GametrailersIE: support multipart videos

Use xml.etree.ElementTree instead of re when possible

youtube_dl/extractor/gametrailers.py

@@ -1,4 +1,5 @@
 import re
+import xml.etree.ElementTree
 
 from .common import InfoExtractor
 from ..utils import (
@@ -11,7 +12,7 @@ class GametrailersIE(InfoExtractor):
     _VALID_URL = r'http://www.gametrailers.com/(?P<type>videos|reviews|full-episodes)/(?P<id>.*?)/(?P<title>.*)'
     _TEST = {
         u'url': u'http://www.gametrailers.com/videos/zbvr8i/mirror-s-edge-2-e3-2013--debut-trailer',
-        u'file': u'zbvr8i.flv',
+        u'file': u'70e9a5d7-cf25-4a10-9104-6f3e7342ae0d.flv',
         u'md5': u'c3edbc995ab4081976e16779bd96a878',
         u'info_dict': {
             u"title": u"E3 2013: Debut Trailer"
@@ -24,45 +25,39 @@ class GametrailersIE(InfoExtractor):
         if mobj is None:
             raise ExtractorError(u'Invalid URL: %s' % url)
         video_id = mobj.group('id')
-        video_type = mobj.group('type')
         webpage = self._download_webpage(url, video_id)
-        if video_type == 'full-episodes':
+        mgid = self._search_regex([r'data-video="(?P<mgid>mgid:.*?)"',
-            mgid_re = r'data-video="(?P<mgid>mgid:.*?)"'
+                                   r'data-contentId=\'(?P<mgid>mgid:.*?)\''],
-        else:
+                                  webpage, u'mgid')
-            mgid_re = r'data-contentId=\'(?P<mgid>mgid:.*?)\''
-        mgid = self._search_regex(mgid_re, webpage, u'mgid')
-        data = compat_urllib_parse.urlencode({'uri': mgid, 'acceptMethods': 'fms'})
 
+        data = compat_urllib_parse.urlencode({'uri': mgid, 'acceptMethods': 'fms'})
         info_page = self._download_webpage('http://www.gametrailers.com/feeds/mrss?' + data,
                                            video_id, u'Downloading video info')
-        links_webpage = self._download_webpage('http://www.gametrailers.com/feeds/mediagen/?' + data,
+        doc = xml.etree.ElementTree.fromstring(info_page.encode('utf-8'))
+        default_thumb = doc.find('./channel/image/url').text
+
+        media_namespace = {'media': 'http://search.yahoo.com/mrss/'}
+        parts = [{
+            'title': video_doc.find('title').text,
+            'ext': 'flv',
+            'id': video_doc.find('guid').text.rpartition(':')[2],
+            # Videos are actually flv not mp4
+            'url': self._get_video_url(video_doc.find('media:group/media:content', media_namespace).attrib['url'], video_id),
+            # The thumbnail may not be defined, it would be ''
+            'thumbnail': video_doc.find('media:group/media:thumbnail', media_namespace).attrib['url'] or default_thumb,
+            'description': video_doc.find('description').text,
+        } for video_doc in doc.findall('./channel/item')]
+        return parts
+
+    def _get_video_url(self, mediagen_url, video_id):
+        if 'acceptMethods' not in mediagen_url:
+            mediagen_url += '&acceptMethods=fms'
+        links_webpage = self._download_webpage(mediagen_url,
                                                video_id, u'Downloading video urls info')
-
+        doc = xml.etree.ElementTree.fromstring(links_webpage)
-        self.report_extraction(video_id)
+        urls = list(doc.iter('src'))
-        info_re = r'''<title><!\[CDATA\[(?P<title>.*?)\]\]></title>.*
+        if len(urls) == 0:
-                      <description><!\[CDATA\[(?P<description>.*?)\]\]></description>.*
-                      <image>.*
-                        <url>(?P<thumb>.*?)</url>.*
-                      </image>'''
-
-        m_info = re.search(info_re, info_page, re.VERBOSE|re.DOTALL)
-        if m_info is None:
-            raise ExtractorError(u'Unable to extract video info')
-        video_title = m_info.group('title')
-        video_description = m_info.group('description')
-        video_thumb = m_info.group('thumb')
-
-        m_urls = list(re.finditer(r'<src>(?P<url>.*)</src>', links_webpage))
-        if m_urls is None or len(m_urls) == 0:
             raise ExtractorError(u'Unable to extract video url')
         # They are sorted from worst to best quality
-        video_url = m_urls[-1].group('url')
+        return urls[-1].text
 
-        return {'url':         video_url,
-                'id':          video_id,
-                'title':       video_title,
-                # Videos are actually flv not mp4
-                'ext':         'flv',
-                'thumbnail':   video_thumb,
-                'description': video_description,
-                }