Keep 'surrogateescape' when reading the manifest, to avoid breaking on bad input.

--HG--
branch : distribute
extra : rebase_source : 6f894cd508e73fae0ad02860654df5181055ba4d

2 files changed, 8 insertions, 4 deletions

diff --git a/setuptools/command/sdist.py b/setuptools/command/sdist.py
index d5259c2b..79eed214 100755
--- a/setuptools/command/sdist.py
+++ b/setuptools/command/sdist.py
@@ -283,7 +283,8 @@ class sdist(_sdist):
         manifest = open(self.manifest, 'rbU')
         for line in manifest:
             if sys.version_info >= (3,):
-                line = line.decode('UTF-8')
+                # Don't break if surrogates have crept into the manifest
+                line = line.decode('UTF-8', 'surrogateescape')
             # ignore comments and blank lines
             line = line.strip()
             if line.startswith('#') or not line:
diff --git a/setuptools/tests/test_sdist.py b/setuptools/tests/test_sdist.py
index 378015a8..04b3db66 100644
--- a/setuptools/tests/test_sdist.py
+++ b/setuptools/tests/test_sdist.py
@@ -262,7 +262,7 @@ class TestSdistTest(unittest.TestCase):
     # Python 3 only
     if sys.version_info >= (3,):
 
-        def test_read_manifest_rejects_surrogates(self):
+        def test_manifest_is_read_with_surrogateescape_error_handler(self):
             # Test for #303.
 
             # This is hard to test on HFS Plus because it quotes unknown
@@ -286,11 +286,14 @@ class TestSdistTest(unittest.TestCase):
                 manifest = open(cmd.manifest, 'ab')
                 manifest.write(filename+b('\n'))
                 manifest.close()
+                # Re-read manifest
+                try:
+                    cmd.read_manifest()
+                except UnicodeDecodeError, e:
+                    self.fail(e)
             finally:
                 unquiet()
 
-            self.assertRaises(UnicodeDecodeError, cmd.read_manifest)
-
     def test_sdist_with_utf8_encoded_filename(self):
         # Test for #303.
         dist = Distribution(SETUP_ATTRS)