TIL: Match unicode letters in regex

Sven — Wed, 20 Dec 2017 12:41:51 +0000

Note: This post was originally posted on sven-dittmer.netlify.com.

Today I faced an interesting challenge on exercism.io.
To me, the most interesting part was about finding out if a sentence was
yelled or not. A sentence counts as yelled if all its word are upper case.

You can find the whole challenge here: Bob in elixir - exercism.io

Here's my first attempt:

String.upcase(input) == input

This one passes some tests, but it failes for sentences that don't contain any word, like "1, 2, 3". Since there's no upper case word, it shouldn't count as yelled.
Next try:

input =~ ~r/\w/u && String.upcase(input) == input

With this regex, I made sure the word contains at least on letter. And since I'm still checking that every letter is upper case, that should do the job, right? Well, not quite.

The exercise's last test was about a sentence yelled in Russian. (Link:
Bob in elixir test suite)
Turns out \w matches only letters of the latin alphabet. It is equivalent to [a-zA-Z]. To make the regex work, I'd have to match any unicode letter. After googling a little bit, I found the solutions here: Regular-Expressions.info. With PCRE we can match different categories of unicode with \p{<some_unicode_category>}. For letters, you can use \p{Letter} or its short form \p{L}. Here's the final code that makes the last test pass:

input =~ ~r/[\p{L}]/u && String.upcase(input) == input

This is my first post on dev.to. Feedback is much appreciated!

Hi, I'm Sven

Sven — Tue, 02 May 2017 21:06:46 +0000

I have been coding for 10 years (3 years professionally).

You can find me on GitHub as svendittmer

I live in Hamburg, Germany.

I work for Brandnooz

I mostly program in these languages: Ruby.

I am currently learning more about Linux.

Nice to meet you.

DEV Community: Sven

TIL: Match unicode letters in regex

Hi, I'm Sven