Python | Create a Dictionary with Values as expressions without evaluating or treating it as string

I need to create a dynamic synthetic Pandas DF in Python and am using Faker and Random functions to create the same.
To do so, I am creating a dictionary as a source to build pandas df where some columns will have static values whereas some column's values are generated through some expressions or Faker objects based on user input.\

for eg:

if user wants to create synthetic dataframe of size 5 with 2 columns, the user provides two inputs as arguments:

col1 = 1
col2 = tuple('1', '2')

so the dictionary should ideally be constructed as:

dict1 = {
        'col1': 1,
        'col2': fake.random_element(elements=('1','2'))
     }

And the Dataframe can be constructed by running the expressions within the dictionary in a loop, as:

from faker import Faker  
import pandas as pd

fake = Faker()  

def generate_fake_data(numRows, inputDict):
    output = [
                 inputDict for x in range(numRows)
            ]
    
    return output

num_records = 5
df_sythetic = pd.DataFrame(generate_fake_data(num_records, dict1))

The problem is that if I try to create the dictionary without letting it evaluate the expression for col2 beforehand, it binds the value as string like:

dict1 = {
        'col1': 1,
        'col2': 'fake.random_element(elements=('1','2'))'
     }

How can I create the dictionary without evaluating the expression columns and also not treating them as string ?

One possible solution is to create a dictionary where keys are column names and values are functions (lambdas):

import pandas as pd
from random import randint


def generate_fake_data(num_rows, input_dict):
    out = []
    for _ in range(num_rows):
        out.append({k: v() for k, v in input_dict.items()})
    return pd.DataFrame(out)


num_records = 5

dict1 = {
    'col1': lambda: 1,
    'col2': lambda: randint(1, 2)
}

df_synthetic = generate_fake_data(num_records, dict1)
print(df_synthetic)

Prints:

   col1  col2
0     1     2
1     1     2
2     1     1
3     1     1
4     1     2

Leveraging the answer by @andrej-kesely here https://stackoverflow.com/a/76408359/218663 and your desire to allow users to enter a more free form set of valid parameters, I believe we can leverage the ast module.

import pandas
import random
import ast

test_rows = 10
template = {}
while True:
    key = input("Enter a column name: ")
    if not key:
        break

    value = input(f"Enter a list [1,2,3] of valid values for column \"{key}\": ")
    try:
        value = ast.literal_eval(value)
    except (ValueError, SyntaxError):
        pass

    template[key] = value if isinstance(value, list) else [value]

df_synthetic = pandas.DataFrame([
    {key: random.choice(value) for key, value in template.items()}
    for _ in range(test_rows)
])
print(df_synthetic)

When prompted, the inputs:

col1
1
col2
[1,2]
<return>

You should get back:

   col1  col2
0     1     2
1     1     1
2     1     1
3     1     2
4     1     1
5     1     1
6     1     1
7     1     2
8     1     1
9     1     2