?palcomp3\.com(?:\.br)?/(?P<id>[^/?&#]+)'
  95     _TESTS = [{
  96         'url': 'https://www.palcomp3.com.br/condedoforro/',
  97         'info_dict': {
  98             'id': '358396',
  99             'title': 'Conde do Forró',
 100         },
 101         'playlist_mincount': 188,
 102     }]
 103     _ARTIST_FIELDS_TMPL = '''artistID
 104     musics {
 105       nodes {
 106         %s
 107       }
 108     }
 109     name'''
 110
 111     @ classmethod
 112     def suitable(cls, url):
 113         return False if PalcoMP3IE._match_valid_url(url) else super(PalcoMP3ArtistIE, cls).suitable(url)
 114
 115     def _real_extract(self, url):
 116         artist_slug = self._match_id(url)
 117         artist = self._call_api(artist_slug, self._ARTIST_FIELDS_TMPL)['artist']
 118
 119         def entries():
 120             for music in (try_get(artist, lambda x: x['musics']['nodes'], list) or []):
 121                 yield self._parse_music(music)
 122
 123         return self.playlist_result(
 124             entries(), str_or_none(artist.get('artistID')), artist.get('name'))
 125
 126
 127 class PalcoMP3VideoIE(PalcoMP3BaseIE):
 128     IE_NAME = 'PalcoMP3:video'
 129     _VALID_URL = r'https?://(?:www\.)?palcomp3\.com(?:\.br)?/(?P<artist>[^/]+)/(?P<id>[^/?&#]+)/?#clipe'
 130     _TESTS = [{
 131         'url': 'https://www.palcomp3.com/maiaraemaraisaoficial/maiara-e-maraisa-voce-faz-falta-aqui-ao-vivo-em-vicosa-mg/#clipe',
 132         'add_ie': ['Youtube'],
 133         'info_dict': {
 134             'id': '_pD1nR2qqPg',
 135             'ext': 'mp4',
 136             'title': 'Maiara e Maraisa - Você Faz Falta Aqui - DVD Ao Vivo Em Campo Grande',
 137             'description': 'md5:7043342c09a224598e93546e98e49282',
 138             'upload_date': '20161107',
 139             'uploader_id': 'maiaramaraisaoficial',
 140             'uploader': 'Maiara e Maraisa',
 141         }
 142     }]
 143     _MUSIC_FIELDS = 'youtubeID'
 144
 145     def _parse_music(self, music):
 146         youtube_id = music['youtubeID']
 147         return self.url_result(youtube_id, 'Youtube', youtube_id)