[yahoo] Fix video extraction (fixes #1521)

There's no need to use two different methods.
Now we can also download videos over http if possible.
Also run the test for rtmp videos, but skip the download.

youtube_dl/extractor/yahoo.py

@@ -1,4 +1,3 @@
-import datetime
 import itertools
 import json
 import re
@@ -6,86 +5,85 @@ import re
 from .common import InfoExtractor, SearchInfoExtractor
 from ..utils import (
     compat_urllib_parse,
-
+    compat_urlparse,
-    ExtractorError,
+    determine_ext,
+    clean_html,
 )
 
+
 class YahooIE(InfoExtractor):
     IE_DESC = u'Yahoo screen'
     _VALID_URL = r'http://screen\.yahoo\.com/.*?-(?P<id>\d*?)\.html'
-    _TEST = {
+    _TESTS = [
-        u'url': u'http://screen.yahoo.com/julian-smith-travis-legg-watch-214727115.html',
+        {
-        u'file': u'214727115.flv',
+            u'url': u'http://screen.yahoo.com/julian-smith-travis-legg-watch-214727115.html',
-        u'md5': u'2e717f169c1be93d84d3794a00d4a325',
+            u'file': u'214727115.mp4',
-        u'info_dict': {
+            u'info_dict': {
-            u"title": u"Julian Smith & Travis Legg Watch Julian Smith"
+                u'title': u'Julian Smith & Travis Legg Watch Julian Smith',
+                u'description': u'Julian and Travis watch Julian Smith',
+            },
         },
-        u'skip': u'Requires rtmpdump'
+        {
-    }
+            u'url': u'http://screen.yahoo.com/wired/codefellas-s1-ep12-cougar-lies-103000935.html',
+            u'file': u'103000935.flv',
+            u'info_dict': {
+                u'title': u'The Cougar Lies with Spanish Moss',
+                u'description': u'Agent Topple\'s mustache does its dirty work, and Nicole brokers a deal for peace. But why is the NSA collecting millions of Instagram brunch photos? And if your waffles have nothing to hide, what are they so worried about?',
+            },
+            u'params': {
+                # Requires rtmpdump
+                u'skip_download': True,
+            },
+        },
+    ]
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        if mobj is None:
-            raise ExtractorError(u'Invalid URL: %s' % url)
         video_id = mobj.group('id')
         webpage = self._download_webpage(url, video_id)
-        m_id = re.search(r'YUI\.namespace\("Media"\)\.CONTENT_ID = "(?P<new_id>.+?)";', webpage)
 
-        if m_id is None: 
+        items_json = self._search_regex(r'YVIDEO_INIT_ITEMS = ({.*?});$',
-            # TODO: Check which url parameters are required
+            webpage, u'items', flags=re.MULTILINE)
-            info_url = 'http://cosmos.bcst.yahoo.com/rest/v2/pops;lmsoverride=1;outputformat=mrss;cb=974419660;id=%s;rd=news.yahoo.com;datacontext=mdb;lg=KCa2IihxG3qE60vQ7HtyUy' % video_id
+        items = json.loads(items_json)
-            webpage = self._download_webpage(info_url, video_id, u'Downloading info webpage')
+        info = items['mediaItems']['query']['results']['mediaObj'][0]
-            info_re = r'''<title><!\[CDATA\[(?P<title>.*?)\]\]></title>.*
+        meta = info['meta']
-                        <description><!\[CDATA\[(?P<description>.*?)\]\]></description>.*
-                        <media:pubStart><!\[CDATA\[(?P<date>.*?)\ .*\]\]></media:pubStart>.*
-                        <media:content\ medium="image"\ url="(?P<thumb>.*?)"\ name="LARGETHUMB"
-                        '''
-            self.report_extraction(video_id)
-            m_info = re.search(info_re, webpage, re.VERBOSE|re.DOTALL)
-            if m_info is None:
-                raise ExtractorError(u'Unable to extract video info')
-            video_title = m_info.group('title')
-            video_description = m_info.group('description')
-            video_thumb = m_info.group('thumb')
-            video_date = m_info.group('date')
-            video_date = datetime.datetime.strptime(video_date, '%m/%d/%Y').strftime('%Y%m%d')
-    
-            # TODO: Find a way to get mp4 videos
-            rest_url = 'http://cosmos.bcst.yahoo.com/rest/v2/pops;element=stream;outputformat=mrss;id=%s;lmsoverride=1;bw=375;dynamicstream=1;cb=83521105;tech=flv,mp4;rd=news.yahoo.com;datacontext=mdb;lg=KCa2IihxG3qE60vQ7HtyUy' % video_id
-            webpage = self._download_webpage(rest_url, video_id, u'Downloading video url webpage')
-            m_rest = re.search(r'<media:content url="(?P<url>.*?)" path="(?P<path>.*?)"', webpage)
-            video_url = m_rest.group('url')
-            video_path = m_rest.group('path')
-            if m_rest is None:
-                raise ExtractorError(u'Unable to extract video url')
 
-        else: # We have to use a different method if another id is defined
+        formats = []
-            long_id = m_id.group('new_id')
+        for s in info['streams']:
-            info_url = 'http://video.query.yahoo.com/v1/public/yql?q=SELECT%20*%20FROM%20yahoo.media.video.streams%20WHERE%20id%3D%22' + long_id + '%22%20AND%20format%3D%22mp4%2Cflv%22%20AND%20protocol%3D%22rtmp%2Chttp%22%20AND%20plrs%3D%2286Gj0vCaSzV_Iuf6hNylf2%22%20AND%20acctid%3D%22389%22%20AND%20plidl%3D%22%22%20AND%20pspid%3D%22792700001%22%20AND%20offnetwork%3D%22false%22%20AND%20site%3D%22ivy%22%20AND%20lang%3D%22en-US%22%20AND%20region%3D%22US%22%20AND%20override%3D%22none%22%3B&env=prod&format=json&callback=YUI.Env.JSONP.yui_3_8_1_1_1368368376830_335'
+            format_info = {
-            webpage = self._download_webpage(info_url, video_id, u'Downloading info json')
+                'width': s.get('width'),
-            json_str = re.search(r'YUI.Env.JSONP.yui.*?\((.*?)\);', webpage).group(1)
+                'height': s.get('height'),
-            info = json.loads(json_str)
+                'bitrate': s.get('bitrate'),
-            res = info[u'query'][u'results'][u'mediaObj'][0]
+            }
-            stream = res[u'streams'][0]
+
-            video_path = stream[u'path']
+            host = s['host']
-            video_url = stream[u'host']
+            path = s['path']
-            meta = res[u'meta']
+            if host.startswith('rtmp'):
-            video_title = meta[u'title']
+                format_info.update({
-            video_description = meta[u'description']
+                    'url': host,
-            video_thumb = meta[u'thumbnail']
+                    'play_path': path,
-            video_date = None # I can't find it
+                    'ext': 'flv',
+                })
+            else:
+                format_url = compat_urlparse.urljoin(host, path)
+                format_info['url'] = format_url
+                format_info['ext'] = determine_ext(format_url)
+                
+            formats.append(format_info)
+        formats = sorted(formats, key=lambda f:(f['height'], f['width']))
+
+        info = {
+            'id': video_id,
+            'title': meta['title'],
+            'formats': formats,
+            'description': clean_html(meta['description']),
+            'thumbnail': meta['thumbnail'],
+        }
+        # TODO: Remove when #980 has been merged
+        info.update(formats[-1])
+
+        return info
 
-        info_dict = {
-                     'id': video_id,
-                     'url': video_url,
-                     'play_path': video_path,
-                     'title':video_title,
-                     'description': video_description,
-                     'thumbnail': video_thumb,
-                     'upload_date': video_date,
-                     'ext': 'flv',
-                     }
-        return info_dict
 
 class YahooSearchIE(SearchInfoExtractor):
     IE_DESC = u'Yahoo screen search'