Component URLs should separate the ead & ref slugs

[seanaery]

Summary

ArcLight currently creates URLs to individual components like this:
{host}/catalog/{ead_ssi}{ref_ssi}

So we get URLs like:
Collection: archives.institution.edu/catalog/slug <-- this is OK
Component: archives.institution.edu/catalog/slugaspace_123 <-- this is not ideal

Desired Change

This ticket is a request to improve the component URLs by using one of these conventions instead:
archives.institution.edu/catalog/slug/aspace_123
archives.institution.edu/catalog/slug_aspace_123

Reasons for concern over the smooshed together URL

• Sometimes creates awkward unintended words (in practice, ead slugs often end in letters and ref values often begin with al or aspace
• Harms SEO, i.e., makes it harder to Google using the EAD slug
• Can make web analytics reports on per-collection activity harder to parse
• Makes it harder to build future pattern-based redirects from these URLs to new ones
• Not very readable

Current ArcLight Usage

No separation in Component URL

These apps follow default core:

• Albany
• CAO
• EmpireADC
• Indiana

Has separation in Component URLs by wedging an underscore between the ead & ref slugs

These apps wedge an underscore between the ead & ref slugs (overriding core):

• Duke
• Michigan
• Princeton

Other Considerations

@gwiedeman & @jrochkind astutely note on code4lib#arclight Slack that if a change is made to core, it's important to be able to redirect from the old style to keep existing URLs working. And even if core stays how it is, I recommend making the convention configurable or otherwise straightforward to override globally.

[cbeer]

#1317 is an idea for a seam for downstream applications to decide what style they want 🤷‍♂️

[jrochkind]

Harms SEO, i.e., makes it harder to Google using the EAD slug

It's a bit of an esoteric art and I'm not necessarily an expert, but I believe that /slug/aspace_123 will be a lot more reliable than /slug_aspace_123 with regard to that.

Can make web analytics reports on per-collection activity harder to parse

Depending on the tool you're using, I think same.

If you're going to do the work to make a change (globally or with a "seam" in an individual app), I predict you'll be a lot happier over the long-term, and less likely to need to change it again, if you can go to /. Even if might require messing with Rails routing a bit more.

[seanaery]

@marlo-longley We are definitely still interested in prioritizing this work. In lieu of having a configurable way to do this, here's what we ended up having to modify in our local 1.0.1-based app:

• Override the to_field 'id' from ead2_component_config.rb to join the two parts with an underscore (commit). This takes care of most of the logic for a component URL, tbh, and wasn't too heavy a lift, but...
• Fix the breadcrumb links to parent components (commit). This was more of a pain and required several local overrides, esp. stemming from the use of app/models/arclight/parent.rb which includes a method global_id that assumes the component ID is "#{eadid}#{id}"

[marlo-longley]

Thank you @seanaery. In earlier comments it seemed like some people had a strong preference for splitting the URL into a new path instead of the underscore separation (so that'd be /slug/aspace_123 rather than /slug_aspace_123) -- that's mainly what I wanted to make sure to consider before moving on a solution.

Your code is really helpful and seems much for straightforward in terms of routing to achieve. I will try to work something with this approach.

[gwiedeman]

We would want to continue the current URLs in our local instance just for persistence, but I the default should probably be the slash (archives.institution.edu/catalog/slug/aspace_123). I think that is okay if its config that you could just change in catalog_controller or similar and its clearly stated in the release notes how to maintain older-style URLs.