def test_equals(self):
cps = CodepointSet('0000..00FF')
self.assertEqual(cps, CodepointSet('0000..00FF'))
self.assertNotEqual(cps, CodepointSet('0000..00FE'))
# Non-CodepointSet always not-equal.
self.assertFalse(cps == 'what?')
def test_parse(self):
cps = CodepointSet('A\nBB\n')
self.assertEqual(repr(cps), r"CodepointSet('000A\n00BB')")
cps = CodepointSet('AAA\nBBB..CCC\n')
self.assertEqual(repr(cps), r"CodepointSet('0AAA\n0BBB..0CCC')")
cps = CodepointSet('\n \n # comment \n \n')
self.assertEqual(repr(cps), "CodepointSet('')")
def test_malformed_range(self):
with self.assertRaises(ValueError):
CodepointSet('0002..0000\n0001')
with self.assertRaises(ValueError):
CodepointSet('0000..0001\n0000..0001\n0002')
with self.assertRaises(ValueError):
CodepointSet('0000\n0002\n0002..0004')
with self.assertRaises(ValueError):
CodepointSet('110000')
with self.assertRaises(ValueError):
CodepointSet('0000\n000G')
def test_len(self):
cps = CodepointSet('0000\n')
self.assertEqual(len(cps), 1)
cps = CodepointSet('0000..0001\n')
self.assertEqual(len(cps), 2)
cps = CodepointSet('0000\n0001\n0002')
self.assertEqual(len(cps), 3)
cps = CodepointSet('0000\n0002')
self.assertEqual(len(cps), 2)
cps = CodepointSet('10000..10FFFF')
self.assertEqual(len(cps), 0x10FFFF - 0x10000 + 1)
def test_even_odd(self):
data = '\n'.join("%04X" % cp for cp in range(0, 10000, 2))
cps = CodepointSet(data)
for cp in range(10000):
if cp in cps:
self.assertTrue((cp % 2) == 0)
else:
self.assertFalse((cp % 2) == 0)
def test_contains(self):
cps = CodepointSet('0000\n')
actual = [cp in cps for cp in range(-1, 4)]
self.assertEqual(actual, [False, True, False, False, False])
self.assertFalse(0x010FFFF in cps)
cps = CodepointSet('0000..0001\n')
actual = [cp in cps for cp in range(-1, 4)]
self.assertEqual(actual, [False, True, True, False, False])
cps = CodepointSet('0000\n0001\n0002')
actual = [cp in cps for cp in range(-1, 4)]
self.assertEqual(actual, [False, True, True, True, False])
cps = CodepointSet('0000\n0002')
actual = [cp in cps for cp in range(-1, 4)]
self.assertEqual(actual, [False, True, False, True, False])
cps = CodepointSet('10000..10FFFF')
self.assertTrue(0x10FFFF in cps)
self.assertFalse(0x110000 in cps)
def test_repr(self):
cps = CodepointSet('')
self.assertEqual(repr(cps), "CodepointSet('')")
cps = CodepointSet('0000')
self.assertEqual(repr(cps), "CodepointSet('0000')")
cps = CodepointSet('0000..00FF')
self.assertEqual(repr(cps), "CodepointSet('0000..00FF')")
cps = CodepointSet('0001..FFFF\n100000..10FFFF')
self.assertEqual(repr(cps),
r"CodepointSet('0001..FFFF\n100000..10FFFF')")
cps = CodepointSet('FFFF..1FFFF')
self.assertEqual(repr(cps), "CodepointSet('FFFF..1FFFF')")
cps = CodepointSet('10000..1FFFF')
self.assertEqual(repr(cps), "CodepointSet('10000..1FFFF')")
cps = CodepointSet('FFFE\n10000..1FFFF')
self.assertEqual(repr(cps), r"CodepointSet('FFFE\n10000..1FFFF')")
# http://www.unicode.org/Public/10.0.0/ucd/DerivedCoreProperties.txt
# Derived Property: Default_Ignorable_Code_Point
_DEFAULT_IGNORABLE = CodepointSet('''
00AD
034F
061C
115F..1160
17B4..17B5
180B..180D
180E
200B..200F
202A..202E
2060..2064
2065
2066..206F
3164
FE00..FE0F
FEFF
FFA0
FFF0..FFF8
1BCA0..1BCA3
1D173..1D17A
E0000
E0001
E0002..E001F
E0020..E007F
E0080..E00FF
E0100..E01EF
E01F0..E0FFF
''')
assert len(_DEFAULT_IGNORABLE) == 4173
def test_coalesce(self):
cps = CodepointSet('0000\n0001\n0002')
self.assertEqual(cps, CodepointSet('0000..0002'))
cps = CodepointSet('0000\n0002\n0003..0004')
self.assertEqual(cps, CodepointSet('0000\n0002..0004'))