DEV Community

Regex isn't that hard

Dinys Monvoisin on September 12, 2020

Regex is the thing that you only learn when you need it. Unless you are processing a considerable amount of data, you likely won’t use it. Does t...

Read full post

Riccardo Bernardini • Sep 12 '20 • Edited

Am I the only one in the world who actually loves regular expressions? I learned them when I was studying about compilers and I always found them a very powerful tool, not necessarily for niche applications or very large amount of data.

I use them to build "tokenizers" or extract information from text files in just a couple of line of codes (in Ruby, mostly), for example (the first thing that came to my mind)

   $stdin.each do |line|
       next unless line =~ /^([a-z]+) *: *(.*)$/
       name=$1
       value=$2
    end

The syntax is not great, I agree, it looks much like line noise. I always wondered about an alternative syntax, but everything I tried (not much, to be honest) was not a really huge improvement.

Oh, yes, and let's not forget search-and-replace-regexp in emacs... You can do wonderful stuff with a single command.

Dinys Monvoisin • Sep 13 '20

This should be a great job working on regular expression all day. But it's not my cup of tea. I would prefer a mixing Regex with some sort of development.

Oh, is Emacs your favourite editor then? I can't image how good it will be to customise a replace operation using Regex.

pentacular • Sep 13 '20

Regex can be useful, but can also be a trap.

When you use regex, be sure to use just enough abstraction that you can swap out the regex implementation with a parser later on.

There are three main traps with regex:

Regex do not handle recursive structures.
Regex do not handle irregular languages.
Regex scale rapidly to become impossible for a human to understand.

It is quite difficult to predict when you'll hit one of these limits, so a little abstraction goes a long way.

Instead of putting regex directly in your code, abstract them with a procedure that does something: e.g., getName(foo) instead of (foo.match(/([^/]+)/) || [])[1]; :)

Catherine Mohan • Sep 12 '20

Great article! I've been using RegEx patterns a lot recently in Powershell as some commands return values as a very long string instead of a proper object. RegEx patterns make pulling the data much easier. I use RegEx 101 to help build my pattern strings. It has very helpful color coding and a dictionary of all the different RegEx operators.

Dinys Monvoisin • Sep 12 '20

Thank you, Catherine, for contributing to this article and providing the readers with addition resources. I wonder what you were using Regex for in PowerShell. Are you using Grep?

Catherine Mohan • Sep 12 '20 • Edited

No, Powershell can use RegEx natively for working with strings. I mostly use Select-String -Pattern to pull substrings out of large string responses. Some string commands even use it by default and you have to remember that or else you'll be a bit confused why some of your code is not responding the way you hope.

-split and -replace will use RegEx to match strings, but .Split() and .Replace() don't. So "catherine.mohan" -split "." returns all the characters and ("catherine.mohan").Split(".") returns catherine and mohan as expected.
You can escape the period and it'll work too.

Dinys Monvoisin • Sep 13 '20

Is your file sparsely found everywhere that's why you are using command line? Often you will just use a program to do all these.

Catherine Mohan • Sep 13 '20

I'm not sure what file you're referring to. I use RegEx in the Powershell CLI, Powershell scripts, and in Powershell apps that I create. Mostly for parsing strings, and occasionally for searching strings. Powershell commands usually return objects with properties, but recently I've had to use some commands that return objects with a single property that is just a long string with all the values in a list Since I can't use the typical $object.property notation to get values, I have to use RegEx to parse the giant string looking for the values I need.

Dinys Monvoisin • Sep 13 '20

Oh, using a string itself in PowerShell. Interesting. May I have more context about the application of it?

Abhigyan • Sep 12 '20

NodeJS 14.1 REPL

> Regex + JS == RegJSex;
> true

Dinys Monvoisin • Sep 13 '20

You make me laugh. It's good to have comments like that sometimes.

Abhigyan • Sep 13 '20

I think that I am a very frank guy. I always make things too funny that it appears easy, even if it's difficult.

John Robertson • Sep 12 '20

Regex is invaluable for software dev and sysadmin. Just put in the effort to learn it - I guarantee it will be worth your time.

Dinys Monvoisin • Sep 13 '20

Try to tell that to the new people learning programming. All of them learning about web development to only create beautiful screen.

John Robertson • Sep 13 '20

I suppose it all depends on your goals. Web development using $WEB_DEV_PLATFORM_DU_JOUR creates an initial perception of rapid progress. If you are one of us that is tasked with completing a complex project all the way to sustainable production, then traditional computer science concepts and tools become essential.

Dinys Monvoisin • Sep 13 '20

With rapid changing market, it is better to build a quick and dirty prototype. However, I do agree that a strong understanding of CS concepts is fundamental.

By the way, do you speak French? What with "DU_JOUR"?

John Robertson • Sep 14 '20

"it is better to build a quick and dirty prototype'
Again, this depends on your goals. As a consultant I only get paid for a working product. If you are working as an employee, then the best strategy is to throw together the prototype, get some kudos, and move on to the next project.
Sadly I do not speak French. "du jour" is a French phrase adopted by English speakers for some time.

Franco Traversaro • Sep 14 '20

Pay attention, the first example is wrong: /.+@.com/ matches something like "name@xcom". A better example (still not covering a lot of peculiar cases) could be /.+@.+\.com/