DZone Snippets is a public source code repository. Easily build up your personal collection of code snippets, categorize them with tags / keywords, and share them with the world
Djikstra's Shunting Yard Algorithm
Python implementation of Djikstra's shunting yard algorithm(aka converting infix to RPN). Read in-code documentation for details. I've coded in some atypical operators(&&, ||), feel free to throw them out if you feel you need to do so. It's quite easy.
So far it works ok(tested on 3+4*2/(1-5)^2^3 and on x(y(90),90)). If I find some mistake, I'll correct it. Fine :)
Example:
a = Djikstrer() a.run(['2','+','2']);
a.output now contains ['2','2','+']
'''Hey, it's public domain :)'''
class MismatchedParenthesesException(Exception):
'''Raised when Djikstrer detects that there is something wrong with separators or parentheses'''
pass
class Djikstrer:
'''This class performs conversion from infix expressions to Reverse Polish Notation using Djikstra's shunting yard
algorithm. Function run() accepts a list of tokens(yes, the expression needs to be tokenized first) and after run() finishes
there is a list of tokens in class' attribute 'output' in RPN order. New operators can be added below. Smaller precedence
means the operator will get executed later, ie. multiplication has greater precedence than addition. Obviously, precedence is set in
op_prec dictionary. assoc, lassor and rassoc dictionaries specify whether the operators are associative, left-associative or
right-associative. Note: one class can perform many conversions, as 'run' method cleans the stack and output before analysis.
Another note: watch out for unary minus(ie. -5) passed as two tokens, as it will certainly screw the stuff up.
Same with unary plus(but who would type something like that?'''
operators = ('&&','||','+','-','*','/','^')
op_prec = {'+':3,'-':3,'*':4,'/':4,'^':5,'&&':6,'||':6}
assoc = {'+':True,'-':False,'*':True,'/':False,'^':False,'||':True,'&&':True}
lassoc = {'+':True,'-':True,'*':True,'/':True,'^':False,'||':True,'&&':True}
rassoc = {'+':True,'-':False,'*':True,'/':False,'^':True,'||':True,'&&':True}
def _isparenthesis(self, token):
return (token in ('(',')'))
def _isfunction(self, token):
return not (self._isoperator(token) or self._isparenthesis(token) or self._isnumber(token) or self._isseparator(token))
def _isseparator(self, token):
return token== ','
def _isoperator(self, token):
return token in self.operators
def _isnumber(self,x):
try:
float(x)
except:
return False
return True
def areThereOperatorsOnStack(self):
for x in self.stack:
if self._isoperator(x):
return True
return False
def sTo(self):
self.output.append(self.stack.pop())
def spush(self, token):
self.stack.append(token)
def opush(self, token):
self.output.append(token)
def run(self, tokenlist):
self.stack = []
self.output = []
try:
for token in tokenlist:
self.token(token)
while self.areThereOperatorsOnStack():
if self._isparenthesis(self.stack[-1]):
raise MismatchedParenthesesException, 'Parentheses'
self.sTo()
except:
raise MismatchedParenthesesException, 'Parentheses or separators'
def token(self, token):
if self._isnumber(token):
self.opush(token)
return
if self._isfunction(token):
self.spush(token)
return
if self._isseparator(token):
while not self.stack[-1]=='(':
if len(self.stack)==0:
raise MismatchedParenthesesException, 'Parentheses or separators'
self.sTo()
if self.stack[-1]=='(':
break
return
if self._isoperator(token):
if len(self.stack)>0:
while self._isoperator(self.stack[-1]) and (((self.assoc[token] or self.lassoc[token]) and (self.op_prec[token]<=self.op_prec[self.stack[-1]])) or (self.rassoc[token] and (self.op_prec[token]<self.op_prec[self.stack[-1]]))):
self.sTo()
self.spush(token)
return
if token=='(':
self.spush(token)
return
if token==')':
while not self.stack[-1]=='(':
if len(self.stack)==0:
raise MismatchedParenthesesException('Parentheses')
self.sTo()
if self.stack[-1]=='(':
break
self.stack.pop()
if len(self.stack)==0:
return
if self._isfunction(self.stack[-1]):
self.sTo()





