>>> code = """for i in range(5):\n    print('Hello world!')"""
>>> exec(code)
Hello world!
Hello world!
Hello world!
Hello world!
Hello world!

>>> expression = '5 + 3 * a'
>>> a = 5
>>> result = eval(expression)
>>> result
20

compile built-in function can be used to precompile an expression to a code object; this code object can then be passed to eval. This will speed up the repeated executions of the evaluated code. The 3rd parameter to compile needs to be the string 'eval'.

>>> code = compile('a * b + c', '<string>', 'eval')
>>> code
<code object <module> at 0x7f0e51a58830, file "<string>", line 1>
>>> a, b, c = 1, 2, 3
>>> eval(code)
5

>>> variables = {'a': 6, 'b': 7}
>>> eval('a * b', globals=variables)
42

As a plus, with this the code cannot accidentally refer to the names defined outside:

>>> eval('variables')
{'a': 6, 'b': 7}
>>> eval('variables', globals=variables)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<string>", line 1, in <module>
NameError: name 'variables' is not defined

Using defaultdict allows for example having undefined variables set to zero:

>>> from collections import defaultdict
>>> variables = defaultdict(int, {'a': 42})
>>> eval('a * c', globals=variables)  # note that 'c' is not explicitly defined
0

If you have a string that contains Python literals, such as strings, floats etc, you can use ast.literal_eval to evaluate its value instead of eval. This has the added feature of allowing only certain syntax.

>>> import ast
>>> code = """(1, 2, {'foo': 'bar'})"""
>>> object = ast.literal_eval(code)
>>> object
(1, 2, {'foo': 'bar'})
>>> type(object)
<class 'tuple'>

However, this is not secure for execution of code provided by untrusted user, and it is trivial to crash an interpreter with carefully crafted input

>>> import ast
>>> ast.literal_eval('()' * 1000000)
[5]    21358 segmentation fault (core dumped)  python3

Here, the input is a string of () repeated one million times, which causes a crash in CPython parser. CPython developers do not consider bugs in parser as security issues.

It is not possible to use eval or exec to execute code from untrusted user securely. Even ast.literal_eval is prone to crashes in the parser. It is sometimes possible to guard against malicious code execution, but it doesn't exclude the possibility of outright crashes in the parser or the tokenizer.

To evaluate code by an untrusted user you need to turn to some third-party module, or perhaps write your own parser and your own virtual machine in Python.