Z3 with string expressions

问题:

I'm trying to use Z3 to determine if an expression is satisfiable. I could easily do this by defining the context then the int_const variables and the formula. To evaluate an expression programmatically you would have to write everything in code. Let's say the logical expression is given in the form of a string, what then? For example,

"x == y && !x == z"

would be expressed in the C API as:

context c;
expr x = c.int_const("x")
//Same for other variables
...
formula = (x == y) && (!x == z);
solver s(c);
s.add(formula);
//s.check() ...etc etc

Okay, I can write the code for this particular formula, but how could I do that programmatically given a string. I'm open to anything you can think of.

Thank you :)

回答1:

I see the following options:

1) you can implement your own parser, and invoke the Z3 API functions. Pro: you can use your "favorite" language for writing formulas. Con: it is "busy" work.

2) you can use the API Z3_parse_smtlib2_string. Con: your formulas must be in SMT 2.0 format. For example, you would have to write (and (= x y) (not (= x y))) instead of (x == y) && !(x == z).

3) You can use the Z3 Python API, and parse strings using the eval function in Python. Here is an example:

from z3 import *
# Creating x, y 
x = Int('x')
y = Int('y')

# Creating the formula using Python
f = And(x == y, Not(x == y))
print f

# Using eval to parse the string.
s = "And(x == y, Not(x == y))"
f2 = eval(s)
print f2

BTW, this script does not work at rise4fun http://rise4fun.com/z3py because the function eval is not allowed there, but you can use the script above in your local Z3 installation.