Fork import

What it does

Fork Import lets you load a JSON database export (for example, a copy of production data, or a fork's data) and, one table at a time, compare it against this environment's database and import whichever rows you choose. It's a diff-and-import tool, not an all-or-nothing restore.

Where to find it

Go to Fork Import in the admin sidebar (/admin/fork-import). This page is admin-only.

When to use it

Use this to bring a dev or staging environment up to date with real data from another environment — for example, syncing a staging fork with a recent production export so you can test against realistic data, or pulling specific rows (a handful of shops, a category tree) from one export into this database without overwriting everything.

How it works

1. Load export file — choose a JSON export file. Its contents aren't displayed (the file can be large); only the list of tables and row counts are shown.
2. Select table — pick one table from the file (Product, Shop, Category, User, and so on) and click Compare to DB.
3. Compare result — the tool shows, for that table:
   • New rows (exist in the file, not in this DB)
   • Updated rows (exist in both, but differ)
   • Unchanged count
   • For auth tables (see below), a simpler "new vs. already in DB" count instead of new/updated
4. If any row references another record that doesn't exist in this DB yet (for example, a Product pointing at a Shop id this database doesn't have), a missing references panel lets you reassign each missing id to an existing record in this DB — or leave it unmapped, which skips (errors) those rows on import.
5. For Product and Service tables specifically, you can filter the rows shown/imported down to a single store.
6. Every new/updated row has a checkbox — uncheck any you don't want. By default, everything comparable is selected.
7. Click Import N selected rows to apply them.

Auth tables are create-only

User, Account, Session, and VerificationToken are treated specially: rows are mapped into this app's auth schema and only ever created, never updated. If a User/Account/Session/VerificationToken already exists in this DB, the import leaves it alone rather than overwriting it. Every other table supports both creating new rows and updating existing ones that changed.

Cautions

• This tool writes directly to this environment's database. Compare before you import, and use the missing-reference and store-filter controls to scope what actually gets applied.
• Unmapped missing references are imported anyway and will error per-row rather than being silently skipped — check the error list after an import (created, updated, and errors are all reported).
• Only run this against an environment you intend to overwrite/extend with the export's data — not against production, unless the export is production's own data and you know exactly what you're syncing.