mirror of
https://codeberg.org/polarisfm/youtube-dl
synced 2024-11-22 16:44:32 +01:00
88 lines
3.2 KiB
Python
88 lines
3.2 KiB
Python
# coding: utf-8
|
|
from __future__ import unicode_literals
|
|
|
|
import json
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from ..utils import (
|
|
parse_duration,
|
|
unified_strdate,
|
|
)
|
|
|
|
|
|
class LibsynIE(InfoExtractor):
|
|
_VALID_URL = r'(?P<mainurl>https?://html5-player\.libsyn\.com/embed/episode/id/(?P<id>[0-9]+))'
|
|
|
|
_TESTS = [{
|
|
'url': 'http://html5-player.libsyn.com/embed/episode/id/6385796/',
|
|
'md5': '2a55e75496c790cdeb058e7e6c087746',
|
|
'info_dict': {
|
|
'id': '6385796',
|
|
'ext': 'mp3',
|
|
'title': "Champion Minded - Developing a Growth Mindset",
|
|
'description': 'In this episode, Allistair talks about the importance of developing a growth mindset, not only in sports, but in life too.',
|
|
'upload_date': '20180320',
|
|
'thumbnail': 're:^https?://.*',
|
|
},
|
|
}, {
|
|
'url': 'https://html5-player.libsyn.com/embed/episode/id/3727166/height/75/width/200/theme/standard/direction/no/autoplay/no/autonext/no/thumbnail/no/preload/no/no_addthis/no/',
|
|
'md5': '6c5cb21acd622d754d3b1a92b582ce42',
|
|
'info_dict': {
|
|
'id': '3727166',
|
|
'ext': 'mp3',
|
|
'title': 'Clients From Hell Podcast - How a Sex Toy Company Kickstarted my Freelance Career',
|
|
'upload_date': '20150818',
|
|
'thumbnail': 're:^https?://.*',
|
|
}
|
|
}]
|
|
|
|
def _real_extract(self, url):
|
|
m = re.match(self._VALID_URL, url)
|
|
video_id = m.group('id')
|
|
url = m.group('mainurl')
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
podcast_title = self._search_regex(
|
|
r'<h3>([^<]+)</h3>', webpage, 'podcast title', default=None)
|
|
if podcast_title:
|
|
podcast_title = podcast_title.strip()
|
|
episode_title = self._search_regex(
|
|
r'(?:<div class="episode-title">|<h4>)([^<]+)</', webpage, 'episode title')
|
|
if episode_title:
|
|
episode_title = episode_title.strip()
|
|
|
|
title = '%s - %s' % (podcast_title, episode_title) if podcast_title else episode_title
|
|
|
|
description = self._html_search_regex(
|
|
r'<p\s+id="info_text_body">(.+?)</p>', webpage,
|
|
'description', default=None)
|
|
if description:
|
|
# Strip non-breaking and normal spaces
|
|
description = description.replace('\u00A0', ' ').strip()
|
|
release_date = unified_strdate(self._search_regex(
|
|
r'<div class="release_date">Released: ([^<]+)<', webpage, 'release date', fatal=False))
|
|
|
|
data_json = self._search_regex(r'var\s+playlistItem\s*=\s*(\{.*?\});\n', webpage, 'JSON data block')
|
|
data = json.loads(data_json)
|
|
|
|
formats = [{
|
|
'url': data['media_url'],
|
|
'format_id': 'main',
|
|
}, {
|
|
'url': data['media_url_libsyn'],
|
|
'format_id': 'libsyn',
|
|
}]
|
|
thumbnail = data.get('thumbnail_url')
|
|
duration = parse_duration(data.get('duration'))
|
|
|
|
return {
|
|
'id': video_id,
|
|
'title': title,
|
|
'description': description,
|
|
'thumbnail': thumbnail,
|
|
'upload_date': release_date,
|
|
'duration': duration,
|
|
'formats': formats,
|
|
}
|