Update upstream source from tag 'upstream/2018.09.10'

[youtubedl] / youtube_dl / downloader / f4m.py

diff --git a/youtube_dl/downloader/f4m.py b/youtube_dl/downloader/f4m.py
index c8fde9a89093393132262f7b7d5ec60d83de4b8d..15e71be9a4f482e2739419c9fc512d423976dd9e 100644 (file)
--- a/youtube_dl/downloader/f4m.py
+++ b/youtube_dl/downloader/f4m.py
@@ -1,12 +1,12 @@
 from __future__ import division, unicode_literals
 
 from __future__ import division, unicode_literals
 

-import base64

 import io
 import itertools
 import time
 
 from .fragment import FragmentFD
 from ..compat import (
 import io
 import itertools
 import time
 
 from .fragment import FragmentFD
 from ..compat import (

+    compat_b64decode,

     compat_etree_fromstring,
     compat_urlparse,
     compat_urllib_error,
     compat_etree_fromstring,
     compat_urlparse,
     compat_urllib_error,


@@ -243,8 +243,17 @@ def remove_encrypted_media(media):
                        media))
 
 
                        media))
 
 

-def _add_ns(prop):
-    return '{http://ns.adobe.com/f4m/1.0}%s' % prop
+def _add_ns(prop, ver=1):
+    return '{http://ns.adobe.com/f4m/%d.0}%s' % (ver, prop)
+
+
+def get_base_url(manifest):
+    base_url = xpath_text(
+        manifest, [_add_ns('baseURL'), _add_ns('baseURL', 2)],
+        'base URL', default=None)
+    if base_url:
+        base_url = base_url.strip()
+    return base_url

 
 
 class F4mFD(FragmentFD):
 
 
 class F4mFD(FragmentFD):


@@ -303,7 +312,7 @@ class F4mFD(FragmentFD):
             boot_info = self._get_bootstrap_from_url(bootstrap_url)
         else:
             bootstrap_url = None
             boot_info = self._get_bootstrap_from_url(bootstrap_url)
         else:
             bootstrap_url = None

-            bootstrap = base64.b64decode(node.text.encode('ascii'))
+            bootstrap = compat_b64decode(node.text)

             boot_info = read_bootstrap_info(bootstrap)
         return boot_info, bootstrap_url
 
             boot_info = read_bootstrap_info(bootstrap)
         return boot_info, bootstrap_url
 


@@ -330,17 +339,17 @@ class F4mFD(FragmentFD):
             rate, media = list(filter(
                 lambda f: int(f[0]) == requested_bitrate, formats))[0]
 
             rate, media = list(filter(
                 lambda f: int(f[0]) == requested_bitrate, formats))[0]
 

-        base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
+        # Prefer baseURL for relative URLs as per 11.2 of F4M 3.0 spec.
+        man_base_url = get_base_url(doc) or man_url
+
+        base_url = compat_urlparse.urljoin(man_base_url, media.attrib['url'])

         bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
         bootstrap_node = doc.find(_add_ns('bootstrapInfo'))

-        # From Adobe F4M 3.0 spec:
-        # The <baseURL> element SHALL be the base URL for all relative
-        # (HTTP-based) URLs in the manifest. If <baseURL> is not present, said
-        # URLs should be relative to the location of the containing document.
-        boot_info, bootstrap_url = self._parse_bootstrap_node(bootstrap_node, man_url)
+        boot_info, bootstrap_url = self._parse_bootstrap_node(
+            bootstrap_node, man_base_url)

         live = boot_info['live']
         metadata_node = media.find(_add_ns('metadata'))
         if metadata_node is not None:
         live = boot_info['live']
         metadata_node = media.find(_add_ns('metadata'))
         if metadata_node is not None:

-            metadata = base64.b64decode(metadata_node.text.encode('ascii'))
+            metadata = compat_b64decode(metadata_node.text)

         else:
             metadata = None
 
         else:
             metadata = None