mirror/yt-dlc
1
0
Fork 0
mirror of https://github.com/blackjack4494/yt-dlc.git synced 2024-12-22 07:43:08 +00:00
Code Issues Releases Wiki Activity

[audioboom] improve metadata extraction

Browse source
This commit is contained in:
Remita Amine 2019-10-17 13:20:16 +01:00
parent 0001157594
commit 86f63633c8
1 changed files with 19 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

34
youtube_dl/extractor/audioboom.py
View file

@ -2,22 +2,25 @@
from __future__ import unicode_literals from __future__ import unicode_literals
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import float_or_none from ..utils import (
clean_html,
float_or_none,
)
class AudioBoomIE(InfoExtractor): class AudioBoomIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?audioboom\.com/(?:boos|posts)/(?P<id>[0-9]+)' _VALID_URL = r'https?://(?:www\.)?audioboom\.com/(?:boos|posts)/(?P<id>[0-9]+)'
_TESTS = [{ _TESTS = [{
'url': 'https://audioboom.com/boos/4279833-3-09-2016-czaban-hour-3?t=0', 'url': 'https://audioboom.com/posts/7398103-asim-chaudhry',
'md5': '63a8d73a055c6ed0f1e51921a10a5a76', 'md5': '7b00192e593ff227e6a315486979a42d',
'info_dict': { 'info_dict': {
'id': '4279833', 'id': '7398103',
'ext': 'mp3', 'ext': 'mp3',
'title': '3/09/2016 Czaban Hour 3', 'title': 'Asim Chaudhry',
'description': 'Guest: Nate Davis - NFL free agency, Guest: Stan Gans', 'description': 'md5:2f3fef17dacc2595b5362e1d7d3602fc',
'duration': 2245.72, 'duration': 4000.99,
'uploader': 'SB Nation A.M.', 'uploader': 'Sue Perkins: An hour or so with...',
'uploader_url': r're:https?://(?:www\.)?audioboom\.com/channel/steveczabanyahoosportsradio', 'uploader_url': r're:https?://(?:www\.)?audioboom\.com/channel/perkins',
} }
}, { }, {
'url': 'https://audioboom.com/posts/4279833-3-09-2016-czaban-hour-3?t=0', 'url': 'https://audioboom.com/posts/4279833-3-09-2016-czaban-hour-3?t=0',
@ -32,8 +35,8 @@ class AudioBoomIE(InfoExtractor):
clip = None clip = None
clip_store = self._parse_json( clip_store = self._parse_json(
self._search_regex( self._html_search_regex(
r'data-new-clip-store=(["\'])(?P<json>{.*?"clipId"\s*:\s*%s.*?})\1' % video_id, r'data-new-clip-store=(["\'])(?P<json>{.+?})\1',
webpage, 'clip store', default='{}', group='json'), webpage, 'clip store', default='{}', group='json'),
video_id, fatal=False) video_id, fatal=False)
if clip_store: if clip_store:
@ -47,14 +50,15 @@ class AudioBoomIE(InfoExtractor):
audio_url = from_clip('clipURLPriorToLoading') or self._og_search_property( audio_url = from_clip('clipURLPriorToLoading') or self._og_search_property(
'audio', webpage, 'audio url') 'audio', webpage, 'audio url')
title = from_clip('title') or self._og_search_title(webpage) title = from_clip('title') or self._html_search_meta(
description = from_clip('description') or self._og_search_description(webpage) ['og:title', 'og:audio:title', 'audio_title'], webpage)
description = from_clip('description') or clean_html(from_clip('formattedDescription')) or self._og_search_description(webpage)
duration = float_or_none(from_clip('duration') or self._html_search_meta( duration = float_or_none(from_clip('duration') or self._html_search_meta(
'weibo:audio:duration', webpage)) 'weibo:audio:duration', webpage))
uploader = from_clip('author') or self._og_search_property( uploader = from_clip('author') or self._html_search_meta(
'audio:artist', webpage, 'uploader', fatal=False) ['og:audio:artist', 'twitter:audio:artist_name', 'audio_artist'], webpage, 'uploader')
uploader_url = from_clip('author_url') or self._html_search_meta( uploader_url = from_clip('author_url') or self._html_search_meta(
'audioboo:channel', webpage, 'uploader url') 'audioboo:channel', webpage, 'uploader url')