Modeling a language in python then using its ast module to parse and interpet the model is very powerful. You completely shortcut the usual tasks of writing a grammar, tokenizer, and parser, and get an instant abstract syntax tree (AST).

To demonstrate, here is the toy language of Chapter 2 of the book Introduction to Static Analysis (ITSA), here Example 2.1:

init([0,1] x [0,1]);
translation(1,0);
iter {
  {
    translation(1,0)
  } or {
    rotation(0, 0, 90)
  }
}

Is modeled in example_2_1.py:

init(0, 1, 0, 1)
translation(1, 0)
while unknown:
    if unknown:
        translation(1, 0)
        translation(2, 3)
    else:
        rotation(0, 0, 90)

And can be interpreted with interpret.py:

$ ./interpret.py ./example_2_1.py
-------- EXECUTED AST --------
init([0, 1, 0, 1])
translation([1, 0])
iteration taken? True
or taken? False
rotation([0, 0, 90])
iteration taken? True
or taken? True
translation([1, 0])
translation([2, 3])
iteration taken? True
or taken? False
rotation([0, 0, 90])
iteration taken? True
or taken? True
translation([1, 0])
translation([2, 3])
iteration taken? True
or taken? False
rotation([0, 0, 90])
iteration taken? False

The AST can be printed with parse.py:

[   'block',
    ['init', 0, 1, 0, 1],
    ['translation', 1, 0],
    [   'iter',
        [   'block',
            [   'or',
                ['block', ['translation', 1, 0], ['translation', 2, 3]],
                ['block', ['rotation', 0, 0, 90]]]]]]

To get the control flow graph, I have tested:

• convert AST to C using py2c.py and compiled with gcc
• convert AST to assembly using py2asm.py and assembled with nasm

Then opened with a binary analysis tool like Binary Ninja. See Makefile.

graph TD
	b0["init(0, 1, 0, 1)<br>translation(1, 0)"]
	b1["(end)"]
	b2["or"]
	b3["rotation(0, 0, 90)"]
	b4["translation(1, 0)<br>translation(2, 3)"]
	b5["iter"]
	b6["(machine code jmp)"]
	b0 --> b5
	b2 --> b3
	b2 --> b4
	b3 --> b6
	b4 --> b6
	b5 --> b1
	b5 --> b2
	b6 --> b5