def test_add_variant_sequence(self):
c = CharSequence([0x002A, 0x002B])
c.add_variant([0x0030])
self.assertEqual(len(c._variants), 1)
self.assertIn((0x0030, ), c._variants)
self.assertEqual(len(c._variants[(0x0030, )]), 1)
self.assertEqual(c._variants[(0x0030, )][0].cp, (0x0030, ))
self.assertRaises(VariantAlreadyExists, c.add_variant, [0x0030])
c.add_variant([0x0030], when='w1')
self.assertRaises(VariantAlreadyExists,
c.add_variant, [0x0030],
when='w1')
c.add_variant([0x0030], not_when='w1')
self.assertRaises(VariantAlreadyExists,
c.add_variant, [0x0030],
not_when='w1')
c.add_variant([0x0030], when='w1', not_when='nw-1')
self.assertRaises(VariantAlreadyExists,
c.add_variant, [0x0030],
when='w1',
not_when='nw-1')
self.assertEqual(len(c._variants[(0x0030, )]), 4)
def test_all_repertoire(self):
self.cd.add_char([0x0010])
self.cd.add_range(0x0001, 0x0005)
self.cd.add_char([0x0000])
self.cd.add_char([0x0011, 0x0012])
self.cd.add_char([0x0013])
# Full output
expected_output = {
Char(0x0000),
RangeChar(0x0001, 0x0001, 0x0005),
RangeChar(0x0002, 0x0001, 0x0005),
RangeChar(0x0003, 0x0001, 0x0005),
RangeChar(0x0004, 0x0001, 0x0005),
RangeChar(0x0005, 0x0001, 0x0005),
Char(0x0010),
CharSequence([0x0011, 0x0012]),
Char(0x0013),
}
self.assertEqual(set(self.cd.all_repertoire()), expected_output)
# Exclude ranges
expected_output = {
Char(0x0000),
Char(0x0010),
CharSequence([0x0011, 0x0012]),
Char(0x0013),
}
self.assertEqual(set(self.cd.all_repertoire(include_ranges=False)),
expected_output)
# Exclude sequences
expected_output = {
Char(0x0000),
RangeChar(0x0001, 0x0001, 0x0005),
RangeChar(0x0002, 0x0001, 0x0005),
RangeChar(0x0003, 0x0001, 0x0005),
RangeChar(0x0004, 0x0001, 0x0005),
RangeChar(0x0005, 0x0001, 0x0005),
Char(0x0010),
Char(0x0013),
}
self.assertEqual(set(self.cd.all_repertoire(include_sequences=False)),
expected_output)
# Exclude ranges and sequences
expected_output = {
Char(0x0000),
Char(0x0010),
Char(0x0013),
}
self.assertEqual(
set(
self.cd.all_repertoire(include_ranges=False,
include_sequences=False)),
expected_output)
def test_as_index(self):
self.assertEqual(Char(0x002A).as_index(), 0x002A)
self.assertEqual(CharSequence((0x002A, 0x002B)).as_index(), 0x002A)
self.assertEqual(RangeChar(0x002A, 0x002A, 0x02C).as_index(), 0x002A)
self.assertEqual(
CharBase.from_cp_or_sequence(0x002A).as_index(), 0x002A)
self.assertEqual(
CharBase.from_cp_or_sequence([0x002A]).as_index(), 0x002A)
self.assertEqual(
CharBase.from_cp_or_sequence([0x002A, 0x002B]).as_index(), 0x002A)
def test_lgr_sequence(self):
self.lgr.add_cp([0x0061, 0x0062, 0x0063])
self.lgr.add_cp([0x0061, 0x0062])
__, result = compute_stats(self.lgr, {})
stats = self.STATS.copy()
stats['codepoint_number'] = 2
stats['sequence_number'] = 2
stats['largest_sequence'] = CharSequence(cp_or_sequence=(0x0061, 0x0062, 0x0063))
stats['largest_sequence_len'] = 3
self.assertDictEqual(result, {'description': 'Generate stats',
'stats': stats})
def test_iter(self):
self.cd.add_char([0x0010])
self.cd.add_range(0x0001, 0x000A)
self.cd.add_char([0x0000])
self.cd.add_char([0x0011, 0x0012])
self.cd.add_char([0x0013])
expected_output = [
Char(0x0000),
RangeChar(0x0001, 0x0001, 0x000A),
Char(0x0010),
CharSequence([0x0011, 0x0012]),
Char(0x0013),
]
self.assertEqual(list(self.cd), expected_output)
def test_hash_sequence(self):
c1 = CharSequence((0x002A, 0x002B))
c2 = CharSequence((0x002A, 0x002B))
c3 = CharSequence((0x002A, 0x002C))
c4 = CharSequence((0x002A, 0x002B, 0x002C))
c5 = CharSequence((0x002A, 0x002C, 0x002B))
self.assertEqual(c1.__hash__(), c2.__hash__())
self.assertNotEqual(c1.__hash__(), c3.__hash__())
self.assertNotEqual(c1.__hash__(), c4.__hash__())
self.assertNotEqual(c1.__hash__(), c5.__hash__())
self.assertNotEqual(c3.__hash__(), c4.__hash__())
self.assertNotEqual(c3.__hash__(), c5.__hash__())
self.assertNotEqual(c4.__hash__(), c5.__hash__())