I have a python snippet as follows
5 + 6 * 8
Compiling this to AST gives:
>>> ast.dump(compile('5+6*8', '', 'eval', ast.PyCF_ONLY_AST))
Expression(body=BinOp(left=Num(n=5), op=Add(), right=BinOp(left=Num(n=6), op=Mult(), right=Num(n=8))))
Note that in the generated AST 6 * 8
is not optimized into 48.
However, if I compile this generated AST and disassemble, 6 * 8
is replaced by 48
.
>>> astree = compile('5+6*8', '', 'eval', ast.PyCF_ONLY_AST)
>>> dis.disco(compile(astree, '', 'eval'))
1 0 LOAD_CONST 0 (5)
3 LOAD_CONST 3 (48)
6 BINARY_ADD
7 RETURN_VALUE
My question
How to compile the code to bytecode without the constant folding optimization? I need to do this for developing an obfuscator.