def test_gnu_sort_and_the_current_version_of_python_sort_strings_using_byte_value(
self):
''' DUMPER '''
unsorted_data_list = ['z\n', 'a\n', '\xc3\xb1\n']
unsorted_data = ''.join(unsorted_data_list)
sorted_data = ''.join(['a\n', 'z\n', '\xc3\xb1\n'])
path = make_temp_file(self.tmp_dir, unsorted_data)
sorted_file = gnu_sort(path, cache_dir=self.tmp_dir)
assertion_msg = (
'GNU Sort must sort comparing byte by byte (export '
'LC_ALL=C) to be faster and consistent with Python 2.')
if PY3:
with open(sorted_file, encoding='utf-8') as f:
assert f.read() == sorted_data, assertion_msg
else:
with open(sorted_file) as f:
assert f.read() == sorted_data, assertion_msg
os.unlink(path)
os.unlink(sorted_file)
python_sort = ''.join(sorted(unsorted_data_list))
assertion_msg = ('Current Python interpreter must sort strings '
'comparing byte by byte, it is important to use the '
'same ordering as the one used with GNU Sort. Note '
'Python 3 uses unicode by default.')
assert python_sort == sorted_data, assertion_msg
def test_gnu_sort_and_the_current_version_of_python_sort_strings_using_byte_value(
self):
''' DUMPER '''
unsorted_data_list = ['z\n', 'a\n', '\xc3\xb1\n']
unsorted_data = ''.join(unsorted_data_list)
sorted_data = ''.join(['a\n', 'z\n', '\xc3\xb1\n'])
path = make_temp_file(self.tmp_dir, unsorted_data)
sorted_file = gnu_sort(path, cache_dir=self.tmp_dir)
with open(sorted_file) as f:
eq_(
f.read(), sorted_data,
'GNU Sort must sort comparing byte by byte (export '
'LC_ALL=C) to be faster and consistent with Python 2.')
os.unlink(path)
os.unlink(sorted_file)
python_sort = ''.join(sorted(unsorted_data_list))
eq_(
python_sort, sorted_data,
'Current Python interpreter must sort strings comparing byte by '
'byte, it is important to use the same ordering as the one used '
'with GNU Sort. Note Python 3 uses unicode by default.')
def test_gnu_sort_can_sort_by_field(self):
''' DUMPER '''
unsorted_data = ''.join(['1,z\n', '2,a\n', '3,\xc3\xb1\n'])
sorted_data = ''.join(['2,a\n', '1,z\n', '3,\xc3\xb1\n'])
path = make_temp_file(self.tmp_dir, unsorted_data)
sorted_file = gnu_sort(path,
delimiter=',',
fieldspec='2',
cache_dir=self.tmp_dir)
with open(sorted_file) as f:
eq_(f.read(), sorted_data)
os.unlink(path)
os.unlink(sorted_file)
def test_gnu_sort_can_sort_by_field(self):
''' DUMPER '''
unsorted_data = ''.join(['1,z\n', '2,a\n', '3,\xc3\xb1\n'])
sorted_data = ''.join(['2,a\n', '1,z\n', '3,\xc3\xb1\n'])
path = make_temp_file(self.tmp_dir, unsorted_data)
sorted_file = gnu_sort(path,
delimiter=',',
fieldspec='2',
cache_dir=self.tmp_dir)
if PY3:
with open(sorted_file, encoding='utf-8') as f:
assert f.read() == sorted_data
else:
with open(sorted_file) as f:
assert f.read() == sorted_data
os.unlink(path)
os.unlink(sorted_file)