regular expressions resources and regular expression problem

Hi all, I don't know anything about regular expressions and I have two questions:

1. What is the most effective way to learn regular expressions and where to find the best possible tutorials, examples, exercises etc.
2. I have the following problem. I would like to extract (1) a list variables and (2) a list of matlab functions from a string. The variables have a precise definition: (i) they can start with 'a' 'd' or 'ee'; (2) then an underscore '_'; (3) then an integer 'n'. For example, the variables could be 'a_1', 'd_100', 'ee_3289' and the expression to parse could look like: 'cos(exp(a_2/4*d_1^3-ee_4)+d_85)'. In this example, the list of variables is 'a_2', 'd_1', 'ee_4', 'd_85' and the list of matlab functions is 'cos', 'exp'. My question here is what is the (best) regular expression to extract that information?

Thanks, Pat

Answer 1

[Andrei Bobrov]

1. Read here about the Regular Expressions

2.

str = 'cos(exp(a_2/4*d_1^3-ee_4)+d_85)';

fun = regexp(str,'\w*(?=\()','match')
var = regexp(str,'(a|d|ee)_\d*','match')

[Patrick Mboma]

Hi Andrei,

Thanks for the reply. My sense is that the patterns are not robust enough. The variables must start with either 'a', 'd', or 'ee' and nothing else. For instance, 'f_1' would not be a valid variable name. As for the functions, how do you detect that a particular expression is a function? would it also capture functions of several arguments? would it capture functions with no arguments?

Thanks alot Pat.

[Andrei Bobrov]

Corrected of var

[Daniel Shub]

Your regexp foo is so much better than mine. I always have to make things up.

Answer 2

[Ned Gulley]

Another way to get good at regular expressions is to practice on Cody with all the problems that have been tagged "regexp".

Answer 3

[Daniel Shub]

Assuming a string x,

[a, b] = regexp(x, '(a|d|(ee))(_)([0-9]*)', 'start', 'end')

will find the start and end indices of your "variables." This will allow for a variable a_0 which may or may not be desired. Without a rule for defining functions that handles the numerous edge cases, creating the regexp is difficult. Andrei's answer seems reasonable to me.

Answer 4

[Patrick Mboma]

Thank you both. That is one big problem solved. The biggest remains though: how do I become as good as you guys are with regular expressions?

Thanks a lot, Pat.

[Walter Roberson]

I think there is an O'Reilly book on regular expressions.

The regular expressions used in MATLAB are a mish-mash between capabilities in other languages. They have similarities to Perl, so it might make sense to study the resources for regular expressions in Perl (and there is probably a book or three for that topic alone.)

[Patrick Mboma]

Thanks Walter. It is something I will look into.