When working on the date formatting portion of a global web app recently, I really wanted to get two details right.
- Localize the dates – People in different parts of the world are used to seeing dates displayed differently. Give the people what they expect.
- Four-digit years – Two-digit years are confusing. It's difficult enough to parse "10/12/2011" when you don't know whether the month or the day comes first. Parsing "10/12/11" just adds to the cognitive load.
I know the user's locale. PHP has
IntlDateFormatter to handle
localizing the dates. This should be easy.
But for space and legacy reasons, I needed a format like "10/11/2012". Well,
IntlDateFormatter::SHORT gives "10/11/12". That's almost perfect.
Almost. You know, except for those pesky, confusing two-digit years.
I ran a test for
IntlDateFormatter::SHORT for a handful of
different locales anyway. Maybe I'd get lucky.
en_US: 10/11/12en_GB: 11/10/2012de_DE: 11.10.12fr_FR: 11/10/2012ru_RU: 11.10.12en_ZW: 11/10/2012ps_AF: م. ۲۰۱۲/۱۰/۱۱
Kinda lucky. I guess. I mean, some of those locales nail it. But I need every locale to have a four-digit year. And I certainly can't define every localized format myself: I'm completely unfamiliar with how things are done in Afghanistan, Zimbabwe, or the 200 or so states in between.
I took a walk.
On the walk, I came up with a plan.
IntlDateFormatter can get and set the date formatting
pattern. I can get the formatting pattern for a specific locale and coax it into
using a four-digit year. Good. But I needed to tread lightly; I didn't want to
mess with the date formatting pattern too much because I didn't want to, for
example, turn a Russian date into something Russians don't understand.
Treading lightly meant changing any occurrence of
yyyy but being extra
careful to only match exactly two 'y's, not four. Everybody Stand Back. I
know regular expressions.
After much fiddling – because who among us can really write a regex correctly on
the first try? – I ended up with
preg_replace("/(?<!y)yy(?!y)/", "yyyy", $pattern). It seems to work.
And I ran the test again.
en_US: 10/11/2012en_GB: 11/10/2012de_DE: 11.10.2012fr_FR: 11/10/2012ru_RU: 11.10.2012en_ZW: 11/10/2012ps_AF: م. ۲۰۱۲/۱۰/۱۱
Hooray! Localized dates with four-digit years. Exactly the details I need.