[sf-lug] ThinkPython: coding problem

[Jeffrey Malone]

In programming, it's generally best to write "defensively" -- expect
the user to give you garbage, and handle it gracefully.

The exception handling method is valid, but it could cause problems on
being able to gracefully handle an error.  Exceptions are often great
in handling more unexpected errors than input validation.  What I mean
by that, is that you expect the user to be imperfect, and the chance
of a user giving you incorrect data is high.  But when it comes to
calling on functions that are preset and defined, you expect them to
give you valid data.  Exceptions are generally meant to deal with when
they don't.
Input validation can often better be handled in a simple structure, in
my *opinion*.
Here is an example of it, put into a loop that gives the user three
tries to enter valid data:

x = 0
while x < 3:
        response = raw_input("Please enter a number: ")
        if response.isdigit():
                break
        print "That's not a numeric input!"
        x+=1
if not response.isdigit():
        print "You failed to enter a numeric value"


Anyway, I'm not saying the method you have is wrong -- not at all.
I'm simply putting forward an alternate idea on handling it (one often
found in C, although using a do-while loop which doesn't exist in
python).  If nothing else, I hope it gives a different perspective on
things ;)

Jeffrey

On Sat, Oct 11, 2008 at 10:44 AM, Jim Cortez <jim at jimcortez.com> wrote:
> There are 3 simple ways to tackle this (at a beginner's level).
>
> You can also use the input() function to read in a number. The downside,
> however, is that it will result in an error if the user enters in a string.
> To handle that error, you can use the try, except method to recover.
> Sometimes, however, python will throw a different error than you expected.
> Read the message at
> http://docs.python.org/library/functions.html?highlight=input#input for more
> information.
>
> Example:
> =======
> $ python
>>>> a = input()
> 8
>>>> a
> 8
>>>> a = input()
> blah
> Traceback (most recent call last):
>  File "<stdin>", line 1, in <module>
>  File "<string>", line 1, in <module>
> NameError: name 'blah' is not defined
> =======
>
> You may also think of using isinstance for type checking. This takes in an
> object, and a type and then looks to see if the 2 types match up.
> http://docs.python.org/library/functions.html?highlight=isinstance#isinstance
>
> Example:
> =======
>>>> isinstance(8, int)
> True
> =======
>
> And last, if you decide to use raw_input(), then you can also use .isdigit().
> Take a look at
> http://docs.python.org/library/stdtypes.html?highlight=isdigit#str.isalnum
> for isdigit and a couple other handy string functions.
>
> Example:
> =======
>>>> "8".isdigit()
> True
>>>> "87654321".isdigit()
> True
> =======
>
> I hope that helps!
> Jim
>
> On Friday 10 October 2008 9:42:27 pm Alex Kleider wrote:
>> # ThinkPython: question-
>>
>> As part of our ThinkPython class exercises I want to
>> check user input.
>> i.e. the user is asked to enter a number and I want the routine
>> to check that it is indeed a number that is entered.
>> It seems that everything "collected" by raw_input is considered
>> a string, even if only digits are included.
>>
>>
>>
>> def int_check(n):
>>   if type(n)==int:
>>     return True   # as used below, NEVER returns True.
>>   else:
>>     return False
>>
>> response=raw_input('enter a number: ')
>> if int_check(response):
>>   pass #procede with what needs to be done
>> else:
>>   pass #take corrective action
>>
>> Can anyone suggest how I can accomplish this?
>> I'm familiar with Pascal and its strong typing which would prevent such a
>> problem from occurring as a runtime problem (vs a syntactic error.)
>>
>> ak
>> ~
>>
>>
>>
>>
>> _______________________________________________
>> sf-lug mailing list
>> sf-lug at linuxmafia.com
>> http://linuxmafia.com/mailman/listinfo/sf-lug
>
>
> _______________________________________________
> sf-lug mailing list
> sf-lug at linuxmafia.com
> http://linuxmafia.com/mailman/listinfo/sf-lug
>