def test_utf8_filename(self):
if sys.version_info >= (3, ):
filename = 'héhé.html'.encode('utf-8')
else:
filename = 'héhé.html'
# Windows does not like byte filenames under Python 3
if sys.platform == 'win32' and sys.version_info >= (3, ):
filename = filename.decode('utf-8')
self.create_file(filename)
# HFS Plus uses decomposed UTF-8
if sys.platform == 'darwin':
filename = decompose(filename)
self.add_file(filename)
self.assertEqual([fn for fn in os.listdir(os.curdir) if fn[0] != '.'],
[fsdecode(filename)])
# git ls-files returns composed UTF-8
if sys.platform == 'darwin':
filename = compose(filename)
self.assertEqual(set(self.listfiles()), set([fsdecode(filename)]))
def test_utf8_filename(self):
if sys.version_info >= (3,):
filename = 'héhé.html'.encode('utf-8')
else:
filename = 'héhé.html'
# Windows does not like byte filenames under Python 3
if sys.platform == 'win32' and sys.version_info >= (3,):
filename = filename.decode('utf-8')
self.create_file(filename)
# HFS Plus uses decomposed UTF-8
if sys.platform == 'darwin':
filename = decompose(filename)
self.add_file(filename)
self.assertEqual(
[fn for fn in os.listdir(os.curdir) if fn[0] != '.'],
[fsdecode(filename)])
# git ls-files returns composed UTF-8
if sys.platform == 'darwin':
filename = compose(filename)
self.assertEqual(
set(self.listfiles()),
set([fsdecode(filename)]))
def test_nonascii_filename(self):
filename = 'héhé.html'
self.create_and_add(filename)
# HFS Plus uses decomposed UTF-8
if sys.platform == 'darwin':
filename = decompose(filename)
self.assertEqual([fn for fn in os.listdir(os.curdir) if fn[0] != '.'],
[filename])
# git ls-files returns composed UTF-8
if sys.platform == 'darwin':
filename = compose(filename)
self.assertEqual(set(self.listfiles()), set([filename]))
def test_nonascii_filename(self):
filename = 'héhé.html'
self.create_and_add(filename)
# HFS Plus uses decomposed UTF-8
if sys.platform == 'darwin':
filename = decompose(filename)
self.assertEqual(
[fn for fn in os.listdir(os.curdir) if fn[0] != '.'],
[filename])
# git ls-files returns composed UTF-8
if sys.platform == 'darwin':
filename = compose(filename)
self.assertEqual(
set(self.listfiles()),
set([filename]))
def listfiles(dirname=''):
git_files = gitlsfiles(dirname)
cwd = realpath(dirname or os.curdir)
prefix_length = len(cwd) + 1
if sys.version_info >= (2, 6):
walker = os.walk(cwd, followlinks=True)
else:
walker = os.walk(cwd)
for root, dirs, files in walker:
for file in files:
filename = join(root, file)
if posix(realpath(filename)) in git_files:
yield filename[prefix_length:]
if __name__ == '__main__':
if len(sys.argv) > 1:
dirname = sys.argv[1]
else:
dirname = ''
for filename in listfiles(dirname):
try:
print(compose(filename))
except UnicodeEncodeError:
print(repr(filename)[1:-1])
def listfiles(dirname=""):
git_files = gitlsfiles(dirname)
cwd = realpath(dirname or os.curdir)
prefix_length = len(cwd) + 1
if sys.version_info >= (2, 6):
walker = os.walk(cwd, followlinks=True)
else:
walker = os.walk(cwd)
for root, dirs, files in walker:
for file in files:
filename = join(root, file)
if posix(realpath(filename)) in git_files:
yield filename[prefix_length:]
if __name__ == "__main__":
if len(sys.argv) > 1:
dirname = sys.argv[1]
else:
dirname = ""
for filename in listfiles(dirname):
try:
print(compose(filename))
except UnicodeEncodeError:
print(repr(filename)[1:-1])