Add an extractor for tlc.de (fixes #2748)
authorJaime Marquínez Ferrándiz <jaime.marquinez.ferrandiz@gmail.com>
Wed, 16 Apr 2014 13:45:05 +0000 (15:45 +0200)
committerJaime Marquínez Ferrándiz <jaime.marquinez.ferrandiz@gmail.com>
Wed, 16 Apr 2014 13:45:05 +0000 (15:45 +0200)
youtube_dl/extractor/__init__.py
youtube_dl/extractor/tlc.py [new file with mode: 0644]

index 3a91e1a465a1f7d0dcd2b169d7dc514d41b055c7..ae5296d90fd495d1dfaece8584c6ef7eeb10f8d5 100644 (file)
@@ -251,6 +251,7 @@ from .tf1 import TF1IE
 from .theplatform import ThePlatformIE
 from .thisav import ThisAVIE
 from .tinypic import TinyPicIE
+from .tlc import TlcDeIE
 from .toutv import TouTvIE
 from .toypics import ToypicsUserIE, ToypicsIE
 from .traileraddict import TrailerAddictIE
diff --git a/youtube_dl/extractor/tlc.py b/youtube_dl/extractor/tlc.py
new file mode 100644 (file)
index 0000000..0a092ea
--- /dev/null
@@ -0,0 +1,42 @@
+# encoding: utf-8
+from __future__ import unicode_literals
+import re
+
+from .common import InfoExtractor
+from .brightcove import BrightcoveIE
+
+
+class TlcDeIE(InfoExtractor):
+    IE_NAME = 'tlc.de'
+    _VALID_URL = r'http://www\.tlc\.de/sendungen/[^/]+/videos/(?P<title>[^/?]+)'
+
+    _TEST = {
+        'url': 'http://www.tlc.de/sendungen/breaking-amish/videos/#3235167922001',
+        'info_dict': {
+            'id': '3235167922001',
+            'ext': 'mp4',
+            'title': 'Breaking Amish: Die Welt da draußen',
+            'uploader': 'Discovery Networks - Germany',
+            'description': 'Vier Amische und eine Mennonitin wagen in New York'
+                '  den Sprung in ein komplett anderes Leben. Begleitet sie auf'
+                ' ihrem spannenden Weg.',
+        },
+    }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        title = mobj.group('title')
+        webpage = self._download_webpage(url, title)
+        iframe_url = self._search_regex(
+            '<iframe src="(http://www\.tlc\.de/wp-content/.+?)"', webpage,
+            'iframe url')
+        # Otherwise we don't get the correct 'BrightcoveExperience' element,
+        # example: http://www.tlc.de/sendungen/cake-boss/videos/cake-boss-cannoli-drama/
+        iframe_url = iframe_url.replace('.htm?', '.php?')
+        iframe = self._download_webpage(iframe_url, title)
+
+        return {
+            '_type': 'url',
+            'url': BrightcoveIE._extract_brightcove_url(iframe),
+            'ie': BrightcoveIE.ie_key(),
+        }