Closed karenmajewicz closed 7 months ago
We want to make faceting configurable.
Okay... this is a lot harder than originally imagined. The values in that field are not normalized.
Examples:
Screenshot
Unfortunately, the field is just a multi-valued string, with no sanity checking/control.
Some steps to move ahead:
I normalized the records so that they are all single values of yyyy-mm-dd or blank. Our newer harvesting scripts automatically insert these values, so most of the records were already in that format. https://geo.btaa.org/admin/blazer/queries/43-accessioned-dates
Another thing to consider here is the idea we had to automatically tag imports with some kind of generated accession code. That is main purpose of these dates - to be able to differentiate between successive uploads.
The various date/tracking fields we use:
Internal Metadata:
created_at
: Generated by GBL Admin and tracking the 1st time an ID was created/uploadedupdated_a
t: Generated by GBL Admin and tracking whenever a record was changed, either through CSV upload or manual edit.Kithe Model:
b1g_dateAccessioned_sm
: Part of the B1G profile of Aardvark for keeping track of harvest cycles. This reflects batches/uploads of a harvest/reharvest. It may or may not match created_at
or updated_at
.Adding to Advanced Search: https://github.com/geobtaa/geoportal/issues/552
When we reharvest from data portals, we create a fresh list of current datasets. We then upload them to GBL Admin and publish them.
Then, we then need to isolate the the outdated records and unpublish them. To do so, we check the "Date Accessioned" field. It would be easier to find and unpublish them if the Date Accessioned field were a facet right under the Publication State.