My First RegEx

Published by FIGBERT on
< Back

(?<=\.\/IssuerIcons\/).*(?=\.png)

§ What it does

Let’s break it apart piece by piece. It’s meant to return the filename from a given path in a specific format. When given ./IssuerIcons/Example.png, the script should return Example.

The first section is a positive lookbehind: (?<=\.\/IssuerIcons\/). This asserts that the selected portion must by preceeded by ./IssuerIcons/.

The section section is a pattern match: .*. This matches all characters, excluding newlines, of any length.

The third and final section is a positive lookforward: (?=\.png). This asserts that the selected portion must be followed by .png.

§ Background

Yesterday, I wrote my first regular expression. I’ve always regarded regex as an arcane art of true shell wizards – and for the most part, I still do. Now though, I’ve gotten a glimpse of their world.

I wrote this expression for Tofu, an open-source 2FA app for iOS. I had been thinking of making more “issuer icons” for the app when I noticed that the icons were all designed in one Sketch file. This could become a problem if two people were to edit the file at one time, so I suggested a solution.

The author got back to me with another idea: ditch Sketch altogether, and replace it with a shell script that generates icons from a directory of pngs. I wrote it. I’ve since refined it, with help from ThinkChaos, to make it independent of any preceding directories using sed: sed -E 's:.+/(.+)\.png:\1:'.

It was a really fun experience – I’ve never written bash before! It reminded me a lot of Drew DeVault’s post Become shell literate. I work primarily from my terminal, and am slowly working my way up to true mastery – I’m nowhere near close, but this was a step in the right direction.