[youtube] Fix extraction.

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

diff --git a/youtube_dl/extractor/mitele.py b/youtube_dl/extractor/mitele.py
index 7091f3335e8223ea0a089ba7cff127a983b12d7d..ad9da96125b92d66d4e6367e23b7ed9a1a633aaf 100644 (file)
--- a/youtube_dl/extractor/mitele.py
+++ b/youtube_dl/extractor/mitele.py
@@ -1,73 +1,93 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
-import json
-
 from .common import InfoExtractor
-from ..compat import (
-    compat_urllib_parse,
-    compat_urlparse,
-)
 from ..utils import (
-    get_element_by_attribute,
-    parse_duration,
-    strip_jsonp,
+    int_or_none,
+    parse_iso8601,
+    smuggle_url,
 )
 
 
 class MiTeleIE(InfoExtractor):
-    IE_NAME = 'mitele.es'
-    _VALID_URL = r'http://www\.mitele\.es/[^/]+/[^/]+/[^/]+/(?P<id>[^/]+)/'
+    IE_DESC = 'mitele.es'
+    _VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P<id>[^/]+)/player'
 
     _TESTS = [{
-        'url': 'http://www.mitele.es/programas-tv/diario-de/la-redaccion/programa-144/',
+        'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player',
         'info_dict': {
-            'id': '0fce117d',
+            'id': 'FhYW1iNTE6J6H7NkQRIEzfne6t2quqPg',
             'ext': 'mp4',
-            'title': 'Programa 144 - Tor, la web invisible',
-            'description': 'md5:3b6fce7eaa41b2d97358726378d9369f',
-            'display_id': 'programa-144',
+            'title': 'Diario de La redacción Programa 144',
+            'description': 'md5:07c35a7b11abb05876a6a79185b58d27',
+            'series': 'Diario de',
+            'season': 'Season 14',
+            'season_number': 14,
+            'episode': 'Tor, la web invisible',
+            'episode_number': 3,
+            'thumbnail': r're:(?i)^https?://.*\.jpg$',
             'duration': 2913,
+            'age_limit': 16,
+            'timestamp': 1471209401,
+            'upload_date': '20160814',
+        },
+        'add_ie': ['Ooyala'],
+    }, {
+        # no explicit title
+        'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player',
+        'info_dict': {
+            'id': 'oyNG1iNTE6TAPP-JmCjbwfwJqqMMX3Vq',
+            'ext': 'mp4',
+            'title': 'Cuarto Milenio Temporada 6 Programa 226',
+            'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f',
+            'series': 'Cuarto Milenio',
+            'season': 'Season 6',
+            'season_number': 6,
+            'episode': 'Episode 24',
+            'episode_number': 24,
+            'thumbnail': r're:(?i)^https?://.*\.jpg$',
+            'duration': 7313,
+            'age_limit': 12,
+            'timestamp': 1471209021,
+            'upload_date': '20160814',
         },
         'params': {
-            # m3u8 download
             'skip_download': True,
         },
+        'add_ie': ['Ooyala'],
+    }, {
+        '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):
-        episode = self._match_id(url)
-        webpage = self._download_webpage(url, episode)
-        embed_data_json = self._search_regex(
-            r'(?s)MSV\.embedData\[.*?\]\s*=\s*({.*?});', webpage, 'embed data',
-        ).replace('\'', '"')
-        embed_data = json.loads(embed_data_json)
-
-        domain = embed_data['mediaUrl']
-        if not domain.startswith('http'):
-            # only happens in telecinco.es videos
-            domain = 'http://' + domain
-        info_url = compat_urlparse.urljoin(
-            domain,
-            compat_urllib_parse.unquote(embed_data['flashvars']['host'])
-        )
-        info_el = self._download_xml(info_url, episode).find('./video/info')
-
-        video_link = info_el.find('videoUrl/link').text
-        token_query = compat_urllib_parse.urlencode({'id': video_link})
-        token_info = self._download_json(
-            embed_data['flashvars']['ov_tk'] + '?' + token_query,
-            episode,
-            transform_source=strip_jsonp
-        )
-        formats = self._extract_m3u8_formats(
-            token_info['tokenizedUrl'], episode, ext='mp4')
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+        pre_player = self._parse_json(self._search_regex(
+            r'window\.\$REACTBASE_STATE\.prePlayer_mtweb\s*=\s*({.+})',
+            webpage, 'Pre Player'), display_id)['prePlayer']
+        title = pre_player['title']
+        video = pre_player['video']
+        video_id = video['dataMediaId']
+        content = pre_player.get('content') or {}
+        info = content.get('info') or {}
 
         return {
-            'id': embed_data['videoId'],
-            'display_id': episode,
-            'title': info_el.find('title').text,
-            'formats': formats,
-            'description': get_element_by_attribute('class', 'text', webpage),
-            'thumbnail': info_el.find('thumb').text,
-            'duration': parse_duration(info_el.find('duration').text),
+            '_type': 'url_transparent',
+            # for some reason only HLS is supported
+            'url': smuggle_url('ooyala:' + video_id, {'supportedformats': 'm3u8,dash'}),
+            'id': video_id,
+            'title': title,
+            'description': info.get('synopsis'),
+            'series': content.get('title'),
+            'season_number': int_or_none(info.get('season_number')),
+            'episode': content.get('subtitle'),
+            'episode_number': int_or_none(info.get('episode_number')),
+            'duration': int_or_none(info.get('duration')),
+            'thumbnail': video.get('dataPoster'),
+            'age_limit': int_or_none(info.get('rating')),
+            'timestamp': parse_iso8601(pre_player.get('publishedTime')),
         }