[youtube] Fix extraction.

[youtube-dl] / youtube_dl / extractor / la7.py

diff --git a/youtube_dl/extractor/la7.py b/youtube_dl/extractor/la7.py
index db2028e9f5148d37f69f0d7c4c41fbbb77d88928..c3b4ffa7e3f523bf9e0d10ad4660e72d18078b6f 100644 (file)
--- a/youtube_dl/extractor/la7.py
+++ b/youtube_dl/extractor/la7.py
@@ -1,63 +1,67 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
-import re
-
 from .common import InfoExtractor
 from ..utils import (
-    parse_duration,
+    js_to_json,
+    smuggle_url,
 )
 
 
 class LA7IE(InfoExtractor):
-    IE_NAME = 'la7.tv'
-    _VALID_URL = r'''(?x)
-        https?://(?:www\.)?la7\.tv/
-        (?:
-            richplayer/\?assetid=|
-            \?contentId=
-        )
-        (?P<id>[0-9]+)'''
-
-    _TEST = {
-        'url': 'http://www.la7.tv/richplayer/?assetid=50355319',
-        'file': '50355319.mp4',
-        'md5': 'ec7d1f0224d20ba293ab56cf2259651f',
+    IE_NAME = 'la7.it'
+    _VALID_URL = r'''(?x)(https?://)?(?:
+        (?:www\.)?la7\.it/([^/]+)/(?:rivedila7|video)/|
+        tg\.la7\.it/repliche-tgla7\?id=
+    )(?P<id>.+)'''
+
+    _TESTS = [{
+        # 'src' is a plain URL
+        'url': 'http://www.la7.it/crozza/video/inccool8-02-10-2015-163722',
+        'md5': '8b613ffc0c4bf9b9e377169fc19c214c',
+        'info_dict': {
+            'id': '0_42j6wd36',
+            'ext': 'mp4',
+            'title': 'Inc.Cool8',
+            'description': 'Benvenuti nell\'incredibile mondo della INC. COOL. 8. dove “INC.” sta per “Incorporated” “COOL” sta per “fashion” ed Eight sta per il gesto  atletico',
+            'thumbnail': 're:^https?://.*',
+            'uploader_id': 'kdla7pillole@iltrovatore.it',
+            'timestamp': 1443814869,
+            'upload_date': '20151002',
+        },
+    }, {
+        # 'src' is a dictionary
+        'url': 'http://tg.la7.it/repliche-tgla7?id=189080',
+        'md5': '6b0d8888d286e39870208dfeceaf456b',
         'info_dict': {
-            'title': 'IL DIVO',
-            'description': 'Un film di Paolo Sorrentino con Toni Servillo, Anna Bonaiuto, Giulio Bosetti  e Flavio Bucci',
-            'duration': 6254,
+            'id': '189080',
+            'ext': 'mp4',
+            'title': 'TG LA7',
         },
-        'skip': 'Blocked in the US',
-    }
+    }, {
+        'url': 'http://www.la7.it/omnibus/rivedila7/omnibus-news-02-07-2016-189077',
+        'only_matching': True,
+    }]
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-
-        xml_url = 'http://www.la7.tv/repliche/content/index.php?contentId=%s' % video_id
-        doc = self._download_xml(xml_url, video_id)
-
-        video_title = doc.find('title').text
-        description = doc.find('description').text
-        duration = parse_duration(doc.find('duration').text)
-        thumbnail = doc.find('img').text
-        view_count = int(doc.find('views').text)
+        video_id = self._match_id(url)
 
-        prefix = doc.find('.//fqdn').text.strip().replace('auto:', 'http:')
+        webpage = self._download_webpage(url, video_id)
 
-        formats = [{
-            'format': vnode.find('quality').text,
-            'tbr': int(vnode.find('quality').text),
-            'url': vnode.find('fms').text.strip().replace('mp4:', prefix),
-        } for vnode in doc.findall('.//videos/video')]
-        self._sort_formats(formats)
+        player_data = self._parse_json(
+            self._search_regex(
+                [r'(?s)videoParams\s*=\s*({.+?});', r'videoLa7\(({[^;]+})\);'],
+                webpage, 'player data'),
+            video_id, transform_source=js_to_json)
 
         return {
+            '_type': 'url_transparent',
+            'url': smuggle_url('kaltura:103:%s' % player_data['vid'], {
+                'service_url': 'http://nkdam.iltrovatore.it',
+            }),
             'id': video_id,
-            'title': video_title,
-            'description': description,
-            'thumbnail': thumbnail,
-            'duration': duration,
-            'formats': formats,
-            'view_count': view_count,
+            'title': player_data['title'],
+            'description': self._og_search_description(webpage, default=None),
+            'thumbnail': player_data.get('poster'),
+            'ie_key': 'Kaltura',
         }