def test_invert():
assert compile(Operator('not',
Literal('a'))) == asm.CharacterClass(['a'],
inverted=True)
assert compile(Operator('not',
Either([Literal('a'), Literal('b')
]))) == asm.CharacterClass(['a', 'b'],
inverted=True)
assert compile(Operator('not',
Either([Literal('a'), Macro('#d')
]))) == asm.CharacterClass(['a', r'\d'],
inverted=True)
assert compile(Operator('not', Either(
[Literal('a'),
Macro('#l')]))) == asm.CharacterClass(['a', ['a', 'z'], ['A', 'Z']],
inverted=True)
def invert_operator(expr):
if isinstance(expr, asm.Literal) and len(expr.string) == 1:
return asm.CharacterClass([expr.string], True)
try:
return expr.invert()
except AttributeError:
raise CompileError('Expression %s cannot be inverted' % expr.to_regex())
def test_range_macros():
assert compile(Range('a', 'f')) == asm.CharacterClass([('a', 'f')], False)
assert compile(Range('B', 'Z')) == asm.CharacterClass([('B', 'Z')], False)
assert compile(Range('2', '6')) == asm.CharacterClass([('2', '6')], False)
assert compile(Either([Range('a', 'f'), Macro('#digit')
])) == asm.CharacterClass([('a', 'f'), r'\d'],
False)
with pytest.raises(CompileError):
compile(Range('a', '5'))
with pytest.raises(CompileError):
compile(Range('a', 'F'))
with pytest.raises(CompileError):
compile(Range('c', 'a'))
with pytest.raises(CompileError):
compile(Range('a', 'a'))
with pytest.raises(AssertionError):
compile(Range('!', ','))
def test_character_class():
assert compile(Either([Literal('a'),
Literal('b')])) == asm.CharacterClass(['a', 'b'],
False)
assert compile(Operator('not',
Either([Literal('a'), Literal('b')
]))) == asm.CharacterClass(['a', 'b'],
inverted=True)
with pytest.raises(CompileError):
compile(Operator('not', Either([Literal('a'), Literal('bc')])))
assert compile(Either([Literal('a'),
Literal('b'),
Literal('0')
])) == asm.CharacterClass(['a', 'b', '0'], False)
assert compile(Either([Literal('a'),
Macro('#d')])) == asm.CharacterClass(['a', r'\d'],
False)
def compile_range(range, _):
if character_category(range.start) != character_category(range.end):
raise CompileError(
"Range start and end not of the same category: '%s' is a %s but '%s' is a %s"
% (range.start, character_category(
range.start), range.end, character_category(range.end)))
if range.start >= range.end:
raise CompileError("Range start not before range end: '%s' >= '%s'" %
(range.start, range.end))
return asm.CharacterClass([(range.start, range.end)], False)
def compile_either(e, macros):
compiled = [compile_ast(s, macros) for s in e.items]
if all(is_single_char(c) for c in compiled):
characters = []
for c in compiled:
if isinstance(c, asm.Literal) and len(c.string) == 1:
characters.append(c.string)
elif isinstance(c, asm.CharacterClass):
characters += c.characters
return asm.CharacterClass(characters, False)
return asm.Either(compiled)
def test_builtin_macros():
assert compile(Macro('#any')) == asm.ANY
not_linefeed = asm.CharacterClass([r'\n'], inverted=True)
assert compile(Macro('#not_linefeed')) == not_linefeed
assert compile(Macro('#nlf')) == not_linefeed
assert compile(Macro('#crlf')) == asm.Literal('\r\n')
assert compile(Concat([Macro('#sl'),
Literal('yo'),
Macro('#el')])) == asm.Concat([
asm.Boundary('^', None),
asm.Literal('yo'),
asm.Boundary('$', None)
])
def test_builtin_macros():
assert compile(Macro('#any')) == asm.ANY
not_linefeed = asm.CharacterClass([r'\n'], inverted=True)
assert compile(Macro('#not_linefeed')) == not_linefeed
assert compile(Macro('#not_linefeed')) == not_linefeed
assert compile(Macro('#windows_newline')) == asm.Literal('\r\n')
assert compile(Concat([Macro('#sl'),
Literal('yo'),
Macro('#el')])) == asm.Concat([
asm.Boundary('^', None),
asm.Literal('yo'),
asm.Boundary('$', None)
])
assert compile(Macro('#quote')) == asm.Literal("'")
assert compile(Macro('#double_quote')) == asm.Literal('"')
assert compile(Macro('#left_brace')) == asm.Literal("[")
assert compile(Macro('#right_brace')) == asm.Literal(']')