2 from __future__ import unicode_literals
4 from .common import InfoExtractor
5 from ..compat import compat_urlparse
6 from ..utils import month_by_abbreviation
10 class YamIE(InfoExtractor):
11 _VALID_URL = r'http://mymedia.yam.com/m/(?P<id>\d+)'
14 # An audio hosted on Yam
15 'url': 'http://mymedia.yam.com/m/2283921',
16 'md5': 'c011b8e262a52d5473d9c2e3c9963b9c',
20 'title': '發現 - 趙薇 京華煙雲主題曲',
21 'uploader_id': 'princekt',
22 'upload_date': '20080807',
26 # An external video hosted on YouTube
27 'url': 'http://mymedia.yam.com/m/3598173',
28 'md5': '0238ceec479c654e8c2f1223755bf3e9',
32 'upload_date': '20150202',
33 'uploader': '新莊社大瑜伽社',
34 'description': 'md5:f5cc72f0baf259a70fb731654b0d2eff',
35 'uploader_id': '2323agoy',
36 'title': '外婆的澎湖灣KTV-潘安邦',
40 def _real_extract(self, url):
41 media_id = self._match_id(url)
42 page = self._download_webpage(url, media_id)
44 # Is it hosted externally on YouTube?
45 youtube_url = self._html_search_regex(
46 r'<embed src="(http://www.youtube.com/[^"]+)"',
47 page, 'YouTube url', default=None)
49 return self.url_result(youtube_url, 'Youtube')
51 api_page = self._download_webpage(
52 'http://mymedia.yam.com/api/a/?pID=' + media_id, media_id)
53 api_result_obj = compat_urlparse.parse_qs(api_page)
55 author = self._html_search_regex(
56 r'<!-- 發表作者 -->:[\n ]+<a href="/([a-z]+)"', page, 'author')
57 mobj = re.search(r'<!-- 發表於 -->(?P<mon>[A-Z][a-z]{2}) ' +
58 r'(?P<day>\d{1,2}), (?P<year>\d{4})', page)
59 upload_date = '%s%02d%02d' % (mobj.group('year'),
60 month_by_abbreviation(mobj.group('mon')),
61 int(mobj.group('day')))
65 'url': api_result_obj['mp3file'][0],
66 'title': self._html_search_meta('description', page),
67 'duration': float(api_result_obj['totaltime'][0]) / 1000.0,
68 'uploader_id': author,
69 'upload_date': upload_date,