Merge pull request #7045 from remitamine/ign

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

diff --git a/youtube_dl/extractor/ard.py b/youtube_dl/extractor/ard.py
index 2c368d833c71809106f549c900e00db544155f87..687eb9f823d0a7ba34d3f3139141af96a27b8222 100644 (file)
--- a/youtube_dl/extractor/ard.py
+++ b/youtube_dl/extractor/ard.py
@@ -14,8 +14,8 @@ from ..utils import (
     parse_duration,
     unified_strdate,
     xpath_text,
-    parse_xml,
 )
+from ..compat import compat_etree_fromstring
 
 
 class ARDMediathekIE(InfoExtractor):
@@ -23,17 +23,42 @@ class ARDMediathekIE(InfoExtractor):
     _VALID_URL = r'^https?://(?:(?:www\.)?ardmediathek\.de|mediathek\.daserste\.de)/(?:.*/)(?P<video_id>[0-9]+|[^0-9][^/\?]+)[^/\?]*(?:\?.*)?'
 
     _TESTS = [{
-        'url': 'http://mediathek.daserste.de/sendungen_a-z/328454_anne-will/22429276_vertrauen-ist-gut-spionieren-ist-besser-geht',
-        'only_matching': True,
+        'url': 'http://www.ardmediathek.de/tv/Dokumentation-und-Reportage/Ich-liebe-das-Leben-trotzdem/rbb-Fernsehen/Video?documentId=29582122&bcastId=3822114',
+        'info_dict': {
+            'id': '29582122',
+            'ext': 'mp4',
+            'title': 'Ich liebe das Leben trotzdem',
+            'description': 'md5:45e4c225c72b27993314b31a84a5261c',
+            'duration': 4557,
+        },
+        'params': {
+            # m3u8 download
+            'skip_download': True,
+        },
     }, {
-        'url': 'http://www.ardmediathek.de/tv/Tatort/Das-Wunder-von-Wolbeck-Video-tgl-ab-20/Das-Erste/Video?documentId=22490580&bcastId=602916',
+        'url': 'http://www.ardmediathek.de/tv/Tatort/Tatort-Scheinwelten-H%C3%B6rfassung-Video/Das-Erste/Video?documentId=29522730&bcastId=602916',
+        'md5': 'f4d98b10759ac06c0072bbcd1f0b9e3e',
         'info_dict': {
-            'id': '22490580',
+            'id': '29522730',
             'ext': 'mp4',
-            'title': 'Das Wunder von Wolbeck (Video tgl. ab 20 Uhr)',
-            'description': 'Auf einem restaurierten Hof bei Wolbeck wird der Heilpraktiker Raffael Lembeck eines morgens von seiner Frau Stella tot aufgefunden. Das Opfer war offensichtlich in seiner Praxis zu Fall gekommen und ist dann verblutet, erklärt Prof. Boerne am Tatort.',
+            'title': 'Tatort: Scheinwelten - Hörfassung (Video tgl. ab 20 Uhr)',
+            'description': 'md5:196392e79876d0ac94c94e8cdb2875f1',
+            'duration': 5252,
         },
-        'skip': 'Blocked outside of Germany',
+    }, {
+        # audio
+        'url': 'http://www.ardmediathek.de/tv/WDR-H%C3%B6rspiel-Speicher/Tod-eines-Fu%C3%9Fballers/WDR-3/Audio-Podcast?documentId=28488308&bcastId=23074086',
+        'md5': '219d94d8980b4f538c7fcb0865eb7f2c',
+        'info_dict': {
+            'id': '28488308',
+            'ext': 'mp3',
+            'title': 'Tod eines Fußballers',
+            'description': 'md5:f6e39f3461f0e1f54bfa48c8875c86ef',
+            'duration': 3240,
+        },
+    }, {
+        'url': 'http://mediathek.daserste.de/sendungen_a-z/328454_anne-will/22429276_vertrauen-ist-gut-spionieren-ist-besser-geht',
+        'only_matching': True,
     }]
 
     def _extract_media_info(self, media_info_url, webpage, video_id):
@@ -85,13 +110,19 @@ class ARDMediathekIE(InfoExtractor):
                 server = stream.get('_server')
                 for stream_url in stream_urls:
                     ext = determine_ext(stream_url)
+                    if quality != 'auto' and ext in ('f4m', 'm3u8'):
+                        continue
                     if ext == 'f4m':
-                        formats.extend(self._extract_f4m_formats(
+                        f4m_formats = self._extract_f4m_formats(
                             stream_url + '?hdcore=3.1.1&plugin=aasp-3.1.1.69.124',
-                            video_id, preference=-1, f4m_id='hds'))
+                            video_id, preference=-1, f4m_id='hds', fatal=False)
+                        if f4m_formats:
+                            formats.extend(f4m_formats)
                     elif ext == 'm3u8':
-                        formats.extend(self._extract_m3u8_formats(
-                            stream_url, video_id, 'mp4', preference=1, m3u8_id='hls'))
+                        m3u8_formats = self._extract_m3u8_formats(
+                            stream_url, video_id, 'mp4', preference=1, m3u8_id='hls', fatal=False)
+                        if m3u8_formats:
+                            formats.extend(m3u8_formats)
                     else:
                         if server and server.startswith('rtmp'):
                             f = {
@@ -136,7 +167,7 @@ class ARDMediathekIE(InfoExtractor):
             raise ExtractorError('This program is only suitable for those aged 12 and older. Video %s is therefore only available between 20 pm and 6 am.' % video_id, expected=True)
 
         if re.search(r'[\?&]rss($|[=&])', url):
-            doc = parse_xml(webpage)
+            doc = compat_etree_fromstring(webpage.encode('utf-8'))
             if doc.tag == 'rss':
                 return GenericIE()._extract_rss(url, video_id, doc)
 
@@ -252,7 +283,7 @@ class ARDIE(InfoExtractor):
 class SportschauIE(ARDMediathekIE):
     IE_NAME = 'Sportschau'
     _VALID_URL = r'(?P<baseurl>https?://(?:www\.)?sportschau\.de/(?:[^/]+/)+video(?P<id>[^/#?]+))\.html'
-    _TEST = {
+    _TESTS = [{
         'url': 'http://www.sportschau.de/tourdefrance/videoseppeltkokainhatnichtsmitklassischemdopingzutun100.html',
         'info_dict': {
             'id': 'seppeltkokainhatnichtsmitklassischemdopingzutun100',
@@ -265,7 +296,7 @@ class SportschauIE(ARDMediathekIE):
             # m3u8 download
             'skip_download': True,
         },
-    }
+    }]
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)