Lexing sum operator and a signed integer with PLY Python

Question

How can I build my raw expression to differentiate between a sum operator and a signed integer? I'm using PLY Python.

This,unfortunately, didn't work:

t_sum=r'\+'
def t_integer(token):
    r'[-+]?\d+'

Answer 1

One way to think about this is, rather than trying to distinguish the purpose of + on lexing, is to wait until parsing (where we determine meaning). So keep:

t_PLUS  = r'\+'
t_MINUS = r'\-'
def t_INTEGER(token):
    r'\d+'
    return int(token)

And define a grammar:

number -> number PLUS number          #sum, binary operator
number -> number MINUS number
number -> PLUS number                 #signed integer, unary operator
number -> MINUS number                
number -> INTEGER

#By writing each as a parse functions like...
def p_number_ADD(p):
    "number : number PLUS number"
    p[0] = p[1]+p[2]

Note: this does allow the following behaviour +-+2 = +(-(+2)) = -2 (as seen in Python).

Answer 2

If you could detect unary operator by its surroundings (which is true only if your language is simple enough to list all the cases):

a ++ b  // binary +, unary +

So we say that ++[a-z] is unary, and +[a-z] is binary (and we list -+, *+ etc. for detecting unary as well).

Then another problem pops up: the surrundings will also attached to the token. In the example: it will be [a] and [++b], instead of [a], [+] and [+b].