packaging: don't use locale encoding to compute MD5 checksums
Open the file in binary mode or use UTF-8 encoding.
diff --git a/Lib/packaging/command/install_distinfo.py b/Lib/packaging/command/install_distinfo.py
index 41fe734..a25a15a 100644
--- a/Lib/packaging/command/install_distinfo.py
+++ b/Lib/packaging/command/install_distinfo.py
@@ -133,9 +133,9 @@
writer.writerow((fpath, '', ''))
else:
size = os.path.getsize(fpath)
- with open(fpath, 'r') as fp:
+ with open(fpath, 'rb') as fp:
hash = hashlib.md5()
- hash.update(fp.read().encode())
+ hash.update(fp.read())
md5sum = hash.hexdigest()
writer.writerow((fpath, md5sum, size))
diff --git a/Lib/packaging/create.py b/Lib/packaging/create.py
index 837d0b6..0676cf1 100644
--- a/Lib/packaging/create.py
+++ b/Lib/packaging/create.py
@@ -400,10 +400,10 @@
self.data['description']).lower().encode())
ref = ref.digest()
for readme in glob.glob('README*'):
- with open(readme) as fp:
+ with open(readme, encoding='utf-8') as fp:
contents = fp.read()
- val = md5(re.sub('\s', '',
- contents.lower()).encode()).digest()
+ contents = re.sub('\s', '', contents.lower()).encode()
+ val = md5(contents).digest()
if val == ref:
del data['description']
data['description-file'] = readme
diff --git a/Lib/packaging/tests/test_command_install_distinfo.py b/Lib/packaging/tests/test_command_install_distinfo.py
index 3d33691..6d40f66 100644
--- a/Lib/packaging/tests/test_command_install_distinfo.py
+++ b/Lib/packaging/tests/test_command_install_distinfo.py
@@ -168,8 +168,8 @@
else:
size = os.path.getsize(f)
md5 = hashlib.md5()
- with open(f) as fp:
- md5.update(fp.read().encode())
+ with open(f, 'rb') as fp:
+ md5.update(fp.read())
hash = md5.hexdigest()
expected.append([f, hash, str(size)])