From 41cdb942636b1fd1b91bf0898b57e6af6d3a688c Mon Sep 17 00:00:00 2001
From: Alistair Coles <alistairncoles@gmail.com>
Date: Mon, 4 Dec 2017 16:16:10 -0800
Subject: [PATCH] Make 'directory' objects have zero-length etags in container
 listings

'Directory' objects are zero length so their etags can be forced to the
hash of an empty string when pfs middleware constructs container listings.
---
 pfs_middleware/pfs_middleware/middleware.py | 17 +++++++++--------
 pfs_middleware/tests/test_pfs_middleware.py |  8 ++++++--
 2 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/pfs_middleware/pfs_middleware/middleware.py b/pfs_middleware/pfs_middleware/middleware.py
index d51cb146b..838ab1f2c 100644
--- a/pfs_middleware/pfs_middleware/middleware.py
+++ b/pfs_middleware/pfs_middleware/middleware.py
@@ -417,7 +417,10 @@ def extract_container_metadata_from_headers(req):
     return meta_headers
 
 
-def best_possible_etag(obj_metadata, account_name, ino, num_writes):
+def best_possible_etag(obj_metadata, account_name, ino, num_writes,
+                       is_dir=False):
+    if is_dir:
+        return EMPTY_OBJECT_ETAG
     if ORIGINAL_MD5_HEADER in obj_metadata:
         val = obj_metadata[ORIGINAL_MD5_HEADER]
         try:
@@ -1159,7 +1162,7 @@ def _json_container_get_response(self, container_entries, account_name):
                                                   ent["IsDir"])
             etag = best_possible_etag(
                 obj_metadata, account_name,
-                ent["InodeNumber"], ent["NumWrites"])
+                ent["InodeNumber"], ent["NumWrites"], is_dir=ent["IsDir"])
             json_entry = {
                 "name": name,
                 "bytes": size,
@@ -1183,7 +1186,8 @@ def _xml_container_get_response(self, container_entries, account_name,
             etag = best_possible_etag(
                 obj_metadata, account_name,
                 container_entry["InodeNumber"],
-                container_entry["NumWrites"])
+                container_entry["NumWrites"],
+                is_dir=container_entry["IsDir"])
             container_node = ET.Element('object')
 
             name_node = ET.Element('name')
@@ -1591,11 +1595,8 @@ def head_object(self, ctx):
             headers["Content-Type"] = guess_content_type(req.path, is_dir)
 
         headers["Content-Length"] = file_size
-        if is_dir:
-            headers["ETag"] = EMPTY_OBJECT_ETAG
-        else:
-            headers["ETag"] = best_possible_etag(
-                headers, ctx.account_name, ino, num_writes)
+        headers["ETag"] = best_possible_etag(
+            headers, ctx.account_name, ino, num_writes, is_dir=is_dir)
         headers["Last-Modified"] = last_modified_from_epoch_ns(
             last_modified_ns)
         headers["X-Timestamp"] = x_timestamp_from_epoch_ns(
diff --git a/pfs_middleware/tests/test_pfs_middleware.py b/pfs_middleware/tests/test_pfs_middleware.py
index bb897fd23..79004472c 100644
--- a/pfs_middleware/tests/test_pfs_middleware.py
+++ b/pfs_middleware/tests/test_pfs_middleware.py
@@ -1585,8 +1585,7 @@ def test_json(self):
             "name": "images",
             "bytes": 0,
             "content_type": "application/directory",
-            "hash": mware.construct_etag(
-                "AUTH_test", 2489682, 0),
+            "hash": "d41d8cd98f00b204e9800998ecf8427e",
             "last_modified": "2016-08-23T01:30:16.359210"})
         self.assertEqual(resp_data[1], {
             "name": "images/avocado.png",
@@ -1693,6 +1692,11 @@ def test_xml(self):
         self.assertEqual(name_node.tag, 'name')
         self.assertEqual(name_node.text, 'images')
 
+        hash_node = obj_attr_tags[1]
+        self.assertEqual(hash_node.tag, 'hash')
+        self.assertEqual(hash_node.text, "d41d8cd98f00b204e9800998ecf8427e")
+        self.assertEqual(hash_node.attrib, {})
+
         content_type_node = obj_attr_tags[3]
         self.assertEqual(content_type_node.tag, 'content_type')
         self.assertEqual(content_type_node.text, 'application/directory')