[mitele] fix extraction(closes #21354)(closes #23456)

youtube_dl/extractor/mitele.py

@@ -4,8 +4,8 @@ from __future__ import unicode_literals
 from .common import InfoExtractor
 from ..utils import (
     int_or_none,
+    parse_iso8601,
     smuggle_url,
-    parse_duration,
 )
 
 
@@ -18,16 +18,18 @@ class MiTeleIE(InfoExtractor):
         'info_dict': {
             'id': 'FhYW1iNTE6J6H7NkQRIEzfne6t2quqPg',
             'ext': 'mp4',
-            'title': 'Tor, la web invisible',
+            'title': 'Diario de La redacción Programa 144',
-            'description': 'md5:3b6fce7eaa41b2d97358726378d9369f',
+            'description': 'md5:07c35a7b11abb05876a6a79185b58d27',
             'series': 'Diario de',
-            'season': 'La redacción',
+            'season': 'Season 14',
             'season_number': 14,
-            'season_id': 'diario_de_t14_11981',
+            'episode': 'Tor, la web invisible',
-            'episode': 'Programa 144',
             'episode_number': 3,
             'thumbnail': r're:(?i)^https?://.*\.jpg$',
             'duration': 2913,
+            'age_limit': 16,
+            'timestamp': 1471209401,
+            'upload_date': '20160814',
         },
         'add_ie': ['Ooyala'],
     }, {
@@ -39,13 +41,15 @@ class MiTeleIE(InfoExtractor):
             'title': 'Cuarto Milenio Temporada 6 Programa 226',
             'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f',
             'series': 'Cuarto Milenio',
-            'season': 'Temporada 6',
+            'season': 'Season 6',
             'season_number': 6,
-            'season_id': 'cuarto_milenio_t06_12715',
+            'episode': 'Episode 24',
-            'episode': 'Programa 226',
             'episode_number': 24,
             'thumbnail': r're:(?i)^https?://.*\.jpg$',
             'duration': 7313,
+            'age_limit': 12,
+            'timestamp': 1471209021,
+            'upload_date': '20160814',
         },
         'params': {
             'skip_download': True,
@@ -54,67 +58,36 @@ class MiTeleIE(InfoExtractor):
     }, {
         'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player',
         'only_matching': True,
+    }, {
+        'url': 'https://www.mitele.es/programas-tv/diario-de/la-redaccion/programa-144-40_1006364575251/player/',
+        'only_matching': True,
     }]
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        display_id = self._match_id(url)
-
+        webpage = self._download_webpage(url, display_id)
-        paths = self._download_json(
+        pre_player = self._parse_json(self._search_regex(
-            'https://www.mitele.es/amd/agp/web/metadata/general_configuration',
+            r'window\.\$REACTBASE_STATE\.prePlayer_mtweb\s*=\s*({.+})',
-            video_id, 'Downloading paths JSON')
+            webpage, 'Pre Player'), display_id)['prePlayer']
-
+        title = pre_player['title']
-        ooyala_s = paths['general_configuration']['api_configuration']['ooyala_search']
+        video = pre_player['video']
-        base_url = ooyala_s.get('base_url', 'cdn-search-mediaset.carbyne.ps.ooyala.com')
+        video_id = video['dataMediaId']
-        full_path = ooyala_s.get('full_path', '/search/v1/full/providers/')
+        content = pre_player.get('content') or {}
-        source = self._download_json(
+        info = content.get('info') or {}
-            '%s://%s%s%s/docs/%s' % (
-                ooyala_s.get('protocol', 'https'), base_url, full_path,
-                ooyala_s.get('provider_id', '104951'), video_id),
-            video_id, 'Downloading data JSON', query={
-                'include_titles': 'Series,Season',
-                'product_name': ooyala_s.get('product_name', 'test'),
-                'format': 'full',
-            })['hits']['hits'][0]['_source']
-
-        embedCode = source['offers'][0]['embed_codes'][0]
-        titles = source['localizable_titles'][0]
-
-        title = titles.get('title_medium') or titles['title_long']
-
-        description = titles.get('summary_long') or titles.get('summary_medium')
-
-        def get(key1, key2):
-            value1 = source.get(key1)
-            if not value1 or not isinstance(value1, list):
-                return
-            if not isinstance(value1[0], dict):
-                return
-            return value1[0].get(key2)
-
-        series = get('localizable_titles_series', 'title_medium')
-
-        season = get('localizable_titles_season', 'title_medium')
-        season_number = int_or_none(source.get('season_number'))
-        season_id = source.get('season_id')
-
-        episode = titles.get('title_sort_name')
-        episode_number = int_or_none(source.get('episode_number'))
-
-        duration = parse_duration(get('videos', 'duration'))
 
         return {
             '_type': 'url_transparent',
             # for some reason only HLS is supported
-            'url': smuggle_url('ooyala:' + embedCode, {'supportedformats': 'm3u8,dash'}),
+            'url': smuggle_url('ooyala:' + video_id, {'supportedformats': 'm3u8,dash'}),
             'id': video_id,
             'title': title,
-            'description': description,
+            'description': info.get('synopsis'),
-            'series': series,
+            'series': content.get('title'),
-            'season': season,
+            'season_number': int_or_none(info.get('season_number')),
-            'season_number': season_number,
+            'episode': content.get('subtitle'),
-            'season_id': season_id,
+            'episode_number': int_or_none(info.get('episode_number')),
-            'episode': episode,
+            'duration': int_or_none(info.get('duration')),
-            'episode_number': episode_number,
+            'thumbnail': video.get('dataPoster'),
-            'duration': duration,
+            'age_limit': int_or_none(info.get('rating')),
-            'thumbnail': get('images', 'url'),
+            'timestamp': parse_iso8601(pre_player.get('publishedTime')),
         }