cpoms / arbor-rb Goto Github PK

https://developers-portal.arbor.sc/pages/graphql

The uber-REST nature of the API means that to extract data in bulk is a bit unwieldy. This is why the API developers have a changelog method, so you can see which resources have changed since the revision id at the time you last synced (if that's how you're utilising the API)

This is far more efficient than the "snapshot" technique of pulling out everything and working out what's changed at the client end.

The API supports this by appending /changelog to the usual resource endpoints, and there's a from-revision query parameter to instruct the API from which revision ID to return changes.

At the moment it just tries to parse the empty response body, ignoring the 401 response code, making it difficult to tell that it is an authentication error.

Should handle different response codes accordingly, before looking at the response body.

Replace with load/dump as is the standard naming for such operations in Ruby libraries.

https://github.com/meritec/arbor-rb/blob/842d9a3175f8a7a4a57e7007ed7423a1b1aa2c1f/lib/arbor/utils.rb#L37

There are certain resources, e.g. uk-dfe_upn-assignments, which don't conform to the convention of dasherized resource names used for all other resources (the underscore), this could cause a problem when we convert from underscore to dasherize & camelcase, etc.

Also need to add this new resource type.

we assumed (fairly) that all records have a href. when they don't, however, the call to href hits the method_missing, which tries to load the full resource, which calls href .. and the loop begins.

if we add href to attr_accessor it won't hit method missing.