[Python-checkins] bpo-35907, CVE-2019-9948: urllib rejects local_file:// scheme (GH-13513)

[Ned Deily]

https://github.com/python/cpython/commit/4f06dae5d8d4400ba38d8502da620f07d4a5696e
commit: 4f06dae5d8d4400ba38d8502da620f07d4a5696e
branch: 3.6
author: Victor Stinner <vstinner at redhat.com>
committer: Ned Deily <nad at python.org>
date: 2019-05-28T22:30:47-04:00
summary:

bpo-35907, CVE-2019-9948: urllib rejects local_file:// scheme (GH-13513)

CVE-2019-9948: Avoid file reading by disallowing local-file:// and
local_file:// URL schemes in URLopener().open() and
URLopener().retrieve() of urllib.request.

Co-Authored-By: SH <push0ebp at gmail.com>
(cherry picked from commit 0c2b6a3943aa7b022e8eb4bfd9bffcddebf9a587)
(cherry picked from commit 34bab215596671d0dec2066ae7d7450cd73f638b)

files:
A Misc/NEWS.d/next/Security/2019-05-21-23-20-18.bpo-35907.NC_zNK.rst
M Lib/test/test_urllib.py
M Lib/urllib/request.py

diff --git a/Lib/test/test_urllib.py b/Lib/test/test_urllib.py
index 649a5b81575b..0061a5297cb3 100644
--- a/Lib/test/test_urllib.py
+++ b/Lib/test/test_urllib.py
@@ -16,6 +16,7 @@
     ssl = None
 import sys
 import tempfile
+import warnings
 from nturl2path import url2pathname, pathname2url
 
 from base64 import b64encode
@@ -1463,6 +1464,23 @@ def open_spam(self, url):
                 "spam://c:|windows%/:=&?~#+!$,;'@()*[]|/path/"),
                 "//c:|windows%/:=&?~#+!$,;'@()*[]|/path/")
 
+    def test_local_file_open(self):
+        # bpo-35907, CVE-2019-9948: urllib must reject local_file:// scheme
+        class DummyURLopener(urllib.request.URLopener):
+            def open_local_file(self, url):
+                return url
+
+        with warnings.catch_warnings(record=True):
+            warnings.simplefilter("ignore", DeprecationWarning)
+
+            for url in ('local_file://example', 'local-file://example'):
+                self.assertRaises(OSError, urllib.request.urlopen, url)
+                self.assertRaises(OSError, urllib.request.URLopener().open, url)
+                self.assertRaises(OSError, urllib.request.URLopener().retrieve, url)
+                self.assertRaises(OSError, DummyURLopener().open, url)
+                self.assertRaises(OSError, DummyURLopener().retrieve, url)
+
+
 # Just commented them out.
 # Can't really tell why keep failing in windows and sparc.
 # Everywhere else they work ok, but on those machines, sometimes
diff --git a/Lib/urllib/request.py b/Lib/urllib/request.py
index d28f2f8369c7..c9945d951821 100644
--- a/Lib/urllib/request.py
+++ b/Lib/urllib/request.py
@@ -1747,7 +1747,7 @@ def open(self, fullurl, data=None):
         name = 'open_' + urltype
         self.type = urltype
         name = name.replace('-', '_')
-        if not hasattr(self, name):
+        if not hasattr(self, name) or name == 'open_local_file':
             if proxy:
                 return self.open_unknown_proxy(proxy, fullurl, data)
             else:
diff --git a/Misc/NEWS.d/next/Security/2019-05-21-23-20-18.bpo-35907.NC_zNK.rst b/Misc/NEWS.d/next/Security/2019-05-21-23-20-18.bpo-35907.NC_zNK.rst
new file mode 100644
index 000000000000..37b567a5b6f9
--- /dev/null
+++ b/Misc/NEWS.d/next/Security/2019-05-21-23-20-18.bpo-35907.NC_zNK.rst
@@ -0,0 +1,3 @@
+CVE-2019-9948: Avoid file reading by disallowing ``local-file://`` and
+``local_file://`` URL schemes in ``URLopener().open()`` and
+``URLopener().retrieve()`` of :mod:`urllib.request`.