Nova Duplicate Field Laravel Package

Getting Started

This package adds a replication layer on top of Laravel models so you can duplicate a record (and optionally its relations) without hand-writing replicate() + fill() boilerplate. v0.3.0 adds per-field control — you can now ignore fields (never copied) or override them (copied but replaced with a computed value) during replication.

Minimal setup:

composer require <vendor>/<package>
php artisan vendor:publish --tag=replication-config

Then replicate your first record:

use Vendor\Package\Facades\Replicator;

$copy = Replicator::make($product)
    ->ignore(['sku', 'slug'])
    ->override(['status' => 'draft', 'created_by' => auth()->id()])
    ->save();

Where to look first, in order:

1. config/replication.php — global ignore/override defaults and the model resolver. This is where 90% of "why was this field copied?" questions get answered.
2. The Replicator facade / service contract — the fluent API (make, ignore, override, withRelations, save).
3. Any Replicates trait or replicationConfig() method you can add to models — per-model defaults live here, not in config.
4. The package tests — the fastest way to see how nested relations and casted attributes are handled, which the README tends to undersell.

First real use case: "duplicate this product/listing/page as a draft". You want a new row with a fresh ID, a handful of identity fields (sku) stripped, and ownership/status reset — exactly the ignore/override split.

Implementation Patterns

Ignore vs. override — pick deliberately. ignore() removes a field entirely (the DB default or null applies). override() keeps the column in the payload but swaps the value. For anything with a NOT NULL + no-default column, use override, not ignore, or the insert fails.

Identity fields belong in ignore, ownership in override:

$copy = Replicator::make($page)
    ->ignore(['slug', 'uuid', 'published_at'])      // regenerated downstream
    ->override(['status' => 'draft', 'author_id' => auth()->id()])
    ->save();

$copy->update(['slug' => Str::slug($copy->title).'-'.Str::random(6)]);

Per-model defaults via a trait/config method so controllers stay thin:

class Product extends Model
{
    use Replicates;

    public function replicationConfig(): array
    {
        return [
            'ignore'   => ['sku', 'slug'],
            'override' => ['status' => 'draft'],
        ];
    }
}

Call-site overrides should merge on top of (not replace) model defaults — verify this in your version, since it's the most common source of surprise.

Controller + form request pattern. Validate the user-supplied overrides only, then hand them to the replicator. Don't let request input reach the override map unfiltered — that's a mass-assignment hole with a different name.

Bulk replication. Wrap in a job and chunk; the fluent builder is per-record. For "duplicate this template into 50 tenants", loop the builder inside DB::transaction() per record rather than one giant transaction.

Relations. If the package supports withRelations(), replicate children after the parent saves so you can point foreign_id at the new ID via override. Do not assume children()->replicate() cascades — it doesn't in most implementations.

Testing. Assert the negative space: assertNull($copy->sku), assertNotEquals($original->uuid, $copy->uuid), assertSame('draft', $copy->status). Snapshot-style assertions on the whole attribute array break on every schema change.

Gotchas and Tips

• Ignoring a unique column doesn't generate a new value. You still need a slug/SKU strategy. The feature removes the copy, not the obligation.
• ignore() on a column with no DB default throws on insert. Prefer override when in doubt; check SHOW CREATE TABLE before trusting an ignore list.
• Timestamps and soft deletes. Confirm whether created_at/updated_at are excluded by default. If you ignore deleted_at but copy deleted_at's value implicitly, a soft-deleted original can produce a soft-deleted copy — usually not what you want.
• Observers and model events fire on the replica. creating/created/saved hooks (audit logs, search indexing, slug generation) run against the new record. If the package's own save bypasses Eloquent, your observers won't fire — test both directions and pick one behaviour intentionally.
• Casts and JSON columns are copied by value from the hydrated model, so array/json casts round-trip cleanly, but Carbon instances are shared objects unless cloned — mutate the copy's dates carefully.
• $fillable/$guarded still apply if the package uses fill(). An override for a non-fillable column silently disappears. Use forceFill semantics or add the field to $fillable.
• Config caching. php artisan config:cache in deploy means edits to config/replication.php won't take effect until re-cached. Classic "works locally, not in prod" for this class of package.
• Debugging the resolved payload. Before save(), dump the resolved ignore/override arrays and the final attribute set. When a field "mysteriously" copies, it's almost always precedence: config default < model replicationConfig() < call-site override. Know which layer won.
• Extension points to reach for, in order: model-level replicationConfig(), then call-site ignore/override, then a custom field resolver or macro if the package exposes one. Avoid forking the replicator for one quirky model.
• Version pin. v0.3.0 is pre-1.0; the fluent API surface is still moving. Pin the minor version and re-read the changelog on bumps — the ignore/override precedence rules are exactly the kind of thing that shifts between 0.x releases.