Merge pull request #1035 from whummer/fix/py2-s3-names-unicode

Allow non-ascii characters in request URLs

moto/server.py

@@ -3,6 +3,7 @@ import json
 import re
 import sys
 import argparse
+import six
 
 from six.moves.urllib.parse import urlencode
 
@@ -47,6 +48,13 @@ class DomainDispatcherApplication(object):
 
     def get_application(self, environ):
         path_info = environ.get('PATH_INFO', '')
+
+        # The URL path might contain non-ASCII text, for instance unicode S3 bucket names
+        if six.PY2 and isinstance(path_info, str):
+            path_info = six.u(path_info)
+        if six.PY3 and isinstance(path_info, six.binary_type):
+            path_info = path_info.decode('utf-8')
+
         if path_info.startswith("/moto-api") or path_info == "/favicon.ico":
             host = "moto_api"
         elif path_info.startswith("/latest/meta-data/"):

tests/test_s3/test_server.py

@@ -1,3 +1,5 @@
+# coding=utf-8
+
 from __future__ import unicode_literals
 import sure  # noqa
 
@@ -78,3 +80,18 @@ def test_s3_server_post_without_content_length():
 
     res = test_client.post('/', "https://tester.localhost:5000/", environ_overrides={'CONTENT_LENGTH': ''})
     res.status_code.should.equal(411)
+
+
+def test_s3_server_post_unicode_bucket_key():
+    # Make sure that we can deal with non-ascii characters in request URLs (e.g., S3 object names)
+    dispatcher = server.DomainDispatcherApplication(server.create_backend_app)
+    backend_app = dispatcher.get_application({
+        'HTTP_HOST': 's3.amazonaws.com',
+        'PATH_INFO': '/test-bucket/test-object-てすと'
+    })
+    assert backend_app
+    backend_app = dispatcher.get_application({
+        'HTTP_HOST': 's3.amazonaws.com',
+        'PATH_INFO': '/test-bucket/test-object-てすと'.encode('utf-8')
+    })
+    assert backend_app