[udemy:course] Skip non-video lectures

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

diff --git a/youtube_dl/extractor/theplatform.py b/youtube_dl/extractor/theplatform.py
index a148f78ce3e60abb429d3cddfa76bc2ad8cff37f..bf6f82f5ae52af83706807f7112c9bdd74138461 100644 (file)
--- a/youtube_dl/extractor/theplatform.py
+++ b/youtube_dl/extractor/theplatform.py
@@ -22,7 +22,6 @@ from ..utils import (
     xpath_with_ns,
     mimetype2ext,
     find_xpath_attr,
-    update_url_query,
 )
 
 default_ns = 'http://www.w3.org/2005/SMIL21/Language'
@@ -31,12 +30,10 @@ _x = lambda p: xpath_with_ns(p, {'smil': default_ns})
 
 class ThePlatformBaseIE(OnceIE):
     def _extract_theplatform_smil(self, smil_url, video_id, note='Downloading SMIL data'):
-        smil_url = update_url_query(smil_url, {'format': 'SMIL'})
-        meta = self._download_xml(smil_url, video_id, note=note)
-        error_element = find_xpath_attr(
-            meta, _x('.//smil:ref'), 'src',
-            'http://link.theplatform.com/s/errorFiles/Unavailable.mp4')
-        if error_element is not None:
+        meta = self._download_xml(smil_url, video_id, note=note, query={'format': 'SMIL'})
+        error_element = find_xpath_attr(meta, _x('.//smil:ref'), 'src')
+        if error_element is not None and error_element.attrib['src'].startswith(
+                'http://link.theplatform.com/s/errorFiles/Unavailable.'):
             raise ExtractorError(error_element.attrib['abstract'], expected=True)
 
         smil_formats = self._parse_smil_formats(
@@ -85,7 +82,7 @@ class ThePlatformBaseIE(OnceIE):
 class ThePlatformIE(ThePlatformBaseIE):
     _VALID_URL = r'''(?x)
         (?:https?://(?:link|player)\.theplatform\.com/[sp]/(?P<provider_id>[^/]+)/
-           (?:(?P<media>(?:(?:[^/]+/)+select/)?media/)|(?P<config>(?:[^/\?]+/(?:swf|config)|onsite)/select/))?
+           (?:(?:(?:[^/]+/)+select/)?(?P<media>media/(?:guid/\d+/)?)|(?P<config>(?:[^/\?]+/(?:swf|config)|onsite)/select/))?
          |theplatform:)(?P<id>[^/\?&]+)'''
 
     _TESTS = [{
@@ -157,7 +154,7 @@ class ThePlatformIE(ThePlatformBaseIE):
         def hex_to_str(hex):
             return binascii.a2b_hex(hex)
 
-        relative_path = url.split('http://link.theplatform.com/s/')[1].split('?')[0]
+        relative_path = re.match(r'https?://link.theplatform.com/s/([^?]+)', url).group(1)
         clear_text = hex_to_str(flags + expiration_date + str_to_hex(relative_path))
         checksum = hmac.new(sig_key.encode('ascii'), clear_text, hashlib.sha1).hexdigest()
         sig = flags + expiration_date + checksum + str_to_hex(sig_secret)
@@ -173,10 +170,10 @@ class ThePlatformIE(ThePlatformBaseIE):
         if not provider_id:
             provider_id = 'dJ5BDC'
 
-        path = provider_id
+        path = provider_id + '/'
         if mobj.group('media'):
-            path += '/media'
-        path += '/' + video_id
+            path += mobj.group('media')
+        path += video_id
 
         qs_dict = compat_parse_qs(compat_urllib_parse_urlparse(url).query)
         if 'guid' in qs_dict: