def test_range_missing_end(self):
with self.assertRaises(ParserError) as context:
parse(u"[a-]")
exception = context.exception
self.assertEqual(exception.reason,
u"expected character, found instruction: ]")
self.assertEqual(exception.annotation, (u"[a-]\n" u" ^"))
def test_zero_or_more_missing_repeatable(self):
with self.assertRaises(ParserError) as context:
parse(u"*")
exception = context.exception
self.assertEqual(exception.reason,
u"* is not preceded by a repeatable expression")
self.assertEqual(exception.annotation, (u"*\n" u"^"))
def test_either_missing_end(self):
with self.assertRaises(ParserError) as context:
parse(u"[ab")
exception = context.exception
self.assertEqual(
exception.reason,
u"unexpected end of string, expected ] corresponding to [")
self.assertEqual(exception.annotation, (u"[ab\n" u"^--^"))
def test_group_missing_end(self):
with self.assertRaises(ParserError) as context:
parse(u"(a")
exception = context.exception
self.assertEqual(
exception.reason,
u"unexpected end of string, expected ) corresponding to (")
self.assertEqual(exception.annotation, (u"(a\n" u"^-^"))
def test_range_missing_start(self):
with self.assertRaises(ParserError) as context:
parse(u"[-c]")
exception = context.exception
self.assertEqual(exception.reason, u"range is missing start")
self.assertEqual(
exception.annotation,
(
u"[-c]\n"
u"^"
)
)
def test_zero_or_more_missing_repeatable(self):
with self.assertRaises(ParserError) as context:
parse(u"*")
exception = context.exception
self.assertEqual(
exception.reason,
u"* is not preceded by a repeatable expression"
)
self.assertEqual(exception.annotation, (
u"*\n"
u"^"
))
def test_group_missing_end(self):
with self.assertRaises(ParserError) as context:
parse(u"(a")
exception = context.exception
self.assertEqual(
exception.reason,
u"unexpected end of string, expected ) corresponding to ("
)
self.assertEqual(
exception.annotation,
(
u"(a\n"
u"^-^"
)
)
def test_either_missing_end(self):
with self.assertRaises(ParserError) as context:
parse(u"[ab")
exception = context.exception
self.assertEqual(
exception.reason,
u"unexpected end of string, expected ] corresponding to ["
)
self.assertEqual(
exception.annotation,
(
u"[ab\n"
u"^--^"
)
)
def test_range_missing_end(self):
with self.assertRaises(ParserError) as context:
parse(u"[a-]")
exception = context.exception
self.assertEqual(
exception.reason,
u"expected character, found instruction: ]"
)
self.assertEqual(
exception.annotation,
(
u"[a-]\n"
u" ^"
)
)
def test_group_missing_begin(self):
with self.assertRaises(ParserError) as context:
parse(u"a)")
exception = context.exception
self.assertEqual(
exception.reason,
u"found unmatched )"
)
self.assertEqual(
exception.annotation,
(
u"a)\n"
u" ^"
)
)
def test_range(self):
self.assertEqual(
parse(u"[a-c]"),
Either(
frozenset([
Range(Character(u"a"), Character(u"c"), DEFAULT_ALPHABET)
])))
def Regex(regex, optimize_flag=False):
ast = parse(tokens(regex))
if len(ast) == 0:
raise ValueError("Nothing to parse")
automata = generate(ast)
if optimize_flag:
return optimize(automata)
else:
return automata
def test_range(self):
self.assertEqual(
parse(u"[a-c]"),
Either(frozenset([Range(
Character(u"a"),
Character(u"c"),
DEFAULT_ALPHABET
)]))
)
def test_epsilon(self):
self.assertEqual(parse(u""), Epsilon())
def __init__(self, regex):
self.regex = regex
self.ast = parse(regex)
def __init__(self, regex):
self.regex = regex
self.ast = parse(regex)
def test_epsilon(self):
self.assertEqual(parse(u""), Epsilon())
def test_one_or_more(self):
self.assertEqual(
parse(u"a+"),
Concatenation(Character(u"a"), Repetition(Character(u"a"))))
def test_range_missing_start(self):
with self.assertRaises(ParserError) as context:
parse(u"[-c]")
exception = context.exception
self.assertEqual(exception.reason, u"range is missing start")
self.assertEqual(exception.annotation, (u"[-c]\n" u"^"))
def test_union(self):
self.assertEqual(
parse(u"a|b"),
Union(Character(u"a"), Character(u"b"))
)
def test_either(self):
self.assertEqual(
parse(u"[ab]"),
Either(frozenset(map(Character, u"ab")))
)
def test_concatenation(self):
self.assertEqual(parse(u"ab"),
Concatenation(Character(u"a"), Character(u"b")))
def test_union(self):
self.assertEqual(parse(u"a|b"), Union(Character(u"a"),
Character(u"b")))
def test_group(self):
self.assertEqual(
parse(u"(a)"),
Group(Character(u"a"))
)
def test_zero_or_more(self):
self.assertEqual(parse(u"a*"), Repetition(Character(u"a")))
def test_either_missing_begin(self):
with self.assertRaises(ParserError) as context:
parse(u"ab]")
exception = context.exception
self.assertEqual(exception.reason, u"found unmatched ]")
self.assertEqual(exception.annotation, (u"ab]\n" u" ^"))
def test_either(self):
self.assertEqual(parse(u"[ab]"),
Either(frozenset(map(Character, u"ab"))))
def test_character(self):
self.assertEqual(parse(u"a"), Character(u"a"))
def test_concatenation(self):
self.assertEqual(
parse(u"ab"),
Concatenation(Character(u"a"), Character(u"b"))
)
def test_character(self):
self.assertEqual(parse(u"a"), Character(u"a"))
def test_zero_or_more(self):
self.assertEqual(
parse(u"a*"),
Repetition(Character(u"a"))
)
def test_neither(self):
self.assertEqual(
parse(u"[^ab]"),
Neither(frozenset(map(Character, u"ab")), DEFAULT_ALPHABET)
)
def test_one_or_more(self):
self.assertEqual(
parse(u"a+"),
Concatenation(Character(u"a"), Repetition(Character(u"a")))
)
def test_group(self):
self.assertEqual(parse(u"(a)"), Group(Character(u"a")))
def __init__(self, definitions):
self.definitions = []
for regex, token_cls in definitions:
self.definitions.append((parse(regex).compile(), token_cls))
def test_neither(self):
self.assertEqual(
parse(u"[^ab]"),
Neither(frozenset(map(Character, u"ab")), DEFAULT_ALPHABET))
