-# coding: utf-8
-from __future__ import unicode_literals
-
import random
import re
import time
from .common import InfoExtractor
from ..compat import compat_str
from ..utils import (
+ KNOWN_EXTENSIONS,
ExtractorError,
float_or_none,
int_or_none,
- KNOWN_EXTENSIONS,
parse_filesize,
str_or_none,
try_get,
- update_url_query,
unified_strdate,
unified_timestamp,
+ update_url_query,
url_or_none,
urljoin,
)
class BandcampIE(InfoExtractor):
- _VALID_URL = r'https?://[^/]+\.bandcamp\.com/track/(?P<id>[^/?#&]+)'
+ _VALID_URL = r'https?://(?P<uploader>[^/]+)\.bandcamp\.com/track/(?P<id>[^/?#&]+)'
+ _EMBED_REGEX = [r'<meta property="og:url"[^>]*?content="(?P<url>.*?bandcamp\.com.*?)"']
_TESTS = [{
'url': 'http://youtube-dl.bandcamp.com/track/youtube-dl-test-song',
'md5': 'c557841d5e50261777a6585648adf439',
'info_dict': {
'id': '1812978515',
'ext': 'mp3',
- 'title': "youtube-dl \"'/\\ä↭ - youtube-dl \"'/\\ä↭ - youtube-dl test song \"'/\\ä↭",
+ 'title': 'youtube-dl "\'/\\ä↭ - youtube-dl "\'/\\ä↭ - youtube-dl test song "\'/\\ä↭',
'duration': 9.8485,
- 'uploader': 'youtube-dl "\'/\\ä↭',
+ 'uploader': 'youtube-dl "\'/\\ä↭',
'upload_date': '20121129',
'timestamp': 1354224127,
+ 'track': 'youtube-dl "\'/\\ä↭ - youtube-dl test song "\'/\\ä↭',
+ 'album_artist': 'youtube-dl "\'/\\ä↭',
+ 'track_id': '1812978515',
+ 'artist': 'youtube-dl "\'/\\ä↭',
+ 'uploader_url': 'https://youtube-dl.bandcamp.com',
+ 'uploader_id': 'youtube-dl',
+ 'thumbnail': 'https://f4.bcbits.com/img/a3216802731_5.jpg',
},
'_skip': 'There is a limit of 200 free downloads / month for the test song'
}, {
'url': 'http://benprunty.bandcamp.com/track/lanius-battle',
'info_dict': {
'id': '2650410135',
- 'ext': 'aiff',
+ 'ext': 'm4a',
+ 'acodec': r're:[fa]lac',
'title': 'Ben Prunty - Lanius (Battle)',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'Ben Prunty',
'track_number': 1,
'track_id': '2650410135',
'artist': 'Ben Prunty',
+ 'album_artist': 'Ben Prunty',
'album': 'FTL: Advanced Edition Soundtrack',
+ 'uploader_url': 'https://benprunty.bandcamp.com',
+ 'uploader_id': 'benprunty',
},
}, {
# no free download, mp3 128
'track_number': 5,
'track_id': '2584466013',
'artist': 'Mastodon',
+ 'album_artist': 'Mastodon',
'album': 'Call of the Mastodon',
+ 'uploader_url': 'https://relapsealumni.bandcamp.com',
+ 'uploader_id': 'relapsealumni',
+ },
+ }, {
+ # track from compilation album (artist/album_artist difference)
+ 'url': 'https://diskotopia.bandcamp.com/track/safehouse',
+ 'md5': '19c5337bca1428afa54129f86a2f6a69',
+ 'info_dict': {
+ 'id': '1978174799',
+ 'ext': 'mp3',
+ 'title': 'submerse - submerse - Safehouse',
+ 'thumbnail': r're:^https?://.*\.jpg$',
+ 'uploader': 'submerse',
+ 'timestamp': 1480779297,
+ 'upload_date': '20161203',
+ 'release_timestamp': 1481068800,
+ 'release_date': '20161207',
+ 'duration': 154.066,
+ 'track': 'submerse - Safehouse',
+ 'track_number': 3,
+ 'track_id': '1978174799',
+ 'artist': 'submerse',
+ 'album_artist': 'Diskotopia',
+ 'album': 'DSK F/W 2016-2017 Free Compilation',
+ 'uploader_url': 'https://diskotopia.bandcamp.com',
+ 'uploader_id': 'diskotopia',
},
}]
attr + ' data', group=2), video_id, fatal=fatal)
def _real_extract(self, url):
- title = self._match_id(url)
+ title, uploader = self._match_valid_url(url).group('id', 'uploader')
webpage = self._download_webpage(url, title)
tralbum = self._extract_data_attr(webpage, title)
thumbnail = self._og_search_thumbnail(webpage)
embed = self._extract_data_attr(webpage, title, 'embed', False)
current = tralbum.get('current') or {}
artist = embed.get('artist') or current.get('artist') or tralbum.get('artist')
+ album_artist = self._html_search_regex(
+ r'<h3 class="albumTitle">[\S\s]*?by\s*<span>\s*<a href="[^>]+">\s*([^>]+?)\s*</a>',
+ webpage, 'album artist', fatal=False)
timestamp = unified_timestamp(
current.get('publish_date') or tralbum.get('album_publish_date'))
'acodec': format_id.split('-')[0],
})
- self._sort_formats(formats)
-
title = '%s - %s' % (artist, track) if artist else track
if not duration:
'title': title,
'thumbnail': thumbnail,
'uploader': artist,
+ 'uploader_id': uploader,
+ 'uploader_url': f'https://{uploader}.bandcamp.com',
'timestamp': timestamp,
'release_timestamp': unified_timestamp(tralbum.get('album_release_date')),
'duration': duration,
'track_id': track_id,
'artist': artist,
'album': embed.get('album_title'),
+ 'album_artist': album_artist,
'formats': formats,
}
-class BandcampAlbumIE(BandcampIE):
+class BandcampAlbumIE(BandcampIE): # XXX: Do not subclass from concrete IE
IE_NAME = 'Bandcamp:album'
_VALID_URL = r'https?://(?:(?P<subdomain>[^.]+)\.)?bandcamp\.com/album/(?P<id>[^/?#&]+)'
}
-class BandcampWeeklyIE(BandcampIE):
+class BandcampWeeklyIE(BandcampIE): # XXX: Do not subclass from concrete IE
IE_NAME = 'Bandcamp:weekly'
_VALID_URL = r'https?://(?:www\.)?bandcamp\.com/?\?(?:.*?&)?show=(?P<id>\d+)'
_TESTS = [{
'ext': ext,
'vcodec': 'none',
})
- self._sort_formats(formats)
title = show.get('audio_title') or 'Bandcamp Weekly'
subtitle = show.get('subtitle')
uploader = self._match_id(url)
webpage = self._download_webpage(url, uploader)
- discography_data = (re.findall(r'<li data-item-id=["\'][^>]+>\s*<a href=["\']([^"\']+)', webpage)
+ discography_data = (re.findall(r'<li data-item-id=["\'][^>]+>\s*<a href=["\'](?![^"\'/]*?/merch)([^"\']+)', webpage)
or re.findall(r'<div[^>]+trackTitle["\'][^"\']+["\']([^"\']+)', webpage))
return self.playlist_from_matches(