[ie/netzkino] Rework extractor (#15351)

Authored by: doe1080

yt_dlp/extractor/netzkino.py

@@ -2,84 +2,59 @@ from .common import InfoExtractor
 from ..utils import (
     clean_html,
     int_or_none,
-    js_to_json,
+    url_or_none,
-    parse_iso8601,
+    urljoin,
 )
+from ..utils.traversal import traverse_obj
 
 
 class NetzkinoIE(InfoExtractor):
-    _WORKING = False
+    _GEO_COUNTRIES = ['DE']
-    _VALID_URL = r'https?://(?:www\.)?netzkino\.de/\#!/[^/]+/(?P<id>[^/]+)'
+    _VALID_URL = r'https?://(?:www\.)?netzkino\.de/details/(?P<id>[^/?#]+)'
-
     _TESTS = [{
-        'url': 'https://www.netzkino.de/#!/scifikino/rakete-zum-mond',
+        'url': 'https://www.netzkino.de/details/snow-beast',
-        'md5': '92a3f8b76f8d7220acce5377ea5d4873',
+        'md5': '1a4c90fe40d3ccabce163287e45e56dd',
         'info_dict': {
-            'id': 'rakete-zum-mond',
+            'id': 'snow-beast',
             'ext': 'mp4',
-            'title': 'Rakete zum Mond \u2013 Jules Verne',
+            'title': 'Snow Beast',
-            'description': 'md5:f0a8024479618ddbfa450ff48ffa6c60',
-            'upload_date': '20120813',
-            'thumbnail': r're:https?://.*\.jpg$',
-            'timestamp': 1344858571,
             'age_limit': 12,
-        },
+            'alt_title': 'Snow Beast',
-        'params': {
+            'cast': 'count:3',
-            'skip_download': 'Download only works from Germany',
+            'categories': 'count:7',
-        },
+            'creators': 'count:2',
-    }, {
+            'description': 'md5:e604a954a7f827a80e96a3a97d48b269',
-        'url': 'https://www.netzkino.de/#!/filme/dr-jekyll-mrs-hyde-2',
+            'location': 'US',
-        'md5': 'c7728b2dadd04ff6727814847a51ef03',
+            'release_year': 2011,
-        'info_dict': {
+            'thumbnail': r're:https?://.+\.jpg',
-            'id': 'dr-jekyll-mrs-hyde-2',
-            'ext': 'mp4',
-            'title': 'Dr. Jekyll & Mrs. Hyde 2',
-            'description': 'md5:c2e9626ebd02de0a794b95407045d186',
-            'upload_date': '20190130',
-            'thumbnail': r're:https?://.*\.jpg$',
-            'timestamp': 1548849437,
-            'age_limit': 18,
-        },
-        'params': {
-            'skip_download': 'Download only works from Germany',
         },
     }]
 
     def _real_extract(self, url):
-        mobj = self._match_valid_url(url)
+        video_id = self._match_id(url)
-        video_id = mobj.group('id')
+        webpage = self._download_webpage(url, video_id)
+        next_js_data = self._search_nextjs_data(webpage, video_id)
 
-        api_url = f'https://api.netzkino.de.simplecache.net/capi-2.0a/movies/{video_id}.json?d=www'
+        query = traverse_obj(next_js_data, (
-        info = self._download_json(api_url, video_id)
+            'props', '__dehydratedState', 'queries', ..., 'state',
-        custom_fields = info['custom_fields']
+            'data', 'data', lambda _, v: v['__typename'] == 'CmsMovie', any))
-
+        if 'DRM' in traverse_obj(query, ('licenses', 'nodes', ..., 'properties', {str})):
-        production_js = self._download_webpage(
+            self.report_drm(video_id)
-            'http://www.netzkino.de/beta/dist/production.min.js', video_id,
-            note='Downloading player code')
-        avo_js = self._search_regex(
-            r'var urlTemplate=(\{.*?"\})',
-            production_js, 'URL templates')
-        templates = self._parse_json(
-            avo_js, video_id, transform_source=js_to_json)
-
-        suffix = {
-            'hds': '.mp4/manifest.f4m',
-            'hls': '.mp4/master.m3u8',
-            'pmd': '.mp4',
-        }
-        film_fn = custom_fields['Streaming'][0]
-        formats = [{
-            'format_id': key,
-            'ext': 'mp4',
-            'url': tpl.replace('{}', film_fn) + suffix[key],
-        } for key, tpl in templates.items()]
 
         return {
             'id': video_id,
-            'formats': formats,
+            **traverse_obj(query, {
-            'title': info['title'],
+                'title': ('originalTitle', {clean_html}),
-            'age_limit': int_or_none(custom_fields.get('FSK')[0]),
+                'age_limit': ('fskRating', {int_or_none}),
-            'timestamp': parse_iso8601(info.get('date'), delimiter=' '),
+                'alt_title': ('originalTitle', {clean_html}, filter),
-            'description': clean_html(info.get('content')),
+                'cast': ('cast', 'nodes', ..., 'person', 'name', {clean_html}, filter),
-            'thumbnail': info.get('thumbnail'),
+                'creators': (('directors', 'writers'), 'nodes', ..., 'person', 'name', {clean_html}, filter),
+                'categories': ('categories', 'nodes', ..., 'category', 'title', {clean_html}, filter),
+                'description': ('longSynopsis', {clean_html}, filter),
+                'duration': ('runtimeInSeconds', {int_or_none}),
+                'location': ('productionCountry', {clean_html}, filter),
+                'release_year': ('productionYear', {int_or_none}),
+                'thumbnail': ('coverImage', 'masterUrl', {url_or_none}),
+                'url': ('videoSource', 'pmdUrl', {urljoin('https://pmd.netzkino-seite.netzkino.de/')}),
+            }),
         }