What is the most "correct", Pythonic way to do user input validation in Python?
I've been using the following:
while True: stuff = input("Please enter foo: ") try: some_test(stuff) print("Thanks.") break except SomeException: print("Invalid input.") Which is nice and readable, I suppose, but I can't help wondering if there isn't some built-in function or something I should be using instead.
I like decorators to separate the checking from the rest of the input handling.
#!/usr/bin/env python def repeatOnError(*exceptions): def checking(function): def checked(*args, **kwargs): while True: try: result = function(*args, **kwargs) except exceptions as problem: print "There was a problem with the input:" print problem.__class__.__name__ print problem print "Please repeat!" else: return result return checked return checking @repeatOnError(ValueError) def getNumberOfIterations(): return int(raw_input("Please enter the number of iterations: ")) iterationCounter = getNumberOfIterations() print "You have chosen", iterationCounter, "iterations." EDIT:
A decorator is more or less a wrapper for an existing function (or method). It takes the existing function (denoted below its @decorator directive) and returns a "replacement" for it. This replacement in our case calls the original function in a loop and catches any exception happening while doing so. If no exception happens, it just returns the result of the original function.
The most Pythonic way to do this kind of validation of "User INput" is to catch an appropriate exception.
Example:
def get_user_input(): while True: try: return int(input("Please enter a number: ")) except ValueError: print("Invalid input. Please try again!") n = get_user_input() print("Thanks! You entered: {0:d}".format(n)) It's also good practice to allow exceptions occur where they lie and allow them to bubble up rather than hide them so you can clearly see what's going wrong in a Python Traceback.
In this case Validating User Input -- Use Python's Duck Typing and catch the error. i.e: If it acts like a duct, it must be a duck. (If it acts like an int, it must be an int).
A bit complicated, but may be interesting:
import re from sys import exc_info,excepthook from traceback import format_exc def condition1(stuff): ''' stuff must be the string of an integer''' try: i = int(stuff) return True except: return False def condition2(stuff): ''' stuff is the string of an integer but the integer must be in the range(10,30)''' return int(stuff) in xrange(10,30) regx = re.compile('assert *\( *([_a-z\d]+)') while True: try: stuff = raw_input("Please enter foo: ") assert(condition1(stuff)) assert ( condition2(stuff)) print("Thanks.") break except AssertionError: tbs = format_exc(exc_info()[0]) funky = globals()[regx.search(tbs).group(1)] excepthook(exc_info()[0], funky.func_doc, None) result
Please enter foo: g AssertionError: stuff must be the string of an integer Please enter foo: 170 AssertionError: stuff is the string of an integer but the integer must be in the range(10,30) Please enter foo: 15 Thanks. .
I found a way to simplify:
from sys import excepthook def condition1(stuff): ''' stuff must be the string of an integer''' try: int(stuff) return True except: return False def another2(stuff): ''' stuff is the string of an integer but the integer must be in the range(10,30)''' return int(stuff) in xrange(10,30) tup = (condition1,another2) while True: try: stuff = raw_input("Please enter foo: ") for condition in tup: assert(condition(stuff)) print("Thanks.") break except AssertionError: excepthook('AssertionError', condition.func_doc, None) 
some_testfunction, please