yt_dlp/extractor/lastfm.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import int_or_none, format_field
   8
   9
  10 class LastFMPlaylistBaseIE(InfoExtractor):
  11     def _entries(self, url, playlist_id):
  12         webpage = self._download_webpage(url, playlist_id)
  13         start_page_number = int_or_none(self._search_regex(
  14             r'\bpage=(\d+)', url, 'page', default=None)) or 1
  15         last_page_number = int_or_none(self._search_regex(
  16             r'>(\d+)</a>[^<]*</li>[^<]*<li[^>]+class="pagination-next', webpage, 'last_page', default=None))
  17
  18         for page_number in range(start_page_number, (last_page_number or start_page_number) + 1):
  19             webpage = self._download_webpage(
  20                 url, playlist_id,
  21                 note='Downloading page %d%s' % (page_number, format_field(last_page_number, template=' of %d')),
  22                 query={'page': page_number})
  23             page_entries = [
  24                 self.url_result(player_url, 'Youtube')
  25                 for player_url in set(re.findall(r'data-youtube-url="([^"]+)"', webpage))
  26             ]
  27
  28             for e in page_entries:
  29                 yield e
  30
  31     def _real_extract(self, url):
  32         playlist_id = self._match_id(url)
  33         return self.playlist_result(self._entries(url, playlist_id), playlist_id)
  34
  35
  36 class LastFMPlaylistIE(LastFMPlaylistBaseIE):
  37     _VALID_URL = r'https?://(?:www\.)?last\.fm/(music|tag)/(?P<id>[^/]+)(?:/[^/]+)?/?(?:[?#]|$)'
  38     _TESTS = [{
  39         'url': 'https://www.last.fm/music/Oasis/(What%27s+the+Story)+Morning+Glory%3F',
  40         'info_dict': {
  41             'id': 'Oasis',
  42         },
  43         'playlist_count': 11,
  44     }, {
  45         'url': 'https://www.last.fm/music/Oasis',
  46         'only_matching': True,
  47     }, {
  48         'url': 'https://www.last.fm/music/Oasis/',
  49         'only_matching': True,
  50     }, {
  51         'url': 'https://www.last.fm/music/Oasis?top_tracks_date_preset=ALL#top-tracks',
  52         'only_matching': True,
  53     }, {
  54         'url': 'https://www.last.fm/music/Oasis/+tracks',
  55         'only_matching': True,
  56     }, {
  57         'url': 'https://www.last.fm/music/Oasis/+tracks?page=2',
  58         'only_matching': True,
  59     }, {
  60         'url': 'https://www.last.fm/music/Oasis/+tracks?date_preset=LAST_90_DAYS#top-tracks',
  61         'only_matching': True,
  62     }, {
  63         'url': 'https://www.last.fm/tag/rock',
  64         'only_matching': True,
  65     }, {
  66         'url': 'https://www.last.fm/tag/rock/tracks',
  67         'only_matching': True,
  68     }]
  69
  70
  71 class LastFMUserIE(LastFMPlaylistBaseIE):
  72     _VALID_URL = r'https?://(?:www\.)?last\.fm/user/[^/]+/playlists/(?P<id>[^/#?]+)'
  73     _TESTS = [{
  74         'url': 'https://www.last.fm/user/mehq/playlists/12319471',
  75         'info_dict': {
  76             'id': '12319471',
  77         },
  78         'playlist_count': 30,
  79     }]
  80
  81
  82 class LastFMIE(InfoExtractor):
  83     _VALID_URL = r'https?://(?:www\.)?last\.fm/music(?:/[^/]+){2}/(?P<id>[^/#?]+)'
  84     _TESTS = [{
  85         'url': 'https://www.last.fm/music/Oasis/_/Wonderwall',
  86         'md5': '9c4a70c2e84c03d54fe24229b9e13b7b',
  87         'info_dict': {
  88             'id': '6hzrDeceEKc',
  89             'ext': 'mp4',
  90             'title': 'Oasis - Wonderwall  (Official Video)',
  91             'thumbnail': r're:^https?://i.ytimg.com/.*\.jpg$',
  92             'description': 'md5:0848669853c10687cc28e88b5756738f',
  93             'uploader': 'Oasis',
  94             'uploader_id': 'oasisinetofficial',
  95             'upload_date': '20080207',
  96             'album': '(What\'s The Story) Morning Glory? (Remastered)',
  97             'track': 'Wonderwall (Remastered)',
  98             'channel_id': 'UCUDVBtnOQi4c7E8jebpjc9Q',
  99             'view_count': int,
 100             'live_status': 'not_live',
 101             'channel_url': 'https://www.youtube.com/channel/UCUDVBtnOQi4c7E8jebpjc9Q',
 102             'tags': 'count:39',
 103             'creator': 'Oasis',
 104             'uploader_url': 're:^https?://www.youtube.com/user/oasisinetofficial',
 105             'duration': 279,
 106             'alt_title': 'Wonderwall (Remastered)',
 107             'age_limit': 0,
 108             'channel': 'Oasis',
 109             'channel_follower_count': int,
 110             'categories': ['Music'],
 111             'availability': 'public',
 112             'like_count': int,
 113             'playable_in_embed': True,
 114             'artist': 'Oasis',
 115         },
 116         'add_ie': ['Youtube'],
 117     }, {
 118         'url': 'https://www.last.fm/music/Oasis/_/Don%27t+Look+Back+In+Anger+-+Remastered/',
 119         'only_matching': True,
 120     }, {
 121         'url': 'https://www.last.fm/music/Guns+N%27+Roses/_/Sweet+Child+o%27+Mine',
 122         'only_matching': True,
 123     }]
 124
 125     def _real_extract(self, url):
 126         video_id = self._match_id(url)
 127         webpage = self._download_webpage(url, video_id)
 128         player_url = self._search_regex(r'(?s)class="header-new-playlink"\s+href="([^"]+)"', webpage, 'player_url')
 129         return self.url_result(player_url, 'Youtube')