Support updating records by id via :on_duplicate_key_update (#187)

[Fivell]

Closes #187.

Context

Issue #187 asked for a way to update existing records based on an id column during import, instead of only inserting. This is already achievable through activerecord-import's upsert support (:on_duplicate_key_update), but the behaviour was untested and the README only mentioned it in passing ("For databases that support upserts you can use :on_duplicate_key_update instead.").

This PR proves it works and documents it properly.

What's here

• Spec (spec/import_spec.rb): a new context under authors already exist that imports authors_with_ids.csv with

  active_admin_import validate: false,
                      on_duplicate_key_update: { conflict_target: [:id], columns: %i[name last_name birthday] }

  and asserts that a single import:

  • updates the colliding row (id 1 — stale birthday replaced, no duplicate created),
  • inserts the non-colliding row (id 2),
  • reports Successfully imported 2 authors.

• README: replaced the one-line note with a concrete example and the two caveats that bite in practice:

  • conflict_target must name the unique column(s); MySQL infers and ignores it.
  • validate must be off for id-based upserts — activerecord-import runs uniqueness validations against the very rows the upsert is about to overwrite, so a model-level validates_uniqueness_of would reject the update (verified: with validate: true the existing-row update is dropped with "Last name has already been taken").
  • Active Record callbacks are not fired for bulk imports.

Verification

bundle exec rspec spec/import_spec.rb → 52 examples, 0 failures (3 new). Run locally on SQLite 3.53 / Rails 7.1 / Active Admin 3.2.