1 files changed, 63 insertions, 0 deletions
diff --git a/youtube_dl/extractor/mdr.py b/youtube_dl/extractor/mdr.py
new file mode 100644
index 000000000..08ce0647f
--- /dev/null
+++ b/youtube_dl/extractor/mdr.py
@@ -0,0 +1,63 @@
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    ExtractorError,
+)
+
+
+class MDRIE(InfoExtractor):
+    _VALID_URL = r'^(?P<domain>(?:https?://)?(?:www\.)?mdr\.de)/mediathek/(?:.*)/(?P<type>video|audio)(?P<video_id>[^/_]+)_.*'
+    
+    # No tests, MDR regularily deletes its videos
+
+    def _real_extract(self, url):
+        m = re.match(self._VALID_URL, url)
+        video_id = m.group('video_id')
+        domain = m.group('domain')
+
+        # determine title and media streams from webpage
+        html = self._download_webpage(url, video_id)
+
+        title = self._html_search_regex(r'<h2>(.*?)</h2>', html, u'title')
+        xmlurl = self._search_regex(
+            r'(/mediathek/(?:.+)/(?:video|audio)[0-9]+-avCustom.xml)', html, u'XML URL')
+
+        doc = self._download_xml(domain + xmlurl, video_id)
+        formats = []
+        for a in doc.findall('./assets/asset'):
+            url_el = a.find('.//progressiveDownloadUrl')
+            if url_el is None:
+                continue
+            abr = int(a.find('bitrateAudio').text) // 1000
+            media_type = a.find('mediaType').text
+            format = {
+                'abr': abr,
+                'filesize': int(a.find('fileSize').text),
+                'url': url_el.text,
+            }
+
+            vbr_el = a.find('bitrateVideo')
+            if vbr_el is None:
+                format.update({
+                    'vcodec': 'none',
+                    'format_id': u'%s-%d' % (media_type, abr),
+                })
+            else:
+                vbr = int(vbr_el.text) // 1000
+                format.update({
+                    'vbr': vbr,
+                    'width': int(a.find('frameWidth').text),
+                    'height': int(a.find('frameHeight').text),
+                    'format_id': u'%s-%d' % (media_type, vbr),
+                })
+            formats.append(format)
+        formats.sort(key=lambda f: (f.get('vbr'), f['abr']))
+        if not formats:
+            raise ExtractorError(u'Could not find any valid formats')
+
+        return {
+            'id': video_id,
+            'title': title,
+            'formats': formats,
+        }