youtube-dl/youtube_dl/extractor/beatportpro.py

100 lines
3.1 KiB
Python
Raw Normal View History

# coding: utf-8
from __future__ import unicode_literals
import re
import json
2015-03-13 16:01:15 +00:00
from .common import InfoExtractor
from ..utils import int_or_none
class BeatportProIE(InfoExtractor):
2015-03-13 16:01:15 +00:00
_VALID_URL = r'https?://pro\.beatport\.com/track/.+/(?P<id>[0-9]+)'
_TESTS = [{
'url': 'https://pro.beatport.com/track/synesthesia-original-mix/5379371',
'md5': 'b3c34d8639a2f6a7f734382358478887',
'info_dict': {
'id': 5379371,
'display-id': 'synesthesia-original-mix',
'ext': 'mp4',
'title': 'Froxic - Synesthesia (Original Mix)',
},
}, {
'url': 'https://pro.beatport.com/track/love-and-war-original-mix/3756896',
'md5': 'e44c3025dfa38c6577fbaeb43da43514',
'info_dict': {
'id': 3756896,
'display-id': 'love-and-war-original-mix',
'ext': 'mp3',
'title': 'Wolfgang Gartner - Love & War (Original Mix)',
},
}, {
'url': 'https://pro.beatport.com/track/birds-original-mix/4991738',
'md5': 'a1fd8e8046de3950fd039304c186c05f',
'info_dict': {
'id': 4991738,
'display-id': 'birds-original-mix',
'ext': 'mp4',
'title': "Tos, Middle Milk, Mumblin' Johnsson - Birds (Original Mix)",
}
}]
def _real_extract(self, url):
track_id = self._match_id(url)
webpage = self._download_webpage(url, track_id)
2015-03-13 16:01:15 +00:00
playables = self._search_regex(
r'window\.Playables\s*=\s*({.*?});', webpage,
'playables info', flags=re.DOTALL)
playables = json.loads(playables)
2015-03-12 20:42:55 +00:00
track = next(t for t in playables['tracks'] if t['id'] == int(track_id))
title = ', '.join((a['name'] for a in track['artists'])) + ' - ' + track['name']
if track['mix']:
title += ' (' + track['mix'] + ')'
formats = []
for ext, info in track['preview'].items():
if info['url'] is None:
continue
fmt = {
'url': info['url'],
'ext': ext,
'format_id': ext,
'vcodec': 'none',
}
if ext == 'mp3':
fmt['preference'] = 0
fmt['acodec'] = 'mp3'
fmt['abr'] = 96
fmt['asr'] = 44100
elif ext == 'mp4':
fmt['preference'] = 1
fmt['acodec'] = 'aac'
fmt['abr'] = 96
fmt['asr'] = 44100
2015-03-13 16:01:15 +00:00
formats.append(fmt)
2015-03-13 15:51:49 +00:00
self._sort_formats(formats)
2015-03-13 16:01:15 +00:00
images = []
for name, info in track['images'].items():
2015-03-13 16:01:15 +00:00
image_url = info.get('url')
if name == 'dynamic' or not image_url:
continue
img = {
'id': name,
2015-03-13 16:01:15 +00:00
'url': image_url,
'height': int_or_none(info.get('height')),
'width': int_or_none(info.get('width')),
}
2015-03-13 16:01:15 +00:00
images.append(img)
return {
'id': track['id'],
'display-id': track['slug'],
'title': title,
'formats': formats,
2015-03-13 16:01:15 +00:00
'thumbnails': images,
}