]> gitweb @ CieloNegro.org - youtube-dl.git/commitdiff
[thisav] Add IE (Fixes #1056)
authorPhilipp Hagemeister <phihag@phihag.de>
Wed, 17 Jul 2013 09:16:53 +0000 (11:16 +0200)
committerPhilipp Hagemeister <phihag@phihag.de>
Wed, 17 Jul 2013 09:16:53 +0000 (11:16 +0200)
youtube_dl/extractor/__init__.py
youtube_dl/extractor/thisav.py [new file with mode: 0644]

index fea7513f9df3a4bc35e6e1e2b761de031ffc3060..6738c41686590d8a12114c5f4ee3c64aae1e5fc9 100644 (file)
@@ -58,6 +58,7 @@ from .steam import SteamIE
 from .teamcoco import TeamcocoIE
 from .ted import TEDIE
 from .tf1 import TF1IE
+from .thisav import ThisAVIE
 from .traileraddict import TrailerAddictIE
 from .tudou import TudouIE
 from .tumblr import TumblrIE
diff --git a/youtube_dl/extractor/thisav.py b/youtube_dl/extractor/thisav.py
new file mode 100644 (file)
index 0000000..35b7379
--- /dev/null
@@ -0,0 +1,46 @@
+#coding: utf-8
+
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    determine_ext,
+)
+
+class ThisAVIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?thisav\.com/video/(?P<id>[0-9]+)/.*'
+    _TEST = {
+        u"url": u"http://www.thisav.com/video/47734/%98%26sup1%3B%83%9E%83%82---just-fit.html",
+        u"file": u"47734.flv",
+        u"info_dict": {
+            u"title": u"高樹マリア - Just fit",
+            u"uploader": u"dj7970",
+            u"uploader_id": u"dj7970"
+        }
+    }
+
+    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'<h1>([^<]*)</h1>', webpage, u'title')
+        video_url = self._html_search_regex(
+            r"addVariable\('file','([^']+)'\);", webpage, u'video url')
+        uploader = self._html_search_regex(
+            r': <a href="http://www.thisav.com/user/[0-9]+/(?:[^"]+)">([^<]+)</a>',
+            webpage, u'uploader name', fatal=False)
+        uploader_id = self._html_search_regex(
+            r': <a href="http://www.thisav.com/user/[0-9]+/([^"]+)">(?:[^<]+)</a>',
+            webpage, u'uploader id', fatal=False)
+        ext = determine_ext(video_url)
+        
+        return {
+            '_type':       'video',
+            'id':          video_id,
+            'url':         video_url,
+            'uploader':    uploader,
+            'uploader_id': uploader_id,
+            'title':       title,
+            'ext':         ext,
+        }