Merge remote-tracking branch 'JGjorgji/fix-leading-zeroes'

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

diff --git a/youtube_dl/extractor/vimeo.py b/youtube_dl/extractor/vimeo.py
index 2e1a09a0b7c0caa7a8947eab2be895b607e6b7ca..55f6cd0d8e1d9e7de2fd49f0e123b23e47496f13 100644 (file)
--- a/youtube_dl/extractor/vimeo.py
+++ b/youtube_dl/extractor/vimeo.py
@@ -18,6 +18,7 @@ from ..utils import (
     std_headers,
     unsmuggle_url,
     urlencode_postdata,
+    int_or_none,
 )
 
 
@@ -74,6 +75,7 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
                 "uploader_id": "user7108434",
                 "uploader": "Filippo Valsorda",
                 "title": "youtube-dl test video - \u2605 \" ' \u5e78 / \\ \u00e4 \u21ad \U0001d550",
+                "duration": 10,
             },
         },
         {
@@ -86,6 +88,7 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
                 'uploader_id': 'openstreetmapus',
                 'uploader': 'OpenStreetMap US',
                 'title': 'Andy Allan - Putting the Carto into OpenStreetMap Cartography',
+                'duration': 1595,
             },
         },
         {
@@ -95,9 +98,10 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
             'info_dict': {
                 'id': '54469442',
                 'ext': 'mp4',
-                'title': 'Kathy Sierra: Building the minimum Badass User, Business of Software',
+                'title': 'Kathy Sierra: Building the minimum Badass User, Business of Software 2012',
                 'uploader': 'The BLN & Business of Software',
                 'uploader_id': 'theblnbusinessofsoftware',
+                'duration': 3610,
             },
         },
         {
@@ -111,11 +115,27 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
                 'upload_date': '20130614',
                 'uploader_id': 'user18948128',
                 'uploader': 'Jaime Marquínez Ferrándiz',
+                'duration': 10,
             },
             'params': {
                 'videopassword': 'youtube-dl',
             },
         },
+        {
+            'url': 'http://vimeo.com/channels/keypeele/75629013',
+            'md5': '2f86a05afe9d7abc0b9126d229bbe15d',
+            'note': 'Video is freely available via original URL '
+                    'and protected with password when accessed via http://vimeo.com/75629013',
+            'info_dict': {
+                'id': '75629013',
+                'ext': 'mp4',
+                'title': 'Key & Peele: Terrorist Interrogation',
+                'description': 'md5:8678b246399b070816b12313e8b4eb5c',
+                'uploader_id': 'atencio',
+                'uploader': 'Peter Atencio',
+                'duration': 187,
+            },
+        },
         {
             'url': 'http://vimeo.com/76979871',
             'md5': '3363dd6ffebe3784d56f4132317fd446',
@@ -128,8 +148,22 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
                 'upload_date': '20131015',
                 'uploader_id': 'staff',
                 'uploader': 'Vimeo Staff',
+                'duration': 62,
             }
         },
+        {
+            'note': 'video player needs Referer',
+            'url': 'http://vimeo.com/user22258446/review/91613211/13f927e053',
+            'md5': '6295fdab8f4bf6a002d058b2c6dce276',
+            'info_dict': {
+                'id': '91613211',
+                'ext': 'mp4',
+                'title': 'Death by dogma versus assembling agile - Sander Hoogendoorn',
+                'uploader': 'DevWeek Events',
+                'duration': 2773,
+                'thumbnail': 're:^https?://.*\.jpg$',
+            }
+        }
     ]
 
     @classmethod
@@ -184,14 +218,14 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
         if data is not None:
             headers = headers.copy()
             headers.update(data)
+        if 'Referer' not in headers:
+            headers['Referer'] = url
 
         # Extract ID from URL
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
         if mobj.group('pro') or mobj.group('player'):
             url = 'http://player.vimeo.com/video/' + video_id
-        else:
-            url = 'https://vimeo.com/' + video_id
 
         # Retrieve video webpage to extract further information
         request = compat_urllib_request.Request(url, None, headers)
@@ -257,7 +291,7 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
         if video_thumbnail is None:
             video_thumbs = config["video"].get("thumbs")
             if video_thumbs and isinstance(video_thumbs, dict):
-                _, video_thumbnail = sorted((int(width), t_url) for (width, t_url) in video_thumbs.items())[-1]
+                _, video_thumbnail = sorted((int(width if width.isdigit() else 0), t_url) for (width, t_url) in video_thumbs.items())[-1]
 
         # Extract video description
         video_description = None
@@ -273,6 +307,9 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
             else:
                 raise
 
+        # Extract video duration
+        video_duration = int_or_none(config["video"].get("duration"))
+
         # Extract upload date
         video_upload_date = None
         mobj = re.search(r'<meta itemprop="dateCreated" content="(\d{4})-(\d{2})-(\d{2})T', webpage)
@@ -344,6 +381,7 @@ class VimeoIE(VimeoBaseInfoExtractor, SubtitlesInfoExtractor):
             'title': video_title,
             'thumbnail': video_thumbnail,
             'description': video_description,
+            'duration': video_duration,
             'formats': formats,
             'webpage_url': url,
             'view_count': view_count,