50 Regex Jokes Every Developer Has Lived Through

I had one problem. I wrote a regex. Now I have two.

"It validates email addresses." No it does not.

The regex worked on the test string. The regex did not work on anything else.

Greedy quantifiers are named greedy for emotional reasons, not technical ones.

A senior engineer is someone who has stared at a backreference for twenty minutes and admitted they had no idea what it did.

"Can you parse HTML with a regex?" No. "But just a little?" Still no.

I once wrote a regex so good I forgot what it matched.

The email pattern is forty-three characters long. It is wrong in seven different ways. It has been in production for eight years.

Catastrophic backtracking is what your CPU does when you nest two quantifiers and walk away.

"Why is the page hanging?" Somebody pasted a long string into the validator and the regex went on a journey.

I trust documentation. I just trust a regex tester more.

The regex matches the URL. It also matches three other things that are not URLs. Nobody noticed for a year.

"Can you make it match phone numbers?" "Which country?" "All of them."

Lookbehinds are great until you remember half the engines do not support them.

Anchors are a love language.

"Just use a regex." The four most expensive words in software.

I wrote the regex on Tuesday. On Friday I had to read it again. On Monday I rewrote it from scratch.

Two kinds of developers: Those who escape special characters, and those whose patterns silently break the first time a dot appears in the input.

Character classes are the friendliest part of regex. This is not a high bar.

"The regex is documented." The documentation is a comment that says: `matches stuff`.

Non-greedy quantifiers exist because greedy ones broke something five years ago and somebody filed a ticket.

I do not always test my regex. But when I do, I regret what I find.

The same regex behaves differently in PCRE, ECMAScript, Python, and Go. This is fine.

"Can you write a regex to extract names from this CSV?" No. "Why not?" Use a CSV parser.

Capturing groups are how I track which part of my pattern is going to confuse me next.

Named groups are capturing groups with self-esteem.

"Why is this regex slow?" Because it backtracks for every character in the input and you wrote it like that on purpose.

A good regex has one job. A great regex has one job and a comment explaining why nobody should touch it.

The regex worked. The edge case did not.

"How long did the regex take?" "Forty minutes to write. Three days to debug."

Regex flags are the small switches that change everything. Forgetting the case-insensitive one has cost me real money.

"I can't read this pattern." Neither can the person who wrote it. They left in 2017.

ReDoS is the security vulnerability you introduce by being clever on a Tuesday afternoon.

The dot matches any character. Except newlines. Unless you set the flag. Which flag varies by engine.

"We sanitize inputs with a regex." That is not what sanitize means.

Regex101 is the only reason any of my patterns work in production.

I once explained a regex out loud and three people quietly left the meeting.

"It's a simple regex." The pattern is on line 14. The bug report is on day six.

Every codebase has one regex that nobody understands and everybody is afraid to remove.

The first version of the pattern was elegant. The production version has eight alternations and a comment that says: `do not touch`.

"Can you make it match Unicode?" "Sure. Which decade of Unicode would you like?"

The intern wrote a regex that matched every IP address. It also matched the company phone number.

Atomic groups are great if you remember they exist and your engine supports them. Mine neither remembered nor supported.

"What does this regex do?" "It used to validate slugs. Now it just exists."

I learned regex by failing to parse a date in 2014. I am still learning regex.

The regex is forty-six characters. The test case is a thousand lines. I am the bug.

"It compiles." That is the lowest possible bar and you should not be proud.

Some people relax by gardening. I relax by writing a pattern that finally matches the date format in column 3.

Negative lookaheads are how I tell the engine what I do not want and then act surprised when I get it anyway.

Writing regex is basically: Solving a small problem with a small tool that has its own small set of medium-sized problems.