def test_recognize():
d = DFA([{"a": 1}, {"b": 0}], [False, True])
assert d.recognize("ababab") == 5
assert d.recognize("c") == -1
d = DFA([{"a": 1}, {DEFAULT: 0}], [False, True])
assert d.recognize("a,a?ab") == 5
assert d.recognize("c") == -1
d = NonGreedyDFA([{"a": 1}, {"b": 0}], [False, True])
assert d.recognize("ababab") == 1
assert d.recognize("c") == -1
d = NonGreedyDFA([{"a": 1}, {DEFAULT: 0}], [False, True])
assert d.recognize("a,a?ab") == 1
assert d.recognize("c") == -1
def test_nonascii():
d = DFA([{"a": 1}, {NON_ASCII: 1}], [False, True])
input = u"aüüüü".encode("utf-8")
assert d.recognize(input) == len(input)
assert d.recognize("c") == -1
assert d.recognize("ü") == -1
d = NonGreedyDFA([{NON_ASCII: 0, "b": 1}, {"b": 0}], [False, True])
input = u"üübbbb".encode("utf-8")
assert d.recognize(input) == len(u"üüb".encode("utf-8"))
assert d.recognize("c") == -1
pytest.raises(ValueError, DFA, [{"\x81": 2}], [True])
def makePyEndDFAMap():
states = []
single = chain(
states, any(states, notGroupStr(states, "'\\")),
any(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT),
any(states, notGroupStr(states, "'\\")))),
newArcPair(states, "'"))
singleDFA = DFA(*nfaToDfa(states, *single))
states = []
double = chain(
states, any(states, notGroupStr(states, '"\\')),
any(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT),
any(states, notGroupStr(states, '"\\')))),
newArcPair(states, '"'))
doubleDFA = DFA(*nfaToDfa(states, *double))
states = []
single3 = chain(
states, any(states, notGroupStr(states, "'\\")),
any(
states,
chain(
states,
group(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT)),
chain(states, newArcPair(states, "'"),
notChainStr(states, "''"))),
any(states, notGroupStr(states, "'\\")))),
chainStr(states, "'''"))
single3DFA = NonGreedyDFA(*nfaToDfa(states, *single3))
states = []
double3 = chain(
states, any(states, notGroupStr(states, '"\\')),
any(
states,
chain(
states,
group(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT)),
chain(states, newArcPair(states, '"'),
notChainStr(states, '""'))),
any(states, notGroupStr(states, '"\\')))),
chainStr(states, '"""'))
double3DFA = NonGreedyDFA(*nfaToDfa(states, *double3))
map = {
"'": singleDFA,
'"': doubleDFA,
"r": None,
"R": None,
"u": None,
"U": None,
"b": None,
"B": None
}
for uniPrefix in (
"",
"u",
"U",
"b",
"B",
):
for rawPrefix in ("", "r", "R"):
prefix = uniPrefix + rawPrefix
map[prefix + "'''"] = single3DFA
map[prefix + '"""'] = double3DFA
return map
def makePyEndDFAMap():
states = []
single = chain(
states, any(states, notGroupStr(states, "'\\")),
any(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT),
any(states, notGroupStr(states, "'\\")))),
newArcPair(states, "'"))
states, accepts = nfaToDfa(states, *single)
singleDFA = DFA(states, accepts)
states_singleDFA = states
states = []
double = chain(
states, any(states, notGroupStr(states, '"\\')),
any(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT),
any(states, notGroupStr(states, '"\\')))),
newArcPair(states, '"'))
states, accepts = nfaToDfa(states, *double)
doubleDFA = DFA(states, accepts)
states_doubleDFA = states
states = []
single3 = chain(
states, any(states, notGroupStr(states, "'\\")),
any(
states,
chain(
states,
group(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT)),
chain(states, newArcPair(states, "'"),
notChainStr(states, "''"))),
any(states, notGroupStr(states, "'\\")))),
chainStr(states, "'''"))
states, accepts = nfaToDfa(states, *single3)
single3DFA = NonGreedyDFA(states, accepts)
states_single3DFA = states
states = []
double3 = chain(
states, any(states, notGroupStr(states, '"\\')),
any(
states,
chain(
states,
group(
states,
chain(states, newArcPair(states, "\\"),
newArcPair(states, DEFAULT)),
chain(states, newArcPair(states, '"'),
notChainStr(states, '""'))),
any(states, notGroupStr(states, '"\\')))),
chainStr(states, '"""'))
states, accepts = nfaToDfa(states, *double3)
double3DFA = NonGreedyDFA(states, accepts)
states_double3DFA = states
return {
"'": (singleDFA, states_singleDFA),
'"': (doubleDFA, states_doubleDFA),
"'''": (single3DFA, states_single3DFA),
'"""': (double3DFA, states_double3DFA)
}