Regular Expressions: Genius/Idiot Dichotomy

Today, I was faced with a little bug that I introduced into some code nearly a month ago, while fixing another bug.

I had enabled use of regular expressions for a Javascript DataTables fnFilter API method.

Originally, I thought I was being quite clever, when I wrapped the string to search for (i.e. the “needle”) in start and end string anchors (e.g. ^needle$). It fixed a problem I was having when I searched for “D” but got results for “D” and “DD”. Done and dusted.

Or so I thought.

A few weeks later, people aren’t finding any results.

That’s because while my solution worked great for times when the string to search (i.e. the “haystack”) was a single value (e.g. “D” or “DD”), it didn’t work so great when the haystack string was “A<br/>B<br/>D”.

Suddenly, I felt like an idiot.

How had I missed that? It was so obvious, in hindsight, that my original solution wasn’t going to cut it. So, I started trying out a few different ideas, but I wasn’t entirely sure how DataTables was filtering the data, so it was more guesswork than I would’ve liked.

Then, I realized that my original solution was all right, if I added another option. I should look for the start and end of the string, OR an angle bracket. So, “^needle$” became” (^|>)needle(<|$)”. This meant that I was able to find “needle<br/>”, “<br/>needle”, “needle”, or “<br/>needle</br>”.

Perfection! Genius! That fixed it! I tested it with lots of different examples and it worked perfectly!

Amazing how just a few characters can make the difference between feeling like an idiot and feeling like a genius.

(PS: Of course, I’m neither a genius or an idiot, but probably somewhere in between. Alas, it seems to me that programming is often about dancing between these two extremes.)

(PPS: Regular expressions are actually a lot of fun. I think the temptation is to avoid them because they can be difficult to understand, but they’re incredibly powerful and a valuable tool to anyone who needs to manipulate data.)