[Tutor] Regex not working as desired

[Steven D'Aprano]

On Mon, Feb 26, 2018 at 11:01:49AM -0800, Roger Lea Scherer wrote:
>   The first step is to input data and then I want to check to make sure
> there are only digits and no other type of characters. I thought regex
> would be great for this.

I'm going to quote Jamie Zawinski:

    Some people, when confronted with a problem, think "I know, 
    I'll use regular expressions." Now they have two problems.


Welcome to the club of people who discovered that regexes are just as 
likely to make things worse as better :-(

Here's another, simpler way to check for all digits:

value = '12345'  # for example
value.isdigit()

The isdigit() method will return True if value contains nothing but 
digits (or the empty string), and False otherwise.


Sounds like just what you want, right? Nope. It *seems* good right up to 
the moment you enter a negative number:

py> '-123'.isdigit()
False

Or you want a number including a decimal point. Floating point numbers 
are *especially* tricky to test for, as you have to include:

# mantissa
optional + or - sign
zero or more digits
optional decimal point (but no more than one!)
zero or more digits
but at least one digit either before or after the decimal point;
# optional exponent
E or e
optional + or - sign
one or more digits


It is hard to write a regex to match floats.

Which brings us to a better tactic for ensuring that values are a valid 
int or float: try it and see!

Instead of using the Look Before You Leap tactic:

if string looks like an int:
    number = int(string)  # hope this works, if not, we're in trouble!
else:
    handle the invalid input


we can use the "Easier To Ask For Forgiveness Than Permission" tactic, 
and just *try* converting it, and deal with it if it fails:

try:
    number = int(string)
except ValueError:
    handle the invalid input

The same applies for floats, of course.

Now, one great benefit of this is that the interpreter already knows 
what makes a proper int (or float), and *you don't have to care*. Let 
the interpreter deal with it, and only if it fails do you have to deal 
with the invalid string.




By the way: absolute *none* of the turtle graphics code is the least bit 
relevant to your question, and we don't need to see it all. That's a bit 
like going to the supermarket to return a can of beans that you bought 
because they had gone off:

"Hi, I bought this can of beans yesterday, but when I got it home and 
opened it, they were all mouldy and green inside. Here's my receipt, 
and the can, and here's the can opener I used to open them, and the bowl 
I was going to put the beans into, and the microwave oven I would have 
used to heat them up, and the spoon for stirring them, and the toast I had 
made to put the beans on, and the salt and pepper shakers I use."

:-)



-- 
Steve