From: Sergey M․ <dstftw@gmail.com>
Date: Sat, 23 Jul 2016 10:56:11 +0000 (+0700)
Subject: [lcp] Improve extraction
X-Git-Url: http://git.oshgnacknak.de/?a=commitdiff_plain;h=c8e170b2092f5e2ad9ea8fd7fb2eedd35e307a1c;p=youtube-dl

[lcp] Improve extraction
---

diff --git a/youtube_dl/extractor/lcp.py b/youtube_dl/extractor/lcp.py
index 38d7502df..ade27a99e 100644
--- a/youtube_dl/extractor/lcp.py
+++ b/youtube_dl/extractor/lcp.py
@@ -1,39 +1,90 @@
 # coding: utf-8
 from __future__ import unicode_literals
+
 from .common import InfoExtractor
+from .arkena import ArkenaIE
+
+
+class LcpPlayIE(ArkenaIE):
+    _VALID_URL = r'https?://play\.lcp\.fr/embed/(?P<id>[^/]+)/(?P<account_id>[^/]+)/[^/]+/[^/]+'
+    _TESTS = [{
+        'url': 'http://play.lcp.fr/embed/327336/131064/darkmatter/0',
+        'md5': 'b8bd9298542929c06c1c15788b1f277a',
+        'info_dict': {
+            'id': '327336',
+            'ext': 'mp4',
+            'title': '327336',
+            'timestamp': 1456391602,
+            'upload_date': '20160225',
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }]
+
 
 class LcpIE(InfoExtractor):
-    IE_NAME = 'LCP'
-    _VALID_URL = r'https?://(?:www\.)?lcp\.fr/(?:[^\/]+/)*(?P<id>[^/]+)'
+    _VALID_URL = r'https?://(?:www\.)?lcp\.fr/(?:[^/]+/)*(?P<id>[^/]+)'
 
     _TESTS = [{
+        # arkena embed
         'url': 'http://www.lcp.fr/la-politique-en-video/schwartzenberg-prg-preconise-francois-hollande-de-participer-une-primaire',
-        'md5': 'ab96c4dae94322ece1e98d97c8dc7807',
+        'md5': 'b8bd9298542929c06c1c15788b1f277a',
         'info_dict': {
             'id': 'd56d03e9',
-            'url': 're:http://httpod.scdn.arkena.com/11970/d56d03e9_.*',
             'ext': 'mp4',
-            'title': 'd56d03e9',
+            'title': 'Schwartzenberg (PRG) préconise à François Hollande de participer à une primaire à gauche',
+            'description': 'md5:96ad55009548da9dea19f4120c6c16a8',
+            'timestamp': 1456488895,
             'upload_date': '20160226',
-            'timestamp': 1456488895
-        }
+        },
+        'params': {
+            'skip_download': True,
+        },
     }, {
+        # dailymotion live stream
         'url': 'http://www.lcp.fr/le-direct',
         'info_dict': {
-            'title': 'Le direct | LCP Assembl\xe9e nationale',
-            'id': 'le-direct',
+            'id': 'xji3qy',
+            'ext': 'mp4',
+            'title': 'La Chaine Parlementaire (LCP), Live TNT',
+            'description': 'md5:5c69593f2de0f38bd9a949f2c95e870b',
+            'uploader': 'LCP',
+            'uploader_id': 'xbz33d',
+            'timestamp': 1308923058,
+            'upload_date': '20110624',
+        },
+        'params': {
+            # m3u8 live stream
+            'skip_download': True,
         },
-        'playlist_mincount': 1
+    }, {
+        'url': 'http://www.lcp.fr/emissions/277792-les-volontaires',
+        'only_matching': True,
     }]
 
     def _real_extract(self, url):
         display_id = self._match_id(url)
+
         webpage = self._download_webpage(url, display_id)
 
-        embed_url_regex = r'"(?P<url>(?:https?://(?:www\.)?)?play\.lcp\.fr/embed/[A-za-z0-9]+/[A-za-z0-9]+/[A-za-z0-9]+/[A-za-z0-9]+)"'
-        embed_url = self._html_search_regex(embed_url_regex, webpage, 'player_url', default=None, fatal=False)
-        if not embed_url:
+        play_url = self._search_regex(
+            r'<iframe[^>]+src=(["\'])(?P<url>%s?(?:(?!\1).)*)\1' % LcpPlayIE._VALID_URL,
+            webpage, 'play iframe', default=None, group='url')
+
+        if not play_url:
             return self.url_result(url, 'Generic')
 
-        title = self._og_search_title(webpage, default=None)
-        return self.url_result(embed_url, 'ArkenaPlay', video_id=display_id, video_title=title)
+        title = self._og_search_title(webpage, default=None) or self._html_search_meta(
+            'twitter:title', webpage, fatal=True)
+        description = self._html_search_meta(
+            ('description', 'twitter:description'), webpage)
+
+        return {
+            '_type': 'url_transparent',
+            'ie_key': LcpPlayIE.ie_key(),
+            'url': play_url,
+            'display_id': display_id,
+            'title': title,
+            'description': description,
+        }