From e14f963a22bebde41e8ee31371d295c6bbf86a82 Mon Sep 17 00:00:00 2001
From: Nisha Agarwal <agarwalnisha1980@gmail.com>
Date: Tue, 13 Jun 2017 15:42:16 +0000
Subject: [PATCH] Fixes hashing issues for py3.5

This fix uses encoding 'utf-8' before doing hashing.

Closes-Bug: 1697719
Co-Authored-By: Vladyslav Drok <vdrok@mirantis.com>
Change-Id: Ia61bed31ebd98032847504d6c96080330088d9ec
---
 ironic/common/utils.py                 | 13 ++++++++++---
 ironic/tests/unit/common/test_utils.py | 18 ++++++++++++++++++
 2 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/ironic/common/utils.py b/ironic/common/utils.py
index 1d1b5a6b17..c6085d74ed 100644
--- a/ironic/common/utils.py
+++ b/ironic/common/utils.py
@@ -281,8 +281,13 @@ def hash_file(file_like_object, hash_algo='md5'):
     :returns: a condensed digest of the bytes of contents.
     """
     checksum = _get_hash_object(hash_algo)
-    for chunk in iter(lambda: file_like_object.read(32768), b''):
-        checksum.update(chunk)
+    while True:
+        chunk = file_like_object.read(32768)
+        if not chunk:
+            break
+        encoded_chunk = (chunk.encode(encoding='utf-8')
+                         if isinstance(chunk, six.string_types) else chunk)
+        checksum.update(encoded_chunk)
     return checksum.hexdigest()
 
 
@@ -298,7 +303,9 @@ def file_has_content(path, content, hash_algo='md5'):
     with open(path, 'rb') as existing:
         file_hash_hex = hash_file(existing, hash_algo=hash_algo)
     ref_hash = _get_hash_object(hash_algo)
-    ref_hash.update(content)
+    encoded_content = (content.encode(encoding='utf-8')
+                       if isinstance(content, six.string_types) else content)
+    ref_hash.update(encoded_content)
     return file_hash_hex == ref_hash.hexdigest()
 
 
diff --git a/ironic/tests/unit/common/test_utils.py b/ironic/tests/unit/common/test_utils.py
index ef040b9143..6c0096eaf2 100644
--- a/ironic/tests/unit/common/test_utils.py
+++ b/ironic/tests/unit/common/test_utils.py
@@ -127,6 +127,16 @@ class GenericUtilsTestCase(base.TestCase):
         # | THEN |
         self.assertEqual(expected, actual)
 
+    def test_hash_file_for_md5_not_binary(self):
+        # | GIVEN |
+        data = u'Mary had a little lamb, its fleece as white as sno\u0449'
+        file_like_object = six.StringIO(data)
+        expected = hashlib.md5(data.encode('utf-8')).hexdigest()
+        # | WHEN |
+        actual = utils.hash_file(file_like_object)  # using default, 'md5'
+        # | THEN |
+        self.assertEqual(expected, actual)
+
     def test_hash_file_for_sha1(self):
         # | GIVEN |
         data = b'Mary had a little lamb, its fleece as white as snow'
@@ -163,6 +173,14 @@ class GenericUtilsTestCase(base.TestCase):
             self.assertTrue(utils.file_has_content('foo', ref))
             mopen.assert_called_once_with('foo', 'rb')
 
+    def test_file_has_content_equal_not_binary(self):
+        data = u'Mary had a little lamb, its fleece as white as sno\u0449'
+        ref = data
+        with mock.patch('ironic.common.utils.open',
+                        mock.mock_open(read_data=data)) as mopen:
+            self.assertTrue(utils.file_has_content('foo', ref))
+            mopen.assert_called_once_with('foo', 'rb')
+
     def test_file_has_content_differ(self):
         data = b'Mary had a little lamb, its fleece as white as snow'
         ref = data + b'!'