[twitter] Fix metadata extraction and test_Twitter_1

This commit is contained in:
Yen Chi Hsuan 2016-02-21 17:29:28 +08:00
parent 132e3b74bd
commit 7efc1c2b49

View File

@ -165,6 +165,7 @@ class TwitterIE(InfoExtractor):
'uploader': 'Gifs', 'uploader': 'Gifs',
'uploader_id': 'giphz', 'uploader_id': 'giphz',
}, },
'expected_warnings': ['height', 'width'],
}, { }, {
'url': 'https://twitter.com/starwars/status/665052190608723968', 'url': 'https://twitter.com/starwars/status/665052190608723968',
'md5': '39b7199856dee6cd4432e72c74bc69d4', 'md5': '39b7199856dee6cd4432e72c74bc69d4',
@ -212,20 +213,24 @@ class TwitterIE(InfoExtractor):
return info return info
mobj = re.search(r'''(?x) mobj = re.search(r'''(?x)
<video[^>]+class="animated-gif"[^>]+ <video[^>]+class="animated-gif"(?P<more_info>[^>]+)>\s*
(?:data-height="(?P<height>\d+)")?[^>]+
(?:data-width="(?P<width>\d+)")?[^>]+
(?:poster="(?P<poster>[^"]+)")?[^>]*>\s*
<source[^>]+video-src="(?P<url>[^"]+)" <source[^>]+video-src="(?P<url>[^"]+)"
''', webpage) ''', webpage)
if mobj: if mobj:
more_info = mobj.group('more_info')
height = int_or_none(self._search_regex(
r'data-height="(\d+)"', more_info, 'height', fatal=False))
width = int_or_none(self._search_regex(
r'data-width="(\d+)"', more_info, 'width', fatal=False))
thumbnail = self._search_regex(
r'poster="([^"]+)"', more_info, 'poster', fatal=False)
info.update({ info.update({
'id': twid, 'id': twid,
'url': mobj.group('url'), 'url': mobj.group('url'),
'height': int_or_none(mobj.group('height')), 'height': height,
'width': int_or_none(mobj.group('width')), 'width': width,
'thumbnail': mobj.group('poster'), 'thumbnail': thumbnail,
}) })
return info return info