Merge remote-tracking branch 'alphapapa/master'

[youtube-dl] / youtube_dl / extractor / instagram.py

diff --git a/youtube_dl/extractor/instagram.py b/youtube_dl/extractor/instagram.py
index 1ffadf67f120f60d57ec367635701fc709232a4b..ddc42882a436a216cbd24b0b28d03da89ec27b0d 100644 (file)
--- a/youtube_dl/extractor/instagram.py
+++ b/youtube_dl/extractor/instagram.py
@@ -5,12 +5,13 @@ from .common import InfoExtractor
 class InstagramIE(InfoExtractor):
     _VALID_URL = r'(?:http://)?instagram.com/p/(.*?)/'
     _TEST = {
-        u'url': u'http://instagram.com/p/aye83DjauH/#',
+        u'url': u'http://instagram.com/p/aye83DjauH/?foo=bar#abc',
         u'file': u'aye83DjauH.mp4',
         u'md5': u'0d2da106a9d2631273e192b372806516',
         u'info_dict': {
             u"uploader_id": u"naomipq", 
-            u"title": u"Video by naomipq"
+            u"title": u"Video by naomipq",
+            u'description': u'md5:1f17f0ab29bd6fe2bfad705f58de3cb8',
         }
     }
 
@@ -18,19 +19,17 @@ class InstagramIE(InfoExtractor):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group(1)
         webpage = self._download_webpage(url, video_id)
-        html_title = self._html_search_regex(
-            r'<title>(.+?)</title>',
-            webpage, u'title', flags=re.DOTALL)
-        title = re.sub(u'(?: *\(Videos?\))? \u2022 Instagram$', '', html_title).strip()
-        uploader_id = self._html_search_regex(r'content="(.*?)\'s video on Instagram',
-            webpage, u'uploader name', fatal=False)
-        ext = 'mp4'
+        uploader_id = self._search_regex(r'"owner":{"username":"(.+?)"',
+            webpage, u'uploader id', fatal=False)
+        desc = self._search_regex(r'"caption":"(.*?)"', webpage, u'description',
+            fatal=False)
 
         return [{
             'id':        video_id,
             'url':       self._og_search_video_url(webpage),
-            'ext':       ext,
-            'title':     title,
+            'ext':       'mp4',
+            'title':     u'Video by %s' % uploader_id,
             'thumbnail': self._og_search_thumbnail(webpage),
-            'uploader_id' : uploader_id
+            'uploader_id' : uploader_id,
+            'description': desc,
         }]