When not to use a regex August 13, 2017 on Drew DeVault's blog

The other day, I saw Learn regex the easy way. This is a great resource, but I felt the need to pen a post explaining that regexes are usually not the right approach.

Let’s do a little exercise. I googled “URL regex” and here’s the first Stack Overflow result:

https?:\/\/(www\.)?[-a-zA-Z0-9@:%._\+~#=]{2,256}\.[a-z]{2,6}\b([-a-zA-Z0-9@:%_\+.~#?&//=]*)

source

This is a bad regex. Here are some valid URLs that this regex fails to match:

Here are some invalid URLs the regex is fine with:

This answer has been revised 9 times on Stack Overflow, and this is the best they could come up with. Go back and read the regex. Can you tell where each of these bugs are? How long did it take you? If you received a bug report in your application because one of these URLs was handled incorrectly, do you understand this regex well enough to fix it? If your application has a URL regex, go find it and see how it fares with these tests.

Complicated regexes are opaque, unmaintainable, and often wrong. The correct approach to validating a URL is as follows:

from urllib.parse import urlparse

def is_url_valid(url):
    try:
        urlparse(url)
        return True
    except:
        return False

A regex is useful for validating simple patterns and for finding patterns in text. For anything beyond that it’s almost certainly a terrible choice. Say you want to…

validate an email address: try to send an email to it!

validate password strength requirements: estimate the complexity with zxcvbn!

validate a date: use your standard library! datetime.datetime.strptime

validate a credit card number: run the Luhn algorithm on it!

validate a social security number: alright, use a regex. But don’t expect the number to be assigned to someone until you ask the Social Security Administration about it!

Get the picture?

Articles from blogs I read Generated by openring

Disappointing phones

Since 2019, my phone has been a OnePlus 5T running LineageOS, and I've loved it.Unfortunately, it uses 3G for calls, and all New Zealand networks plan to turn off their 3G network in early 2026. I would still be able to use this phone for taking photo…

via Cadence's Weblog February 8, 2026

Trudging Through Nonsense

Last week Anthropic released a report on disempowerment patterns in real-world AI usage which finds that roughly one in 1,000 to one in 10,000 conversations with their LLM, Claude, fundamentally compromises the user’s beliefs, values, or actions. They not…

via Aphyr: Posts February 4, 2026

Binary Dependencies: Identifying the Hidden Packages We All Depend On

We need better tools for uncovering phantom binary dependencies. Not having these tools makes our global tech infrastructure less secure, and puts a strain on the Open Source maintainers we rely on.

via Vlad's Website January 31, 2026