Make symlink work with Unicode account names

Also, ensure that the stored symlink headers really *are* url-encoded as
we've been assuming.

Change-Id: I1f300d69bec43f0deb430294da05a4ec04308040
Related-Bug: 1774238
Closes-Bug: #1821240

swift/common/middleware/symlink.py

@@ -161,7 +161,7 @@ from cgi import parse_header
 from six.moves.urllib.parse import unquote
 
 from swift.common.utils import get_logger, register_swift_info, split_path, \
-    MD5_OF_EMPTY_STRING, closing_if_possible
+    MD5_OF_EMPTY_STRING, closing_if_possible, quote
 from swift.common.constraints import check_account_format
 from swift.common.wsgi import WSGIContext, make_subrequest
 from swift.common.request_helpers import get_sys_meta_prefix, \
@@ -208,6 +208,7 @@ def _check_symlink_header(req):
         req, TGT_OBJ_SYMLINK_HDR, 2,
         'X-Symlink-Target header must be of the '
         'form <container name>/<object name>')
+    req.headers[TGT_OBJ_SYMLINK_HDR] = quote('%s/%s' % (container, obj))
 
     # Check account format if it exists
     account = check_account_format(
@@ -217,7 +218,9 @@ def _check_symlink_header(req):
     # Extract request path
     _junk, req_acc, req_cont, req_obj = req.split_path(4, 4, True)
 
-    if not account:
+    if account:
+        req.headers[TGT_ACCT_SYMLINK_HDR] = quote(account)
+    else:
         account = req_acc
 
     # Check if symlink targets the symlink itself or not
@@ -378,9 +381,9 @@ class SymlinkObjectContext(WSGIContext):
             :returns: new request for target path if it's symlink otherwise
                       None
             """
-            version, account, _junk = split_path(req.path, 2, 3, True)
+            version, account, _junk = req.split_path(2, 3, True)
             account = self._response_header_value(
-                TGT_ACCT_SYSMETA_SYMLINK_HDR) or account
+                TGT_ACCT_SYSMETA_SYMLINK_HDR) or quote(account)
             target_path = os.path.join(
                 '/', version, account,
                 symlink_target.lstrip('/'))
@@ -485,7 +488,7 @@ class SymlinkObjectContext(WSGIContext):
         if tgt_co:
             version, account, _junk = req.split_path(2, 3, True)
             target_acc = self._response_header_value(
-                TGT_ACCT_SYSMETA_SYMLINK_HDR) or account
+                TGT_ACCT_SYSMETA_SYMLINK_HDR) or quote(account)
             location_hdr = os.path.join(
                 '/', version, target_acc, tgt_co)
             req.environ['swift.leave_relative_location'] = True

test/functional/test_symlink.py

@@ -270,23 +270,45 @@ class TestSymlink(Base):
         target_obj = 'dealde%2Fl04 011e%204c8df/flash.png'
         link_obj = uuid4().hex
 
-        # Now let's write a new target object and symlink will be able to
-        # return object
+        # create target using unnormalized path
         resp = retry(
             self._make_request, method='PUT', container=self.env.tgt_cont,
             obj=target_obj, body=TARGET_BODY)
-
         self.assertEqual(resp.status, 201)
+        # you can get it using either name
+        resp = retry(
+            self._make_request, method='GET', container=self.env.tgt_cont,
+            obj=target_obj)
+        self.assertEqual(resp.status, 200)
+        self.assertEqual(resp.content, TARGET_BODY)
+        normalized_quoted_obj = 'dealde/l04%20011e%204c8df/flash.png'
+        self.assertEqual(normalized_quoted_obj, urllib.parse.quote(
+            urllib.parse.unquote(target_obj)))
+        resp = retry(
+            self._make_request, method='GET', container=self.env.tgt_cont,
+            obj=normalized_quoted_obj)
+        self.assertEqual(resp.status, 200)
+        self.assertEqual(resp.content, TARGET_BODY)
 
-        # PUT symlink
+        # create a symlink using the un-normalized target path
         self._test_put_symlink(link_cont=self.env.link_cont, link_obj=link_obj,
                                tgt_cont=self.env.tgt_cont,
                                tgt_obj=target_obj)
-
+        # and it's normalized
         self._assertSymlink(
             self.env.link_cont, link_obj,
-            expected_content_location="%s/%s" % (self.env.tgt_cont,
-                                                 target_obj))
+            expected_content_location='%s/%s' % (
+                self.env.tgt_cont, normalized_quoted_obj))
+
+        # create a symlink using the normalized target path
+        self._test_put_symlink(link_cont=self.env.link_cont, link_obj=link_obj,
+                               tgt_cont=self.env.tgt_cont,
+                               tgt_obj=normalized_quoted_obj)
+        # and it's ALSO normalized
+        self._assertSymlink(
+            self.env.link_cont, link_obj,
+            expected_content_location='%s/%s' % (
+                self.env.tgt_cont, normalized_quoted_obj))
 
     def test_symlink_put_head_get(self):
         link_obj = uuid4().hex

test/functional/test_versioned_writes.py

@@ -18,7 +18,7 @@ from copy import deepcopy
 import json
 import time
 import unittest2
-from six.moves.urllib.parse import quote
+from six.moves.urllib.parse import quote, unquote
 
 import test.functional as tf
 
@@ -652,7 +652,7 @@ class TestObjectVersioning(Base):
         tgt_b.write("bbbbb")
 
         symlink_name = Utils.create_name()
-        sym_tgt_header = '%s/%s' % (container.name, tgt_a_name)
+        sym_tgt_header = quote(unquote('%s/%s' % (container.name, tgt_a_name)))
         sym_headers_a = {'X-Symlink-Target': sym_tgt_header}
         symlink = container.file(symlink_name)
         symlink.write("", hdrs=sym_headers_a)
@@ -684,7 +684,8 @@ class TestObjectVersioning(Base):
         sym_info = symlink.info(parms={'symlink': 'get'})
         self.assertEqual("aaaaa", symlink.read())
         self.assertEqual(MD5_OF_EMPTY_STRING, sym_info['etag'])
-        self.assertEqual('%s/%s' % (self.env.container.name, target.name),
+        self.assertEqual(
+            quote(unquote('%s/%s' % (self.env.container.name, target.name))),
             sym_info['x_symlink_target'])
 
     def _setup_symlink(self):