Struggling with regular expressions? Then visit 'Regex Crossword', a site to learn them through a Sudoku-like game

Really fun!

Nice. Next step - regex golf! https://xkcd.com/1313/

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

The infinity combinations you can get and that they aren't easy to read. IMO is one of the most powerful tool for text search but also a dangerous one (the day you create your RE it totally makes sense and it is hyper intuitive... six weeks later you want to cry).

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

The infinity combinations you can get and that they aren't easy to read. IMO is one of the most powerful tool for text search but also a dangerous one (the day you create your RE it totally makes sense and it is hyper intuitive... six weeks later you want to cry).

I see. So the cautionary tale for young programmers is, if you aren't solid in your commenting and variable names, you'll be condemned to use this stuff.

Oh man, finally a -hopefully- way to learn better regex!

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

The infinity combinations you can get and that they aren't easy to read. IMO is one of the most powerful tool for text search but also a dangerous one (the day you create your RE it totally makes sense and it is hyper intuitive... six weeks later you want to cry).

I see. So the cautionary tale for young programmers is, if you aren't solid in your commenting and variable names, you'll be condemned to use this stuff.

I did some commenting in few regex I had to do at work the other day... and two days later when I revisit them it wouldn't work. Turned out with all the rush to leave the office I had confused some symbols and had messed it xD

Fun, when you're not tearing your hair out

Nice. Next step - regex golf! https://xkcd.com/1313/

This might have been inspiration for the comic: https://alf.nu/RegexGolf

Oh man, finally a -hopefully- way to learn better regex!

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

The infinity combinations you can get and that they aren't easy to read. IMO is one of the most powerful tool for text search but also a dangerous one (the day you create your RE it totally makes sense and it is hyper intuitive... six weeks later you want to cry).

I see. So the cautionary tale for young programmers is, if you aren't solid in your commenting and variable names, you'll be condemned to use this stuff.

I did some commenting in few regex I had to do at work the other day... and two days later when I revisit them it wouldn't work. Turned out with all the rush to leave the office I had confused some symbols and had messed it xD

The linked site, https://regex101.com/ is actually nice for situations like that. It parses the regular expression into components and explains what they do.

As for the crossword itself, it's actually doable and I completed almost all the official puzzles, only the hexagonal one is unfinished.

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

Unlike programming languages, you don’t have meaningful variable or function names to rely on. It’s just symbols and a lot of nesting of rules and patterns. So it’s about a legible as minified code, if not less.

For example, this very simple pattern will match semi-colons that appear at the start or end of a line, or are followed by another semi-colon (I use it to clean up a list after removing duplicate entries):
^;+|;(?=;)|;$

Every single character has significance, a purpose of its own. Namely:
<start of line><semi-colon><1 or more><OR><semi-colon><start of group><look ahead (without including in results)><match><semi-colon><end of group><OR><semi-colon><end of line>

It’s very hard to come up with the exact pattern you are trying to match (without getting false positives), and often as hard to remember what it’s trying to match if you come back to it after the fact. If someone else made it and you don’t even know what it’s looking for, you have to decipher it piece by piece.


Last edited by Salvatos on 1 Apr 2020 at 7:29 pm UTC

Or, as jwz so eloquently put it:

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

Doesn't make sense. I'm better with Regex than with Sudoku.

Looks brilliant. I have managed to mostly avoid learning regex so far. This is a good way to learn.

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

The infinity combinations you can get and that they aren't easy to read. IMO is one of the most powerful tool for text search but also a dangerous one (the day you create your RE it totally makes sense and it is hyper intuitive... six weeks later you want to cry).

There are some things in life that are easy to understand, and there are some things that are difficult. The really tricky ones are those that you think you understand perfectly, but it turns out you couldn't be more wrong. Regexes are like that for me. The rules all make perfect logical sense, but whenever I try to follow them I come up with something that totally fails to work as expected.

Fortunately, we have things like txt2regex. (The output from which always leaves me saying, “But that's what I did! Isn't it?”. Oh, well. It works.)

This looks interesting. I have to check it out. Regex is finicky as all hell. The rules are simple enough really, but the syntax ends up looking totally bonkers. And regex has a really nasty habit of doing stuff you don't expect. Like replacing way too much text.

This is pretty straighforward as far as regex goes, but I'm struggling with crap like this while editing a wiki. Not least because I need about 50 rules probably, not all looking like this. So yeah, a million things can go wrong. Thankfully there are excellent sites such as https://regexr.com/ (LOVE the cheatsheet on the left), that make it a bit more tolerable.

(\+?\d*\s?[–-]?\s?\d*\s?%?\s?)(\{\{Bleeding3\}\})
(\+?\d*%?)(\s?[–-]\s?)(%?\s)(\{\{\w*\}\})

This looks really interesting, I'll have to give it a shot! Always good to get more practice with regexes.

Weird. The definition sounds like a straightforward, fairly intuitive thing. What makes them so hard?

The main point of regexes is that they can use wildcards—symbols which don't match a literal symbol, but any of a (possibly very large set) of (possibly combinations of) literal symbols. (Like using '*' on the command line, but much more powerful and complex.) Using them for any but the simplest tasks thus usually requires a lot abstract, symbolic thought, and generally people aren't too good at being able to follow the combinatorially-explosive number of consequences that can come from changing a single symbol in the regex.

Like, imagine scanning a webpage to find a company's mailing address. As humans, we've got a lot of in-built tolerance for what an "address" might look like—depending on where the company is in the world it might have more or fewer fields than would be the norm in our home country, maybe the street or city name contains multiple separate words, etc. But we have a mental "prototype" of roughly what an address looks like and can use fuzzy logic to recognize things which, while deviating from the strict image of that prototype, are still addresses. Now imagine trying to get a computer to do the same thing; regexes allow you to encapsulate that insane amount of flexibility via (appropriately complex) combinations of wildcards. (Actually attempting to set up a regex that could handle addresses sounds like an unutterable nightmare. :O)