[freesound] Minor improvements

This commit is contained in:
Jaime Marquínez Ferrándiz 2013-07-15 21:33:45 +02:00
parent 66400c470c
commit 67de24e449
2 changed files with 17 additions and 17 deletions

View file

@ -20,7 +20,7 @@
from .escapist import EscapistIE from .escapist import EscapistIE
from .facebook import FacebookIE from .facebook import FacebookIE
from .flickr import FlickrIE from .flickr import FlickrIE
from .freesound import FreeSoundIE from .freesound import FreesoundIE
from .funnyordie import FunnyOrDieIE from .funnyordie import FunnyOrDieIE
from .gamespot import GameSpotIE from .gamespot import GameSpotIE
from .gametrailers import GametrailersIE from .gametrailers import GametrailersIE

View file

@ -1,36 +1,36 @@
# -*- coding: utf-8 -*-
import re import re
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import determine_ext
class FreeSoundIE(InfoExtractor): class FreesoundIE(InfoExtractor):
_VALID_URL = r'(?:http://)?(?:www\.)?freesound\.org/people/([^/]+)/sounds/([^/]+)' _VALID_URL = r'(?:https?://)?(?:www\.)?freesound\.org/people/([^/]+)/sounds/(?P<id>[^/]+)'
_TEST = { _TEST = {
u'url': u'http://www.freesound.org/people/miklovan/sounds/194503/', u'url': u'http://www.freesound.org/people/miklovan/sounds/194503/',
u'file': u'194503.mp3', u'file': u'194503.mp3',
u'md5': u'12280ceb42c81f19a515c745eae07650', u'md5': u'12280ceb42c81f19a515c745eae07650',
u'info_dict': { u'info_dict': {
u"title": u"gulls in the city.wav by miklovan", u"title": u"gulls in the city.wav",
u"uploader" : u"miklovan" u"uploader" : u"miklovan",
u'description': u'the sounds of seagulls in the city',
} }
} }
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
music_id = mobj.group(2) music_id = mobj.group('id')
webpage = self._download_webpage(url, music_id) webpage = self._download_webpage(url, music_id)
title = self._html_search_regex(r'<meta property="og:title" content="([^"]*)"', title = self._html_search_regex(r'<div id="single_sample_header">.*?<a href="#">(.+?)</a>',
webpage, 'music title') webpage, 'music title', flags=re.DOTALL)
music_url = self._html_search_regex(r'<meta property="og:audio" content="([^"]*)"', music_url = self._og_search_property('audio', webpage, 'music url')
webpage, 'music url') description = self._html_search_regex(r'<div id="sound_description">(.*?)</div>',
uploader = self._html_search_regex(r'<meta property="og:audio:artist" content="([^"]*)"', webpage, 'description', fatal=False, flags=re.DOTALL)
webpage, 'music uploader')
ext = music_url.split('.')[-1]
return [{ return [{
'id': music_id, 'id': music_id,
'title': title, 'title': title,
'url': music_url, 'url': music_url,
'uploader': uploader, 'uploader': self._og_search_property('audio:artist', webpage, 'music uploader'),
'ext': ext, 'ext': determine_ext(music_url),
}] 'description': description,
}]