def test_subtraction(self):
c = unicode5.CharClass("abc")
c.subtract_char("b")
result = self.class_test(c)
self.assertTrue(result == "ac", "subtract_char: %s" % result)
c.subtract_range("b", "d")
self.assertTrue(self.class_test(c) == "a", "subtract_range")
tests = [[[], [], ""], [[['a', 'b']], [['c', 'd']], "ab"],
[[['a', 'b']], [['b', 'c']], "a"],
[[['a', 'c']], [['b', 'd']], "a"],
[[['a', 'd']], [['b', 'd']], "a"],
[[['a', 'd']], [['b', 'c']], "ad"],
[[['a', 'c']], [['a', 'd']], ""],
[[['a', 'c']], [['a', 'c']], ""],
[[['a', 'd']], [['a', 'b']], "cd"],
[[['b', 'c']], [['a', 'd']], ""],
[[['b', 'c']], [['a', 'c']], ""],
[[['b', 'd']], [['a', 'c']], "d"],
[[['a', 'z']], [['f', 'h'], ['s', 'u']],
"abcdeijklmnopqrvwxyz"],
[[['a', 'e'], ['i', 'r'], ['v', 'z']], [['m', 'x']],
"abcdeijklyz"]]
for test in tests:
c1 = unicode5.CharClass(*test[0])
c2 = unicode5.CharClass(*test[1])
c3 = unicode5.CharClass(c1)
c3.subtract_class(c2)
result = self.class_test(c3)
self.assertTrue(
result == test[2],
"Subtract: %s - %s, found %s" % (repr(c1), repr(c2), repr(c3)))
def test_add(self):
c = unicode5.CharClass(u"ac")
c.add_char(u"b")
self.assertTrue(self.class_test(c) == "abc", "add_char")
c.add_range(u"b", u"e")
self.assertTrue(self.class_test(c) == "abcde", "add_range")
c.add_class(unicode5.CharClass(["m", "s"]))
self.assertTrue(self.class_test(c) == "abcdemnopqrs", "add_class")
def test_representation(self):
repr_tests = [
[[], "CharClass()", ""],
[[['a', 'z']],
"CharClass((u'a',u'z'))" if py2 else "CharClass(('a','z'))",
"a-z"],
[[['a', 'd'], ['f', 'k']], "CharClass((u'a',u'd'), (u'f',u'k'))"
if py2 else "CharClass(('a','d'), ('f','k'))", "a-df-k"],
[[['-', '-']], "CharClass(u'-')" if py2 else "CharClass('-')",
"\\-"],
[[['[', ']']],
"CharClass((u'[',u']'))" if py2 else "CharClass(('[',']'))",
"[-\\]"],
[[['\\', '\\']],
"CharClass(u'\\\\')" if py2 else "CharClass('\\\\')", "\\\\"],
]
for test in repr_tests:
c = unicode5.CharClass(*test[0])
self.assertTrue(
repr(c) == test[1],
"CharClass repr test: expected %s, found %s" %
(test[1], repr(c)))
result = c.format_re()
self.assertTrue(
result == test[2],
"CharClass Re test: expected %s, found %s" % (test[2], result))
def test_negate_char_class(self):
"""Check the Negation function"""
min_char = character(0)
max_char = character(maxunicode)
char_class_tests = [[[], [[min_char, max_char]]],
[[['b', 'c']], [[min_char, 'a'], ['d', max_char]]],
[[['b', 'c'], ['e', 'f']],
[[min_char, 'a'], ['d', 'd'], ['g', max_char]]]]
for test in char_class_tests:
c1 = unicode5.CharClass(*test[0])
c2 = unicode5.CharClass(c1)
c2.negate()
c3 = unicode5.CharClass(*test[1])
self.assertTrue(
c2 == c3, "%s negated to %s, expected %s" %
(repr(c1), repr(c2), repr(c3)))
c2.negate()
self.assertTrue(c2 == c1,
"%s double negation got %s" % (repr(c1), repr(c2)))
def test_complex_constructors(self):
init_tests = [
[[], ""],
[[['a', 'z']], "abcdefghijklmnopqrstuvwxyz"],
[[['a', 'd'], ['f', 'k']], "abcdfghijk"],
[[['b', 'b']], "b"],
[[['a', 'b'], ['c', 'd'], ['e', 'f'], ['g', 'h'], ['i', 'j'],
['k', 'k']], "abcdefghijk"],
[[['a', 'b'], ['d', 'f'], ['h', 'h']], "abdefh"],
[[['h', 'h'], ['d', 'f'], ['a', 'b']], "abdefh"],
]
for test in init_tests:
c = unicode5.CharClass(*test[0])
result = self.class_test(c)
self.assertTrue(
result == test[1],
"CharClass test: expected %s, found %s" % (test[1], result))
def test_constructor(self):
c = unicode5.CharClass()
if MAX_CHAR < 0x10FFFF:
logging.warn("unicode5 tests truncated to character(0x%X) by "
"narrow python build" % MAX_CHAR)
for code in range3(MAX_CHAR + 1):
self.assertFalse(c.test(character(code)))
c = unicode5.CharClass('a')
self.assertTrue(self.class_test(c) == 'a')
c = unicode5.CharClass(('a', 'z'))
self.assertTrue(self.class_test(c) == 'abcdefghijklmnopqrstuvwxyz')
c = unicode5.CharClass('abcxyz')
self.assertTrue(
len(c.ranges) == 2, "No range optimization: %s" % repr(c.ranges))
self.assertTrue(self.class_test(c) == 'abcxyz')
cc = unicode5.CharClass(c)
self.assertTrue(self.class_test(cc) == 'abcxyz')
c = unicode5.CharClass(('a', 'c'), ('e', 'g'), 'd')
self.assertTrue(
len(c.ranges) == 1,
"Missing range optimization: %s" % repr(c.ranges))