Make sure the manifest never contains decomposed UTF-8.

--HG--
branch : distribute
extra : rebase_source : 0e0fb3beac56f66f12670ec69ebfd3996d12d912

2 files changed, 48 insertions, 3 deletions

diff --git a/setuptools/command/egg_info.py b/setuptools/command/egg_info.py
index 46cdf4e0..e932d46a 100755
--- a/setuptools/command/egg_info.py
+++ b/setuptools/command/egg_info.py
@@ -287,6 +287,19 @@ class FileList(FileList):
 
 
 
+def compose(path):
+    # Apple's HFS Plus returns decomposed UTF-8. Since just about
+    # everyone else chokes on it, we must make sure to return fully
+    # composed UTF-8 only.
+    if sys.getfilesystemencoding().lower() == 'utf-8':
+        from unicodedata import normalize
+        if sys.version_info >= (3,):
+            path = normalize('NFC', path)
+        else:
+            path = normalize('NFC', path.decode('utf-8')).encode('utf-8')
+    return path
+
+
 class manifest_maker(sdist):
 
     template = "MANIFEST.in"
@@ -311,6 +324,7 @@ class manifest_maker(sdist):
         self.prune_file_list()
         self.filelist.sort()
         self.filelist.remove_duplicates()
+        self.filelist.files = [compose(path) for path in self.filelist.files]
         self.write_manifest()
 
     def write_manifest (self):
diff --git a/setuptools/tests/test_sdist.py b/setuptools/tests/test_sdist.py
index 8d9ed922..34123545 100644
--- a/setuptools/tests/test_sdist.py
+++ b/setuptools/tests/test_sdist.py
@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 """sdist tests"""
 
 
@@ -74,6 +75,36 @@ class TestSdistTest(unittest.TestCase):
 
         manifest = cmd.filelist.files
 
-        self.assert_(os.path.join('sdist_test', 'a.txt') in manifest)
-        self.assert_(os.path.join('sdist_test', 'b.txt') in manifest)
-        self.assert_(os.path.join('sdist_test', 'c.rst') not in manifest)
+        self.assertTrue(os.path.join('sdist_test', 'a.txt') in manifest)
+        self.assertTrue(os.path.join('sdist_test', 'b.txt') in manifest)
+        self.assertTrue(os.path.join('sdist_test', 'c.rst') not in manifest)
+
+    def test_filelist_is_fully_composed(self):
+        # Test for #303. Requires HFS Plus to fail.
+
+        # Add file with non-ASCII filename
+        filename = os.path.join('sdist_test', 'smörbröd.py')
+        open(filename, 'w').close()
+
+        dist = Distribution(SETUP_ATTRS)
+        dist.script_name = 'setup.py'
+        cmd = sdist(dist)
+        cmd.ensure_finalized()
+
+        # squelch output
+        old_stdout = sys.stdout
+        old_stderr = sys.stderr
+        sys.stdout = StringIO()
+        sys.stderr = StringIO()
+        try:
+            cmd.run()
+        finally:
+            sys.stdout = old_stdout
+            sys.stderr = old_stderr
+
+        self.assertTrue(filename in cmd.filelist.files)
+
+
+def test_suite():
+    return unittest.defaultTestLoader.loadTestsFromName(__name__)
+