The 12th Annual Regular Expression Day - June 1st 2019

Of late I've been doing more work with Docker, which has necessitated writing shell scripts and learning more about the command line (full endorsement of Julia Evans Bite Size Command Line zine - very accesible).

One of the really powerful commands is sed, which can utilize regular expressions (https://www.gnu.org/software/sed/manual/sed.html#sed-regular-expressions).

So, for example, in one instance I use it to set a variable for stepping up and over to a different directory, adjacent to where the script is located.

DIR=$( sed 's+/build+/.secrets+g' <(echo $DIR) )

Regex for the win!

This is a little bit more meta, but once I had to resolve the following problem:

How do you know if two different regex expressions match will match the same set of entries?

For example:

• /a/ will match "a" without quotes
• /.*/ will match anything, including "a"

The problem is resolved by converting each regex to a Finite State machine and then computing the intersection of both by generating a third FSM and then converting it back to regex.

I ended up using a open source library for it https://github.com/qntm/greenery

Not sure if this will qualify as a proper answer, but... I haven't had my "unforgettable regex moment". It's more of a long-term relationship: don't worry, I'll explain.

See, most regex I've done is - to put it bluntly - uninteresting.
Most patterns I come up with are well-tread paths to solutions that aren't revolutionary in any way. They aren't some flash-in-the-pan, fancy approach. They aren't some new sugar for my syntactical sweet tooth. Rarely are they anything you would draw inspiration from, or that spurns you on as a dev. Were programming analogous to fiction, regex would not be found in that section of the library with titles like "A Night to Surrender" or "40 Nights at Sea".

Rather, regex is unyielding, inflexible, and often stubbornly unforgiving. Yet, it's in those qualities that I find a deeper love for regex: reliable predicably!

Libraries come and go, that form data you're parsing is crazy, and your code style will change over time; yet that regex you wrote remains an obelisk of utility. It just works the same way every time. Consistency is its draw, and it doesn't disappoint.

So, I don't particularly remember any regex I've written. It's all unremarkably the unsung hero of my career, but the hero nonetheless. For that, I owe regex a debt of gratitude.

Graduated as a mathematician, I decided to take a career into programming some years ago.
One of the first assignments I got was to implement a regular expression for parsing social media urls of several platforms.

Not familiar with regular expressions before, I was - and certainly still am - amazed by the power of it!

I believe it was one of my first ever commits in which I composed this regex for parsing the Twitter variant:
^(?:http(?:s)?://)?(?:[w]{0,3}[.]?)?(twitter.com|@)?(?:/)?([A-Za-z0-9_]*)?(?:/)?(.+)?$

When processing the value "https://www.twitter.com/BenNadel" for example, the username would be extracted by the regex and the value be modified to "@BenNadel".

Beautiful.

On a recent project, I needed to provide "masked" versions of a user's email address and phone number for display. I was doing this in a ColdFusion app but I achieved my results by dropping down to Java and using replaceAll() on the string.

For the email, I needed to match/replace everything but the first character on the left side of the "@".

emailValue.replaceAll("(?<=.{1})[^@](?=[^@]*?@)", "*");
// Result: t***@myemail.com

For the phone number, I needed to match/replace everything but the last 4 digits in the value.

phoneValue.replaceAll(".(?=.{4})", "*");
// Result: ******8888

@Ben,

Booo for low turnout! Woohoo for winning!

With regard to the command line, I am a complete beginner, but I can't say enough good things about this: https://gumroad.com/l/bite-size-command-line - It's extremely practical and makes those incomprehensible man pages actually accessible. If you want something completely different for your next "A Book Apart", or just want to explore the command line, I'd recommend it. Also, in terms of helpfulness, here's another cool reference: https://tldr.sh/

Keep up the awesome blogging! And thanks!

@Ben,

Awesome! I'll be sure to check out JRegEx. Thanks.

For a cool online regex testing utility, check out http://gskinner.com/RegExr/ .
I've gotten a great deal of use out of it throughout the years :)