[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / vgtv.py
index 811ee197d415bec57c355a986a07d65a8e0d8b7c..fe7a26b620deeff53a5443beea96e109f7685dc8 100644 (file)
@@ -8,11 +8,13 @@ from .xstream import XstreamIE
 from ..utils import (
     ExtractorError,
     float_or_none,
+    try_get,
 )
 
 
 class VGTVIE(XstreamIE):
     IE_DESC = 'VGTV, BTTV, FTV, Aftenposten and Aftonbladet'
+    _GEO_BYPASS = False
 
     _HOST_TO_APPNAME = {
         'vgtv.no': 'vgtv',
@@ -20,6 +22,9 @@ class VGTVIE(XstreamIE):
         'aftenbladet.no/tv': 'satv',
         'fvn.no/fvntv': 'fvntv',
         'aftenposten.no/webtv': 'aptv',
+        'ap.vgtv.no/webtv': 'aptv',
+        'tv.aftonbladet.se/abtv': 'abtv',
+        'www.aftonbladet.se/tv': 'abtv',
     }
 
     _APP_NAME_TO_VENDOR = {
@@ -28,6 +33,7 @@ class VGTVIE(XstreamIE):
         'satv': 'sa',
         'fvntv': 'fvn',
         'aptv': 'ap',
+        'abtv': 'ab',
     }
 
     _VALID_URL = r'''(?x)
@@ -35,10 +41,11 @@ class VGTVIE(XstreamIE):
                     (?P<host>
                         %s
                     )
-                    /
+                    /?
                     (?:
-                        \#!/(?:video|live)/|
-                        embed?.*id=
+                        (?:\#!/)?(?:video|live)/|
+                        embed?.*id=|
+                        a(?:rticles)?/
                     )|
                     (?P<appname>
                         %s
@@ -56,7 +63,7 @@ class VGTVIE(XstreamIE):
                 'ext': 'mp4',
                 'title': 'Hevnen er søt: Episode 10 - Abu',
                 'description': 'md5:e25e4badb5f544b04341e14abdc72234',
-                'thumbnail': 're:^https?://.*\.jpg',
+                'thumbnail': r're:^https?://.*\.jpg',
                 'duration': 648.000,
                 'timestamp': 1404626400,
                 'upload_date': '20140706',
@@ -71,7 +78,7 @@ class VGTVIE(XstreamIE):
                 'ext': 'flv',
                 'title': 'OPPTAK: VGTV følger EM-kvalifiseringen',
                 'description': 'md5:3772d9c0dc2dff92a886b60039a7d4d3',
-                'thumbnail': 're:^https?://.*\.jpg',
+                'thumbnail': r're:^https?://.*\.jpg',
                 'duration': 9103.0,
                 'timestamp': 1410113864,
                 'upload_date': '20140907',
@@ -91,7 +98,7 @@ class VGTVIE(XstreamIE):
                 'ext': 'mp4',
                 'title': 'V75 fra Solvalla 30.05.15',
                 'description': 'md5:b3743425765355855f88e096acc93231',
-                'thumbnail': 're:^https?://.*\.jpg',
+                'thumbnail': r're:^https?://.*\.jpg',
                 'duration': 25966,
                 'timestamp': 1432975582,
                 'upload_date': '20150530',
@@ -107,19 +114,48 @@ class VGTVIE(XstreamIE):
             'md5': 'fd828cd29774a729bf4d4425fe192972',
             'info_dict': {
                 'id': '21039',
-                'ext': 'mov',
+                'ext': 'mp4',
                 'title': 'TRAILER: «SWEATSHOP» - I can´t take any more',
                 'description': 'md5:21891f2b0dd7ec2f78d84a50e54f8238',
                 'duration': 66,
                 'timestamp': 1417002452,
                 'upload_date': '20141126',
                 'view_count': int,
-            }
+            },
+            'params': {
+                # m3u8 download
+                'skip_download': True,
+            },
         },
         {
             'url': 'http://www.bt.no/tv/#!/video/100250/norling-dette-er-forskjellen-paa-1-divisjon-og-eliteserien',
             'only_matching': True,
         },
+        {
+            'url': 'http://ap.vgtv.no/webtv#!/video/111084/de-nye-bysyklene-lettere-bedre-gir-stoerre-hjul-og-feste-til-mobil',
+            'only_matching': True,
+        },
+        {
+            # geoblocked
+            'url': 'http://www.vgtv.no/#!/video/127205/inside-the-mind-of-favela-funk',
+            'only_matching': True,
+        },
+        {
+            'url': 'http://tv.aftonbladet.se/abtv/articles/36015',
+            'only_matching': True,
+        },
+        {
+            'url': 'https://www.aftonbladet.se/tv/a/36015',
+            'only_matching': True,
+        },
+        {
+            'url': 'abtv:140026',
+            'only_matching': True,
+        },
+        {
+            'url': 'http://www.vgtv.no/video/84196/hevnen-er-soet-episode-10-abu',
+            'only_matching': True,
+        },
     ]
 
     def _real_extract(self, url):
@@ -144,28 +180,29 @@ class VGTVIE(XstreamIE):
         if len(video_id) == 5:
             if appname == 'bttv':
                 info = self._extract_video_info('btno', video_id)
-            elif appname == 'aptv':
-                info = self._extract_video_info('ap', video_id)
 
         streams = data['streamUrls']
         stream_type = data.get('streamType')
-
+        is_live = stream_type == 'live'
         formats = []
 
         hls_url = streams.get('hls')
         if hls_url:
-            m3u8_formats = self._extract_m3u8_formats(
-                hls_url, video_id, 'mp4', m3u8_id='hls', fatal=False)
-            if m3u8_formats:
-                formats.extend(m3u8_formats)
+            formats.extend(self._extract_m3u8_formats(
+                hls_url, video_id, 'mp4',
+                entry_protocol='m3u8' if is_live else 'm3u8_native',
+                m3u8_id='hls', fatal=False))
 
         hds_url = streams.get('hds')
-        # wasLive hds are always 404
-        if hds_url and stream_type != 'wasLive':
+        if hds_url:
+            hdcore_sign = 'hdcore=3.7.0'
             f4m_formats = self._extract_f4m_formats(
-                hds_url + '?hdcore=3.2.0&plugin=aasp-3.2.0.77.18', video_id, f4m_id='hds', fatal=False)
+                hds_url + '?%s' % hdcore_sign, video_id, f4m_id='hds', fatal=False)
             if f4m_formats:
-                formats.extend(f4m_formats)
+                for entry in f4m_formats:
+                    # URLs without the extra param induce an 404 error
+                    entry.update({'extra_param_to_segment_url': hdcore_sign})
+                    formats.append(entry)
 
         mp4_urls = streams.get('pseudostreaming') or []
         mp4_url = streams.get('mp4')
@@ -175,7 +212,7 @@ class VGTVIE(XstreamIE):
             format_info = {
                 'url': mp4_url,
             }
-            mobj = re.search('(\d+)_(\d+)_(\d+)', mp4_url)
+            mobj = re.search(r'(\d+)_(\d+)_(\d+)', mp4_url)
             if mobj:
                 tbr = int(mobj.group(3))
                 format_info.update({
@@ -188,17 +225,24 @@ class VGTVIE(XstreamIE):
 
         info['formats'].extend(formats)
 
+        if not info['formats']:
+            properties = try_get(
+                data, lambda x: x['streamConfiguration']['properties'], list)
+            if properties and 'geoblocked' in properties:
+                raise self.raise_geo_restricted(
+                    countries=[host.rpartition('.')[-1].partition('/')[0].upper()])
+
         self._sort_formats(info['formats'])
 
         info.update({
             'id': video_id,
-            'title': self._live_title(data['title']) if stream_type == 'live' else data['title'],
+            'title': self._live_title(data['title']) if is_live else data['title'],
             'description': data['description'],
             'thumbnail': data['images']['main'] + '?t[]=900x506q80',
             'timestamp': data['published'],
             'duration': float_or_none(data['duration'], 1000),
             'view_count': data['displays'],
-            'is_live': True if stream_type == 'live' else False,
+            'is_live': is_live,
         })
         return info
 
@@ -206,7 +250,7 @@ class VGTVIE(XstreamIE):
 class BTArticleIE(InfoExtractor):
     IE_NAME = 'bt:article'
     IE_DESC = 'Bergens Tidende Articles'
-    _VALID_URL = 'http://(?:www\.)?bt\.no/(?:[^/]+/)+(?P<id>[^/]+)-\d+\.html'
+    _VALID_URL = r'https?://(?:www\.)?bt\.no/(?:[^/]+/)+(?P<id>[^/]+)-\d+\.html'
     _TEST = {
         'url': 'http://www.bt.no/nyheter/lokalt/Kjemper-for-internatet-1788214.html',
         'md5': '2acbe8ad129b3469d5ae51b1158878df',
@@ -215,7 +259,7 @@ class BTArticleIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Alrekstad internat',
             'description': 'md5:dc81a9056c874fedb62fc48a300dac58',
-            'thumbnail': 're:^https?://.*\.jpg',
+            'thumbnail': r're:^https?://.*\.jpg',
             'duration': 191,
             'timestamp': 1289991323,
             'upload_date': '20101117',
@@ -233,7 +277,7 @@ class BTArticleIE(InfoExtractor):
 class BTVestlendingenIE(InfoExtractor):
     IE_NAME = 'bt:vestlendingen'
     IE_DESC = 'Bergens Tidende - Vestlendingen'
-    _VALID_URL = 'http://(?:www\.)?bt\.no/spesial/vestlendingen/#!/(?P<id>\d+)'
+    _VALID_URL = r'https?://(?:www\.)?bt\.no/spesial/vestlendingen/#!/(?P<id>\d+)'
     _TESTS = [{
         'url': 'http://www.bt.no/spesial/vestlendingen/#!/86588',
         'md5': 'd7d17e3337dc80de6d3a540aefbe441b',