Merge branch 'jython-support'
[youtube-dl] / youtube_dl / extractor / rtve.py
index b701a322ae1020d1bcaedcaf35876486ac5c9e16..8a8c5d2a0e2b27bd6e476ef7d848753fbe648dd5 100644 (file)
@@ -6,16 +6,19 @@ import re
 import time
 
 from .common import InfoExtractor
-from ..compat import compat_urlparse
 from ..utils import (
+    ExtractorError,
     float_or_none,
     remove_end,
+    remove_start,
+    sanitized_Request,
+    std_headers,
     struct_unpack,
 )
 
 
 def _decrypt_url(png):
-    encrypted_data = base64.b64decode(png)
+    encrypted_data = base64.b64decode(png.encode('utf-8'))
     text_index = encrypted_data.find(b'tEXt')
     text_chunk = encrypted_data[text_index - 4:]
     length = struct_unpack('!I', text_chunk[:4])[0]
@@ -84,25 +87,30 @@ class RTVEALaCartaIE(InfoExtractor):
         'only_matching': True,
     }]
 
+    def _real_initialize(self):
+        user_agent_b64 = base64.b64encode(std_headers['User-Agent'].encode('utf-8')).decode('utf-8')
+        manager_info = self._download_json(
+            'http://www.rtve.es/odin/loki/' + user_agent_b64,
+            None, 'Fetching manager info')
+        self._manager = manager_info['manager']
+
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
         info = self._download_json(
             'http://www.rtve.es/api/videos/%s/config/alacarta_videos.json' % video_id,
             video_id)['page']['items'][0]
-        png_url = 'http://www.rtve.es/ztnr/movil/thumbnail/default/videos/%s.png' % video_id
-        png = self._download_webpage(png_url, video_id, 'Downloading url information')
+        if info['state'] == 'DESPU':
+            raise ExtractorError('The video is no longer available', expected=True)
+        png_url = 'http://www.rtve.es/ztnr/movil/thumbnail/%s/videos/%s.png' % (self._manager, video_id)
+        png_request = sanitized_Request(png_url)
+        png_request.add_header('Referer', url)
+        png = self._download_webpage(png_request, video_id, 'Downloading url information')
         video_url = _decrypt_url(png)
         if not video_url.endswith('.f4m'):
-            auth_url = video_url.replace(
+            video_url = video_url.replace(
                 'resources/', 'auth/resources/'
             ).replace('.net.rtve', '.multimedia.cdn.rtve')
-            video_path = self._download_webpage(
-                auth_url, video_id, 'Getting video url')
-            # Use mvod1.akcdn instead of flash.akamaihd.multimedia.cdn to get
-            # the right Content-Length header and the mp4 format
-            video_url = compat_urlparse.urljoin(
-                'http://mvod1.akcdn.rtve.es/', video_path)
 
         subtitles = None
         if info.get('sbtFile') is not None:
@@ -126,10 +134,11 @@ class RTVEALaCartaIE(InfoExtractor):
             (s['lang'], [{'ext': 'vtt', 'url': s['src']}])
             for s in subs)
 
+
 class RTVEInfantilIE(InfoExtractor):
-    IE_NAME = 'rtve.es:alacarta'
-    IE_DESC = 'RTVE a la carta'
-    _VALID_URL = r'https?://(?:www\.)?rtve\.es/infantil/serie/(?P<show>[^/]*)/video/(?P<short_tittle>[^/]*)/(?P<id>[0-9]+)/'
+    IE_NAME = 'rtve.es:infantil'
+    IE_DESC = 'RTVE infantil'
+    _VALID_URL = r'https?://(?:www\.)?rtve\.es/infantil/serie/(?P<show>[^/]*)/video/(?P<short_title>[^/]*)/(?P<id>[0-9]+)/'
 
     _TESTS = [{
         'url': 'http://www.rtve.es/infantil/serie/cleo/video/maneras-vivir/3040283/',
@@ -141,12 +150,10 @@ class RTVEInfantilIE(InfoExtractor):
             'thumbnail': 'http://www.rtve.es/resources/jpg/6/5/1426182947956.JPG',
             'duration': 357.958,
         },
-    },]
+    }]
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-        short_tittle = mobj.group('short_tittle')
+        video_id = self._match_id(url)
         info = self._download_json(
             'http://www.rtve.es/api/videos/%s/config/alacarta_videos.json' % video_id,
             video_id)['page']['items'][0]
@@ -169,18 +176,17 @@ class RTVEInfantilIE(InfoExtractor):
         }
 
 
-
 class RTVELiveIE(InfoExtractor):
     IE_NAME = 'rtve.es:live'
     IE_DESC = 'RTVE.es live streams'
-    _VALID_URL = r'http://www\.rtve\.es/(?:deportes/directo|noticias|television)/(?P<id>[a-zA-Z0-9-]+)'
+    _VALID_URL = r'http://www\.rtve\.es/directo/(?P<id>[a-zA-Z0-9-]+)'
 
     _TESTS = [{
-        'url': 'http://www.rtve.es/noticias/directo-la-1/',
+        'url': 'http://www.rtve.es/directo/la-1/',
         'info_dict': {
-            'id': 'directo-la-1',
-            'ext': 'flv',
-            'title': 're:^La 1 de TVE [0-9]{4}-[0-9]{2}-[0-9]{2}Z[0-9]{6}$',
+            'id': 'la-1',
+            'ext': 'mp4',
+            'title': 're:^La 1 [0-9]{4}-[0-9]{2}-[0-9]{2}Z[0-9]{6}$',
         },
         'params': {
             'skip_download': 'live stream',
@@ -193,23 +199,20 @@ class RTVELiveIE(InfoExtractor):
         video_id = mobj.group('id')
 
         webpage = self._download_webpage(url, video_id)
-        player_url = self._search_regex(
-            r'<param name="movie" value="([^"]+)"/>', webpage, 'player URL')
-        title = remove_end(self._og_search_title(webpage), ' en directo')
+        title = remove_end(self._og_search_title(webpage), ' en directo en RTVE.es')
+        title = remove_start(title, 'Estoy viendo ')
         title += ' ' + time.strftime('%Y-%m-%dZ%H%M%S', start_time)
 
         vidplayer_id = self._search_regex(
-            r' id="vidplayer([0-9]+)"', webpage, 'internal video ID')
-        png_url = 'http://www.rtve.es/ztnr/movil/thumbnail/default/videos/%s.png' % vidplayer_id
+            r'playerId=player([0-9]+)', webpage, 'internal video ID')
+        png_url = 'http://www.rtve.es/ztnr/movil/thumbnail/amonet/videos/%s.png' % vidplayer_id
         png = self._download_webpage(png_url, video_id, 'Downloading url information')
-        video_url = _decrypt_url(png)
+        m3u8_url = _decrypt_url(png)
+        formats = self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4')
 
         return {
             'id': video_id,
-            'ext': 'flv',
             'title': title,
-            'url': video_url,
-            'app': 'rtve-live-live?ovpfv=2.1.2',
-            'player_url': player_url,
-            'rtmp_live': True,
+            'formats': formats,
+            'is_live': True,
         }