Merge remote-tracking branch 'hassaanaliw/snotr'

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

diff --git a/youtube_dl/extractor/nrk.py b/youtube_dl/extractor/nrk.py
index a331a7702fa62dee10ff8ef3865cc2455a14927c..96f0ae1ebde53402a3e651dcd86c1397d141c542 100644 (file)
--- a/youtube_dl/extractor/nrk.py
+++ b/youtube_dl/extractor/nrk.py
@@ -4,7 +4,11 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..utils import ExtractorError
+from ..utils import (
+    ExtractorError,
+    float_or_none,
+    unified_strdate,
+)
 
 
 class NRKIE(InfoExtractor):
@@ -13,7 +17,7 @@ class NRKIE(InfoExtractor):
     _TESTS = [
         {
             'url': 'http://www.nrk.no/video/dompap_og_andre_fugler_i_piip_show/D0FA54B5C8B6CE59/emne/piipshow/',
-            'md5': '12618eef328c9a35c1b47d5583d9c30d',
+            'md5': 'a6eac35052f3b242bb6bb7f43aed5886',
             'info_dict': {
                 'id': '150533',
                 'ext': 'flv',
@@ -23,7 +27,7 @@ class NRKIE(InfoExtractor):
         },
         {
             'url': 'http://www.nrk.no/lyd/lyd_av_oppleser_for_blinde/AEFDDD5473BA0198/',
-            'md5': '390b2ce15c0d6aa376ef5059ac9f865e',
+            'md5': '3471f2a51718195164e88f46bf427668',
             'info_dict': {
                 'id': '154915',
                 'ext': 'flv',
@@ -64,4 +68,77 @@ class NRKIE(InfoExtractor):
             'title': data['title'],
             'description': data['description'],
             'thumbnail': thumbnail,
-        }
\ No newline at end of file
+        }
+
+
+class NRKTVIE(InfoExtractor):
+    _VALID_URL = r'http://tv\.nrk(?:super)?\.no/(?:serie/[^/]+|program)/(?P<id>[a-zA-Z]{4}\d{8})'
+
+    _TESTS = [
+        {
+            'url': 'http://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
+            'md5': '7b96112fbae1faf09a6f9ae1aff6cb84',
+            'info_dict': {
+                'id': 'MUHH48000314',
+                'ext': 'flv',
+                'title': '20 spørsmål',
+                'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
+                'upload_date': '20140523',
+                'duration': 1741.52,
+            }
+        },
+        {
+            'url': 'http://tv.nrk.no/program/mdfp15000514',
+            'md5': 'af01795a31f1cf7265c8657534d8077b',
+            'info_dict': {
+                'id': 'mdfp15000514',
+                'ext': 'flv',
+                'title': 'Kunnskapskanalen: Grunnlovsjubiléet - Stor ståhei for ingenting',
+                'description': 'md5:654c12511f035aed1e42bdf5db3b206a',
+                'upload_date': '20140524',
+                'duration': 4605.0,
+            }
+        },
+    ]
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id')
+
+        page = self._download_webpage(url, video_id)
+
+        title = self._html_search_meta('title', page, 'title')
+        description = self._html_search_meta('description', page, 'description')
+        thumbnail = self._html_search_regex(r'data-posterimage="([^"]+)"', page, 'thumbnail', fatal=False)
+        upload_date = unified_strdate(self._html_search_meta('rightsfrom', page, 'upload date', fatal=False))
+        duration = float_or_none(
+            self._html_search_regex(r'data-duration="([^"]+)"', page, 'duration', fatal=False))
+
+        formats = []
+
+        f4m_url = re.search(r'data-media="([^"]+)"', page)
+        if f4m_url:
+            formats.append({
+                'url': f4m_url.group(1) + '?hdcore=3.1.1&plugin=aasp-3.1.1.69.124',
+                'format_id': 'f4m',
+                'ext': 'flv',
+            })
+
+        m3u8_url = re.search(r'data-hls-media="([^"]+)"', page)
+        if m3u8_url:
+            formats.append({
+                'url': m3u8_url.group(1),
+                'format_id': 'm3u8',
+            })
+
+        self._sort_formats(formats)
+
+        return {
+            'id': video_id,
+            'title': title,
+            'description': description,
+            'thumbnail': thumbnail,
+            'upload_date': upload_date,
+            'duration': duration,
+            'formats': formats,
+        }