projects
/
youtube-dl
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Update description value for the write_info_json test (required after 27dcce19045670f...
[youtube-dl]
/
youtube_dl
/
extractor
/
mit.py
diff --git
a/youtube_dl/extractor/mit.py
b/youtube_dl/extractor/mit.py
index d09d03e36e8be030b889ba9fd4912f5df128a5c2..52be9232fe12d394ed600c349721339c5891a78a 100644
(file)
--- a/
youtube_dl/extractor/mit.py
+++ b/
youtube_dl/extractor/mit.py
@@
-25,23
+25,21
@@
class TechTVMITIE(InfoExtractor):
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
-
web
page = self._download_webpage(
+
raw_
page = self._download_webpage(
'http://techtv.mit.edu/videos/%s' % video_id, video_id)
'http://techtv.mit.edu/videos/%s' % video_id, video_id)
- embed_page = self._download_webpage(
- 'http://techtv.mit.edu/embeds/%s/' % video_id, video_id,
- note=u'Downloading embed page')
+ clean_page = re.compile(u'<!--.*?-->', re.S).sub(u'', raw_page)
base_url = self._search_regex(r'ipadUrl: \'(.+?cloudfront.net/)',
base_url = self._search_regex(r'ipadUrl: \'(.+?cloudfront.net/)',
-
embed
_page, u'base url')
- formats_json = self._search_regex(r'bitrates: (\[.+?\])',
embed
_page,
+
raw
_page, u'base url')
+ formats_json = self._search_regex(r'bitrates: (\[.+?\])',
raw
_page,
u'video formats')
formats = json.loads(formats_json)
formats = sorted(formats, key=lambda f: f['bitrate'])
u'video formats')
formats = json.loads(formats_json)
formats = sorted(formats, key=lambda f: f['bitrate'])
- title = get_element_by_id('edit-title',
web
page)
- description = clean_html(get_element_by_id('edit-description',
web
page))
+ title = get_element_by_id('edit-title',
clean_
page)
+ description = clean_html(get_element_by_id('edit-description',
clean_
page))
thumbnail = self._search_regex(r'playlist:.*?url: \'(.+?)\'',
thumbnail = self._search_regex(r'playlist:.*?url: \'(.+?)\'',
-
embed
_page, u'thumbnail', flags=re.DOTALL)
+
raw
_page, u'thumbnail', flags=re.DOTALL)
return {'id': video_id,
'title': title,
return {'id': video_id,
'title': title,