Browse Source

[aljazeera] Add extractor (closes #4487)

Jaime Marquínez Ferrándiz 10 năm trước cách đây
mục cha
commit
bb18d787b5
2 tập tin đã thay đổi với 33 bổ sung0 xóa
  1. 1 0
      youtube_dl/extractor/__init__.py
  2. 32 0
      youtube_dl/extractor/aljazeera.py

+ 1 - 0
youtube_dl/extractor/__init__.py

@@ -5,6 +5,7 @@ from .academicearth import AcademicEarthCourseIE
 from .addanime import AddAnimeIE
 from .adultswim import AdultSwimIE
 from .aftonbladet import AftonbladetIE
+from .aljazeera import AlJazeeraIE
 from .anitube import AnitubeIE
 from .anysex import AnySexIE
 from .aol import AolIE

+ 32 - 0
youtube_dl/extractor/aljazeera.py

@@ -0,0 +1,32 @@
+from .common import InfoExtractor
+
+
+class AlJazeeraIE(InfoExtractor):
+    _VALID_URL = r'http://www\.aljazeera\.com/programmes/.*?/(?P<id>[^/]+)\.html'
+
+    _TEST = {
+        'url': 'http://www.aljazeera.com/programmes/the-slum/2014/08/deliverance-201482883754237240.html',
+        'info_dict': {
+            'id': '3792260579001',
+            'ext': 'mp4',
+            'title': 'The Slum - Episode 1: Deliverance',
+            'description': 'As a birth attendant advocating for family planning, Remy is on the frontline of Tondo\'s battle with overcrowding.',
+            'uploader': 'Al Jazeera English',
+        },
+        'add_ie': ['Brightcove'],
+    }
+
+    def _real_extract(self, url):
+        program_name = self._match_id(url)
+        webpage = self._download_webpage(url, program_name)
+        brightcove_id = self._search_regex(
+            r'RenderPagesVideo\(\'(.+?)\'', webpage, 'brightcove id')
+
+        return {
+            '_type': 'url',
+            'url':
+                'http://c.brightcove.com/services/viewer/federated_f9?'
+                '&playerKey=AQ~~%2CAAAAmtVJIFk~%2CTVGOQ5ZTwJbeMWnq5d_H4MOM57xfzApc'
+                '&%40videoPlayer={0}'.format(brightcove_id),
+            'ie_key': 'Brightcove',
+        }