[tutv] Add IE (Fixes #965)

diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
index 35b619ccad8f7edf0f204119514cd1e4de119782..98733e39465d6c865d8a0237dea73099f595ad39 100644 (file)
--- a/youtube_dl/extractor/__init__.py
+++ b/youtube_dl/extractor/__init__.py
@@ -50,6 +50,7 @@ from .ted import TEDIE
 from .tf1 import TF1IE
 from .tudou import TudouIE
 from .tumblr import TumblrIE
+from .tutv import TutvIE
 from .ustream import UstreamIE
 from .vbox7 import Vbox7IE
 from .vevo import VevoIE

diff --git a/youtube_dl/extractor/tutv.py b/youtube_dl/extractor/tutv.py
new file mode 100644 (file)
index 0000000..8afe483
--- /dev/null
+++ b/youtube_dl/extractor/tutv.py
@@ -0,0 +1,41 @@
+import base64
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    compat_parse_qs,
+)
+
+class TutvIE(InfoExtractor):
+    _VALID_URL=r'https?://(?:www\.)?tu\.tv/videos/(?P<id>[^/?]+)'
+    _TEST = {
+        u'url': u'http://tu.tv/videos/noah-en-pabellon-cuahutemoc',
+        u'file': u'2742556.flv',
+        u'md5': u'5eb766671f69b82e528dc1e7769c5cb2',
+        u'info_dict': {
+            u"title": u"Noah en pabellon cuahutemoc"
+        }
+    }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id')
+
+        webpage = self._download_webpage(url, video_id)
+        title = self._html_search_regex(
+            r'<meta property="og:title" content="(.*?)">', webpage, u'title')
+        internal_id = self._search_regex(r'codVideo=([0-9]+)', webpage, u'internal video ID')
+
+        data_url = u'http://tu.tv/flvurl.php?codVideo=' + str(internal_id)
+        data_content = self._download_webpage(data_url, video_id, note=u'Downloading video info')
+        data = compat_parse_qs(data_content)
+        video_url = base64.b64decode(data['kpt'][0])
+        ext = video_url.partition('?')[0].rpartition('.')[2]
+
+        info = {
+            'id': internal_id,
+            'url': video_url,
+            'ext': ext,
+            'title': title,
+        }
+        return [info]