Treat filenames from the Wheel file

sigmavirus24 · sigmavirus24 · commit c767fbfcf61b · 2017-05-15T19:58:51.000-05:00
Filenames in a Wheel are returned as bytestrings. When we check to find METADATA files, we need to juggle the fact that strings are unicode by default and that the filenames are bytestrings. Closes pypagh-235
diff --git a/tests/test_wheel.py b/tests/test_wheel.py
@@ -28,3 +28,20 @@ def example_wheel(request):
 
 def test_version_parsing(example_wheel):
     assert example_wheel.py_version == 'py2.py3'
+
+
+def test_find_metadata_files():
+    names = [
+        b'package/lib/__init__.py',
+        b'package/lib/version.py',
+        b'package/METADATA.txt',
+        b'package/METADATA.json',
+        b'package/METADATA',
+    ]
+    expected = [
+        ['package', 'METADATA'],
+        ['package', 'METADATA.json'],
+        ['package', 'METADATA.txt'],
+    ]
+    candidates = wheel.Wheel.find_candidate_metadata_files(names)
+    assert expected == candidates
diff --git a/twine/wheel.py b/twine/wheel.py
@@ -38,6 +38,12 @@
     re.VERBOSE)
 
 
+def try_decode(s):
+    if isinstance(s, bytes):
+        return s.decode('utf8')
+    return s
+
+
 class Wheel(Distribution):
 
     def __init__(self, filename, metadata_version=None):
@@ -51,6 +57,15 @@ def py_version(self):
         wheel_info = wheel_file_re.match(self.basefilename)
         return wheel_info.group("pyver")
 
+    @staticmethod
+    def find_candidate_metadata_files(names):
+        """Filter files that may be METADATA files."""
+        tuples = [
+            x.split('/') for x in map(try_decode, names)
+            if 'METADATA' in x
+        ]
+        return [x[1] for x in sorted([(len(x), x) for x in tuples])]
+
     def read(self):
         fqn = os.path.abspath(os.path.normpath(self.filename))
         if not os.path.exists(fqn):
@@ -66,9 +81,7 @@ def read_file(name):
             raise ValueError('Not a known archive format: %s' % fqn)
 
         try:
-            tuples = [x.split('/') for x in names if 'METADATA' in x]
-            schwarz = sorted([(len(x), x) for x in tuples])
-            for path in [x[1] for x in schwarz]:
+            for path in self.find_candidate_metadata_files(names):
                 candidate = '/'.join(path)
                 data = read_file(candidate)
                 if b'Metadata-Version' in data:
