13 April, 2015
This is part five of the “generating this website” series. To read the rest of the series, go to the series index here
On February 1st, 2015, I decided to change the URL scheme for this site, such that instead of outputting pages to post-name.html it would output to post-name/index.html. Unfortunately, this breaks any existing links to the original pages that might be out there. This module keeps those links active by generating any posts prior to that date at the original URL.
Imports
Let’s get the obvious out the way…
> {-# LANGUAGE UnicodeSyntax #-}
|
> {-# LANGUAGE OverloadedStrings #-}
|
> module OutdatedURLs where
|
|
|
We’re going to need a couple of Maybe-related utilities.
> import Data.Maybe (fromMaybe, isJust)
|
We’ll also import Posts, since for the most part we want to generate these posts exactly like their correctly-addressed counterparts.
Finally, we import Data.Time so that we can check the dates. Note we hide readTime because we’re going to use the version from the Posts module.
> import Data.Time hiding (readTime)
|
Setting up the rules
First we define the cut-off date, before which an old-style .html page will be created. This applies to any posts made prior to midnight, February 1st, 2015.
|
(fromGregorian 2015 02 01)
|
|
|
The Rules managing these posts are similar to those for standard Posts. We make use of the recently-added matchMetadata to compare the date field of the posts with our cutoffDate, and we tag the rule with the version “outdated” to avoid them showing up in the index. And of course we use setExtension rather than simplifyURL, since that’s the whole point of the exercise!
> outdatedURLs :: Tags → Rules ()
|
|
> matchMetadata "posts/*" isOutdated ∘ version "outdated" $
|
|
|
|
metadataRoute dateAndTitle `composeRoutes`
|
|
|
|
|
|
|
|
|
|
= maybe False checkDate ∘ lookupString "date"
|
|
= (> 0) ∘ diffUTCTime cutoffDate ∘ readTime
|
That’s it!
With just a small amount of effort, we’ve managed to completely change our URL scheme without breaking any old links. I wouldn’t want to make a habit of this sort of thing, but it’s good to know it can be resolved fairly easily should the need arise!
…or is it?
Well, it was, until in August 2017 I found myself wanting to make another change that would invalidate some old URLs. This time, to try and give myself a modicum of future-proofing, I’m going to encode the old URL I want to support into the metadata for the post.
> clonedURLs :: Tags -> Rules ()
|
|
> matchMetadata "posts/*" hasCloneField ∘ version "cloned" $
|
|
|
$ metadataRoute (fromMaybe mempty ∘ cloneURL)
|
|
|
|
|
= isJust ∘ lookupString "clone"
|
|
= constRoute <$> lookupString "clone" meta
|
This version is very similar to outdatedURLs above, but rather than try to be clever with the date field, it simply searches for any post containing a clone field. If it finds one, it outputs a cloned version to the location specified in that field.
Hopefully that should cover me from now on! If I need to clone into multiple locations in future, I’ll have to update this function to accept a comma-separated list or something, but it’ll do for now…
