308 lines
10 KiB
Python
308 lines
10 KiB
Python
"""Test script for the dumbdbm module
|
|
Original by Roger E. Masse
|
|
"""
|
|
|
|
import io
|
|
import operator
|
|
import os
|
|
import stat
|
|
import unittest
|
|
import dbm.dumb as dumbdbm
|
|
from test import support
|
|
from functools import partial
|
|
|
|
_fname = support.TESTFN
|
|
|
|
def _delete_files():
|
|
for ext in [".dir", ".dat", ".bak"]:
|
|
try:
|
|
os.unlink(_fname + ext)
|
|
except OSError:
|
|
pass
|
|
|
|
class DumbDBMTestCase(unittest.TestCase):
|
|
_dict = {b'0': b'',
|
|
b'a': b'Python:',
|
|
b'b': b'Programming',
|
|
b'c': b'the',
|
|
b'd': b'way',
|
|
b'f': b'Guido',
|
|
b'g': b'intended',
|
|
'\u00fc'.encode('utf-8') : b'!',
|
|
}
|
|
|
|
def test_dumbdbm_creation(self):
|
|
f = dumbdbm.open(_fname, 'c')
|
|
self.assertEqual(list(f.keys()), [])
|
|
for key in self._dict:
|
|
f[key] = self._dict[key]
|
|
self.read_helper(f)
|
|
f.close()
|
|
|
|
@unittest.skipUnless(hasattr(os, 'umask'), 'test needs os.umask()')
|
|
@unittest.skipUnless(hasattr(os, 'chmod'), 'test needs os.chmod()')
|
|
def test_dumbdbm_creation_mode(self):
|
|
try:
|
|
old_umask = os.umask(0o002)
|
|
f = dumbdbm.open(_fname, 'c', 0o637)
|
|
f.close()
|
|
finally:
|
|
os.umask(old_umask)
|
|
|
|
expected_mode = 0o635
|
|
if os.name != 'posix':
|
|
# Windows only supports setting the read-only attribute.
|
|
# This shouldn't fail, but doesn't work like Unix either.
|
|
expected_mode = 0o666
|
|
|
|
import stat
|
|
st = os.stat(_fname + '.dat')
|
|
self.assertEqual(stat.S_IMODE(st.st_mode), expected_mode)
|
|
st = os.stat(_fname + '.dir')
|
|
self.assertEqual(stat.S_IMODE(st.st_mode), expected_mode)
|
|
|
|
def test_close_twice(self):
|
|
f = dumbdbm.open(_fname)
|
|
f[b'a'] = b'b'
|
|
self.assertEqual(f[b'a'], b'b')
|
|
f.close()
|
|
f.close()
|
|
|
|
def test_dumbdbm_modification(self):
|
|
self.init_db()
|
|
f = dumbdbm.open(_fname, 'w')
|
|
self._dict[b'g'] = f[b'g'] = b"indented"
|
|
self.read_helper(f)
|
|
f.close()
|
|
|
|
def test_dumbdbm_read(self):
|
|
self.init_db()
|
|
f = dumbdbm.open(_fname, 'r')
|
|
self.read_helper(f)
|
|
with self.assertRaisesRegex(ValueError,
|
|
'The database is opened for reading only'):
|
|
f[b'g'] = b'x'
|
|
with self.assertRaisesRegex(ValueError,
|
|
'The database is opened for reading only'):
|
|
del f[b'a']
|
|
f.close()
|
|
|
|
def test_dumbdbm_keys(self):
|
|
self.init_db()
|
|
f = dumbdbm.open(_fname)
|
|
keys = self.keys_helper(f)
|
|
f.close()
|
|
|
|
def test_write_contains(self):
|
|
f = dumbdbm.open(_fname)
|
|
f[b'1'] = b'hello'
|
|
self.assertIn(b'1', f)
|
|
f.close()
|
|
|
|
def test_write_write_read(self):
|
|
# test for bug #482460
|
|
f = dumbdbm.open(_fname)
|
|
f[b'1'] = b'hello'
|
|
f[b'1'] = b'hello2'
|
|
f.close()
|
|
f = dumbdbm.open(_fname)
|
|
self.assertEqual(f[b'1'], b'hello2')
|
|
f.close()
|
|
|
|
def test_str_read(self):
|
|
self.init_db()
|
|
f = dumbdbm.open(_fname, 'r')
|
|
self.assertEqual(f['\u00fc'], self._dict['\u00fc'.encode('utf-8')])
|
|
|
|
def test_str_write_contains(self):
|
|
self.init_db()
|
|
f = dumbdbm.open(_fname)
|
|
f['\u00fc'] = b'!'
|
|
f['1'] = 'a'
|
|
f.close()
|
|
f = dumbdbm.open(_fname, 'r')
|
|
self.assertIn('\u00fc', f)
|
|
self.assertEqual(f['\u00fc'.encode('utf-8')],
|
|
self._dict['\u00fc'.encode('utf-8')])
|
|
self.assertEqual(f[b'1'], b'a')
|
|
|
|
def test_line_endings(self):
|
|
# test for bug #1172763: dumbdbm would die if the line endings
|
|
# weren't what was expected.
|
|
f = dumbdbm.open(_fname)
|
|
f[b'1'] = b'hello'
|
|
f[b'2'] = b'hello2'
|
|
f.close()
|
|
|
|
# Mangle the file by changing the line separator to Windows or Unix
|
|
with io.open(_fname + '.dir', 'rb') as file:
|
|
data = file.read()
|
|
if os.linesep == '\n':
|
|
data = data.replace(b'\n', b'\r\n')
|
|
else:
|
|
data = data.replace(b'\r\n', b'\n')
|
|
with io.open(_fname + '.dir', 'wb') as file:
|
|
file.write(data)
|
|
|
|
f = dumbdbm.open(_fname)
|
|
self.assertEqual(f[b'1'], b'hello')
|
|
self.assertEqual(f[b'2'], b'hello2')
|
|
|
|
|
|
def read_helper(self, f):
|
|
keys = self.keys_helper(f)
|
|
for key in self._dict:
|
|
self.assertEqual(self._dict[key], f[key])
|
|
|
|
def init_db(self):
|
|
f = dumbdbm.open(_fname, 'n')
|
|
for k in self._dict:
|
|
f[k] = self._dict[k]
|
|
f.close()
|
|
|
|
def keys_helper(self, f):
|
|
keys = sorted(f.keys())
|
|
dkeys = sorted(self._dict.keys())
|
|
self.assertEqual(keys, dkeys)
|
|
return keys
|
|
|
|
# Perform randomized operations. This doesn't make assumptions about
|
|
# what *might* fail.
|
|
def test_random(self):
|
|
import random
|
|
d = {} # mirror the database
|
|
for dummy in range(5):
|
|
f = dumbdbm.open(_fname)
|
|
for dummy in range(100):
|
|
k = random.choice('abcdefghijklm')
|
|
if random.random() < 0.2:
|
|
if k in d:
|
|
del d[k]
|
|
del f[k]
|
|
else:
|
|
v = random.choice((b'a', b'b', b'c')) * random.randrange(10000)
|
|
d[k] = v
|
|
f[k] = v
|
|
self.assertEqual(f[k], v)
|
|
f.close()
|
|
|
|
f = dumbdbm.open(_fname)
|
|
expected = sorted((k.encode("latin-1"), v) for k, v in d.items())
|
|
got = sorted(f.items())
|
|
self.assertEqual(expected, got)
|
|
f.close()
|
|
|
|
def test_context_manager(self):
|
|
with dumbdbm.open(_fname, 'c') as db:
|
|
db["dumbdbm context manager"] = "context manager"
|
|
|
|
with dumbdbm.open(_fname, 'r') as db:
|
|
self.assertEqual(list(db.keys()), [b"dumbdbm context manager"])
|
|
|
|
with self.assertRaises(dumbdbm.error):
|
|
db.keys()
|
|
|
|
def test_check_closed(self):
|
|
f = dumbdbm.open(_fname, 'c')
|
|
f.close()
|
|
|
|
for meth in (partial(operator.delitem, f),
|
|
partial(operator.setitem, f, 'b'),
|
|
partial(operator.getitem, f),
|
|
partial(operator.contains, f)):
|
|
with self.assertRaises(dumbdbm.error) as cm:
|
|
meth('test')
|
|
self.assertEqual(str(cm.exception),
|
|
"DBM object has already been closed")
|
|
|
|
for meth in (operator.methodcaller('keys'),
|
|
operator.methodcaller('iterkeys'),
|
|
operator.methodcaller('items'),
|
|
len):
|
|
with self.assertRaises(dumbdbm.error) as cm:
|
|
meth(f)
|
|
self.assertEqual(str(cm.exception),
|
|
"DBM object has already been closed")
|
|
|
|
def test_create_new(self):
|
|
with dumbdbm.open(_fname, 'n') as f:
|
|
for k in self._dict:
|
|
f[k] = self._dict[k]
|
|
|
|
with dumbdbm.open(_fname, 'n') as f:
|
|
self.assertEqual(f.keys(), [])
|
|
|
|
def test_eval(self):
|
|
with open(_fname + '.dir', 'w') as stream:
|
|
stream.write("str(print('Hacked!')), 0\n")
|
|
with support.captured_stdout() as stdout:
|
|
with self.assertRaises(ValueError):
|
|
with dumbdbm.open(_fname) as f:
|
|
pass
|
|
self.assertEqual(stdout.getvalue(), '')
|
|
|
|
def test_missing_data(self):
|
|
for value in ('r', 'w'):
|
|
_delete_files()
|
|
with self.assertRaises(FileNotFoundError):
|
|
dumbdbm.open(_fname, value)
|
|
self.assertFalse(os.path.exists(_fname + '.dir'))
|
|
self.assertFalse(os.path.exists(_fname + '.bak'))
|
|
|
|
def test_missing_index(self):
|
|
with dumbdbm.open(_fname, 'n') as f:
|
|
pass
|
|
os.unlink(_fname + '.dir')
|
|
for value in ('r', 'w'):
|
|
with self.assertRaises(FileNotFoundError):
|
|
dumbdbm.open(_fname, value)
|
|
self.assertFalse(os.path.exists(_fname + '.dir'))
|
|
self.assertFalse(os.path.exists(_fname + '.bak'))
|
|
|
|
def test_invalid_flag(self):
|
|
for flag in ('x', 'rf', None):
|
|
with self.assertRaisesRegex(ValueError,
|
|
"Flag must be one of "
|
|
"'r', 'w', 'c', or 'n'"):
|
|
dumbdbm.open(_fname, flag)
|
|
|
|
@unittest.skipUnless(hasattr(os, 'chmod'), 'test needs os.chmod()')
|
|
def test_readonly_files(self):
|
|
with support.temp_dir() as dir:
|
|
fname = os.path.join(dir, 'db')
|
|
with dumbdbm.open(fname, 'n') as f:
|
|
self.assertEqual(list(f.keys()), [])
|
|
for key in self._dict:
|
|
f[key] = self._dict[key]
|
|
os.chmod(fname + ".dir", stat.S_IRUSR)
|
|
os.chmod(fname + ".dat", stat.S_IRUSR)
|
|
os.chmod(dir, stat.S_IRUSR|stat.S_IXUSR)
|
|
with dumbdbm.open(fname, 'r') as f:
|
|
self.assertEqual(sorted(f.keys()), sorted(self._dict))
|
|
f.close() # don't write
|
|
|
|
@unittest.skipUnless(support.TESTFN_NONASCII,
|
|
'requires OS support of non-ASCII encodings')
|
|
def test_nonascii_filename(self):
|
|
filename = support.TESTFN_NONASCII
|
|
for suffix in ['.dir', '.dat', '.bak']:
|
|
self.addCleanup(support.unlink, filename + suffix)
|
|
with dumbdbm.open(filename, 'c') as db:
|
|
db[b'key'] = b'value'
|
|
self.assertTrue(os.path.exists(filename + '.dat'))
|
|
self.assertTrue(os.path.exists(filename + '.dir'))
|
|
with dumbdbm.open(filename, 'r') as db:
|
|
self.assertEqual(list(db.keys()), [b'key'])
|
|
self.assertTrue(b'key' in db)
|
|
self.assertEqual(db[b'key'], b'value')
|
|
|
|
def tearDown(self):
|
|
_delete_files()
|
|
|
|
def setUp(self):
|
|
_delete_files()
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|