X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fappletrailers.py;h=ef5644aa54fe28002dc4d8c76308941c264252e3;hb=dabc127362ddfe88996e72b7e0d5cd2e4f239c98;hp=a527f10de250596e42f19f0957433e2a72fe5bbf;hpb=693b8b2d310e119417787e7b06d1e1832d16f05d;p=youtube-dl.git
diff --git a/youtube_dl/extractor/appletrailers.py b/youtube_dl/extractor/appletrailers.py
index a527f10de..ef5644aa5 100644
--- a/youtube_dl/extractor/appletrailers.py
+++ b/youtube_dl/extractor/appletrailers.py
@@ -1,5 +1,4 @@
import re
-import xml.etree.ElementTree
import json
from .common import InfoExtractor
@@ -65,18 +64,18 @@ class AppleTrailersIE(InfoExtractor):
uploader_id = mobj.group('company')
playlist_url = compat_urlparse.urljoin(url, u'includes/playlists/itunes.inc')
- playlist_snippet = self._download_webpage(playlist_url, movie)
- playlist_cleaned = re.sub(r'(?s)', u'', playlist_snippet)
- playlist_cleaned = re.sub(r'
', r'
', playlist_cleaned)
- # The ' in the onClick attributes are not escaped, it couldn't be parsed
- # with xml.etree.ElementTree.fromstring
- # like: http://trailers.apple.com/trailers/wb/gravity/
- def _clean_json(m):
- return u'iTunes.playURL(%s);' % m.group(1).replace('\'', ''')
- playlist_cleaned = re.sub(self._JSON_RE, _clean_json, playlist_cleaned)
- playlist_html = u'' + playlist_cleaned + u''
+ def fix_html(s):
+ s = re.sub(r'(?s)', u'', s)
+ s = re.sub(r'
', r'
', s)
+ # The ' in the onClick attributes are not escaped, it couldn't be parsed
+ # like: http://trailers.apple.com/trailers/wb/gravity/
+ def _clean_json(m):
+ return u'iTunes.playURL(%s);' % m.group(1).replace('\'', ''')
+ s = re.sub(self._JSON_RE, _clean_json, s)
+ s = u'' + s + u''
+ return s
+ doc = self._download_xml(playlist_url, movie, transform_source=fix_html)
- doc = xml.etree.ElementTree.fromstring(playlist_html)
playlist = []
for li in doc.findall('./div/ul/li'):
on_click = li.find('.//a').attrib['onClick']