def test_op():
assert compile(Operator('capture', Literal('Yo'))) == asm.Capture(
None, asm.Literal('Yo'))
assert compile(Operator('0-1', Literal('Yo'))) == asm.Multiple(
0, 1, True, asm.Literal('Yo'))
assert compile(Operator('1+', Literal('Yo'))) == asm.Multiple(
1, None, True, asm.Literal('Yo'))
def test_comment():
assert compile(Operator('comment', Either([]))) == asm.Literal('')
assert compile(Operator('comment', Literal('a'))) == asm.Literal('')
assert compile(Operator('comment',
Either([Literal('a'),
Literal('b')]))) == asm.Literal('')
assert compile(
Concat([
Macro('#sl'),
Operator('comment', Literal('yo')),
Macro('#el')
])) == asm.Concat([asm.Boundary('^', None),
asm.Boundary('$', None)])
def test_builtin_macros():
assert compile(Macro('#any')) == asm.ANY
not_linefeed = asm.CharacterClass([r'\n'], inverted=True)
assert compile(Macro('#not_linefeed')) == not_linefeed
assert compile(Macro('#nlf')) == not_linefeed
assert compile(Macro('#crlf')) == asm.Literal('\r\n')
assert compile(Concat([Macro('#sl'),
Literal('yo'),
Macro('#el')])) == asm.Concat([
asm.Boundary('^', None),
asm.Literal('yo'),
asm.Boundary('$', 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(']')
import re
from re2.parser import Concat, Either, Def, Operator, Macro, Literal, Nothing
from re2 import asm
EMPTY = asm.Literal('')
EMPTY_CONCAT = asm.Concat([])
class CompileError(Exception): pass
builtin_macros = {
'#any': asm.ANY,
'#linefeed': asm.LINEFEED,
'#carriage_return': asm.CARRIAGE_RETURN,
'#windows_newline': asm.Literal('\r\n'),
'#tab': asm.TAB,
'#digit': asm.DIGIT,
'#letter': asm.LETTER,
'#lowercase': asm.LOWERCASE,
'#uppercase': asm.UPPERCASE,
'#space': asm.SPACE,
'#word_character': asm.WORD_CHARACTER,
'#start_string': asm.START_STRING,
'#end_string': asm.END_STRING,
'#start_line': asm.START_LINE,
'#end_line': asm.END_LINE,
'#word_boundary': asm.WORD_BOUNDARY,
'#quote': asm.Literal("'"),
'#double_quote': asm.Literal('"'),
}
def test_def_scoping():
assert compile(Concat([Concat([Macro('#x')]),
Def('#x', Literal('x'))])) == asm.Literal('x')
with pytest.raises(CompileError):
compile(Concat([Concat([Def('#x', Literal('x'))]), Macro('#x')]))
def test_def():
assert compile(Concat([Def('#x', Literal('x')),
Macro('#x')])) == asm.Literal('x')
assert compile(Concat([Macro('#x'),
Def('#x', Literal('x'))])) == asm.Literal('x')
def test_nothing():
assert compile(Nothing()) == asm.Literal('')
assert compile(Concat([Nothing(), Nothing()])) == asm.Literal('')
def test_either():
assert compile(Either([Literal('abc'), Literal('def')])) == asm.Either(
[asm.Literal('abc'), asm.Literal('def')])
def test_concat():
assert compile(Concat(map(Literal, 'abc'))) == asm.Concat(
[asm.Literal('a'),
asm.Literal('b'),
asm.Literal('c')])
def test_literal():
assert compile(Literal('abc')) == asm.Literal('abc')
import re
from re2.parser import Parser, Concat, Either, Def, Operator, Macro, Range, Literal, Nothing
from re2 import asm
from re2.errors import CompileError
parser = Parser()
EMPTY = asm.Literal('')
EMPTY_CONCAT = asm.Concat([])
builtin_macros = {
'#any': asm.ANY,
'#linefeed': asm.LINEFEED,
'#carriage_return': asm.CARRIAGE_RETURN,
'#windows_newline': asm.Literal('\r\n'),
'#tab': asm.TAB,
'#digit': asm.DIGIT,
'#letter': asm.LETTER,
'#lowercase': asm.LOWERCASE,
'#uppercase': asm.UPPERCASE,
'#space': asm.SPACE,
'#token_character': asm.TOKEN_CHARACTER,
'#start_string': asm.START_STRING,
'#end_string': asm.END_STRING,
'#start_line': asm.START_LINE,
'#end_line': asm.END_LINE,
'#word_boundary': asm.WORD_BOUNDARY,
'#quote': asm.Literal("'"),
'#double_quote': asm.Literal('"'),
'#left_brace': asm.Literal('['),
