From cba1a5e1ae14134ae47e96b28b786c60c35ab461 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Stefan=20P=C3=B6schel?= <github@basicmaster.de>
Date: Mon, 27 Apr 2020 17:08:16 +0200
Subject: [PATCH] [ntvde] Fix extraction (closes #21204)

---
 youtube_dl/extractor/ntvde.py | 37 ++++++++++++++---------------------
 1 file changed, 15 insertions(+), 22 deletions(-)

diff --git a/youtube_dl/extractor/ntvde.py b/youtube_dl/extractor/ntvde.py
index 101a5374c..40cd1cf45 100644
--- a/youtube_dl/extractor/ntvde.py
+++ b/youtube_dl/extractor/ntvde.py
@@ -4,11 +4,9 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..compat import compat_urlparse
 from ..utils import (
     int_or_none,
     js_to_json,
-    parse_duration,
 )
 
 
@@ -18,7 +16,7 @@ class NTVDeIE(InfoExtractor):
 
     _TESTS = [{
         'url': 'http://www.n-tv.de/mediathek/videos/panorama/Schnee-und-Glaette-fuehren-zu-zahlreichen-Unfaellen-und-Staus-article14438086.html',
-        'md5': '6ef2514d4b1e8e03ca24b49e2f167153',
+        'md5': 'ec439b7980c2a600c357dbe431555c5d',
         'info_dict': {
             'id': '14438086',
             'ext': 'mp4',
@@ -26,7 +24,7 @@ class NTVDeIE(InfoExtractor):
             'title': 'Schnee und Glätte führen zu zahlreichen Unfällen und Staus',
             'alt_title': 'Winterchaos auf deutschen Straßen',
             'description': 'Schnee und Glätte sorgen deutschlandweit für einen chaotischen Start in die Woche: Auf den Straßen kommt es zu kilometerlangen Staus und Dutzenden Glätteunfällen. In Düsseldorf und München wirbelt der Schnee zudem den Flugplan durcheinander. Dutzende Flüge landen zu spät, einige fallen ganz aus.',
-            'duration': 4020,
+            'duration': 67,
             'timestamp': 1422892797,
             'upload_date': '20150202',
         },
@@ -37,31 +35,26 @@ class NTVDeIE(InfoExtractor):
         webpage = self._download_webpage(url, video_id)
 
         info = self._parse_json(self._search_regex(
-            r'(?s)ntv\.pageInfo\.article\s*=\s*(\{.*?\});', webpage, 'info'),
+            r'(?s)article:\s*(\{.*?\})', webpage, 'info'),
             video_id, transform_source=js_to_json)
         timestamp = int_or_none(info.get('publishedDateAsUnixTimeStamp'))
-        vdata = self._parse_json(self._search_regex(
-            r'(?s)\$\(\s*"\#player"\s*\)\s*\.data\(\s*"player",\s*(\{.*?\})\);',
+        pdata = self._parse_json(self._search_regex(
+            r'(?s)\$\(\s*"\#playerwrapper"\s*\)\s*\.data\(\s*"player",\s*(\{.*?\})\);',
             webpage, 'player data'), video_id,
-            transform_source=lambda s: js_to_json(re.sub(r'advertising:\s*{[^}]+},', '', s)))
-        duration = parse_duration(vdata.get('duration'))
+            transform_source=lambda s: js_to_json(re.sub(r',\s*ivw:\s*[^}]*}', '}', s)))  # ignore from composed IVW key/value on
+        vdata = pdata['setup']['source']
+        duration = vdata.get('length')
 
         formats = []
-        if vdata.get('video'):
+        if vdata.get('progressive'):
             formats.append({
-                'format_id': 'flash',
-                'url': 'rtmp://fms.n-tv.de/%s' % vdata['video'],
+                'format_id': 'progressive',
+                'url': vdata['progressive'],
+                'preference': 1,  # prefer, as seems to have the same properties as the highest HLS quality, plus proper fps
             })
-        if vdata.get('videoMp4'):
-            formats.append({
-                'format_id': 'mobile',
-                'url': compat_urlparse.urljoin('http://video.n-tv.de', vdata['videoMp4']),
-                'tbr': 400,  # estimation
-            })
-        if vdata.get('videoM3u8'):
-            m3u8_url = compat_urlparse.urljoin('http://video.n-tv.de', vdata['videoM3u8'])
+        if vdata.get('hls'):
             formats.extend(self._extract_m3u8_formats(
-                m3u8_url, video_id, ext='mp4', entry_protocol='m3u8_native',
+                vdata['hls'], video_id, ext='mp4', entry_protocol='m3u8_native',
                 preference=0, m3u8_id='hls', fatal=False))
         self._sort_formats(formats)
 
@@ -71,7 +64,7 @@ class NTVDeIE(InfoExtractor):
             'description': info.get('intro'),
             'alt_title': info.get('kicker'),
             'timestamp': timestamp,
-            'thumbnail': vdata.get('html5VideoPoster'),
+            'thumbnail': vdata.get('poster'),
             'duration': duration,
             'formats': formats,
         }