def parse(self, string):
for elem in self.padWithSequenceOperators(string):
if isinstance(elem, regOp.Operator):
self.pushOperator(elem)
elif regOp.isOperator(elem):
self.pushOperator(regOp.getOperator(elem))
else: #if character
self.pushCharacter(elem)
while len(self._operatorStack) > 0:
tempOp = self._operatorStack.pop()
if isinstance(tempOp, regOp.BracketOperator):
raise Exception(RegexParser._unbalanced_parenth_except)
else:
self._outputQueue.append(tempOp)
return self._outputQueue
def testParser(self):
self.assertRaises(Exception, p.RegexParser().parse, "(a|b)+)")
self.assertRaises(Exception, p.RegexParser().parse, "((a|b)+")
self.assertRaises(Exception, p.RegexParser().parse, "(a|)b)+")
self.assertListEqual(p.RegexParser().parse(""), [])
self.assertListEqual(p.RegexParser().parse("a"), ['a'])
self.assertListEqual(p.RegexParser().parse("ab"), ['a','b',o.getSquenceOperator()])
self.assertListEqual(p.RegexParser().parse("a|b"), ['a','b',o.getOperator('|')])
self.assertListEqual(p.RegexParser().parse("(a|b)+"), ['a','b',o.getOperator('|'), o.getOperator('+')])
self.assertListEqual(p.RegexParser().parse("((a)|b)+"), ['a','b',o.getOperator('|'), o.getOperator('+')])
self.assertListEqual(p.RegexParser().parse("(a|b)+c"), ['a','b',o.getOperator('|'), o.getOperator('+'), 'c', o.getSquenceOperator()])
self.assertListEqual(p.RegexParser().parse("(a|b)+c*"), ['a','b',o.getOperator('|'), o.getOperator('+'), 'c', o.getOperator('*'), o.getSquenceOperator()])
self.assertListEqual(p.RegexParser().parse("c(a|b)+"), ['c','a','b', o.getOperator('|'), o.getOperator('+'), o.getSquenceOperator()])
self.assertListEqual(p.RegexParser().parse("((a))"), ['a'])
self.assertListEqual(p.RegexParser().parse("a+b+"), ['a',o.getOperator('+'),'b',o.getOperator('+'),o.getSquenceOperator()])
def padWithSequenceOperators(self,string):
paddedString = []
pastChar = ''
for idx,char in enumerate(string):
if idx != 0:
if regOp.isOperator(char):
if regOp.isOpeningBracket(char) and not regOp.isOpeningBracket(pastChar):
paddedString.append(regOp.getSquenceOperator())
else: #is character
if regOp.isOperator(pastChar):
if regOp.isClosingBracket(pastChar) or regOp.isLeftAssociativeUnaryOperator(regOp.getOperator(pastChar)):
paddedString.append(regOp.getSquenceOperator())
else: #pastChar is character
paddedString.append(regOp.getSquenceOperator())
paddedString.append(char)
pastChar = char
return paddedString