Description

Using Knex to run migrations, while keeping Prisma as the source of truth? This script lets you use Prisma to build a starting SQL script, when creating knex migrations.

Setup and Usage

You will probably want to run this other script I made first, to set everything up: prisma-migration-migrator. This will set up Knex on an existing Prisma project. Otherwise, Prisma might output SQL related to the knex migrations history table, which you probably don't want.

Assuming you have Prisma and Knex set up, just add this package to your project to start using it.

1. First, install the dev-only dependency:

   npm install -D prisma-diff-to-knex-migration

2. Then, when you want to change something in your DB (like adding a field or something), update your Prisma Schema file first!

3. Then, pipe this script to your knex migration creation script:

   npx knex migrate:make 'example_migration_name' | npx prisma-diff-to-knex

   The script will replace the starting content of the migration file with the SQL generated by Prisma.

   This script will never run the migration, it will only generate the SQL for it. You will still need to run the migration with npx knex migrate:latest.

How does it works?

The script picks up the path of the migration from the knex migration creation script, and then generates a SQL migration based on the difference between the Prisma datamodel (the schema), and the datasource (your DB).

This means that the SQL generated will depend on the current state of your Prisma Schema file, and the current state of your database. If you have a migration pending, it will generate the SQL for that migration. If you have no pending migrations, it will generate an empty SQL file.

The SQL comes from running npx prisma migrate diff --from-schema-datasource ${schemaPath} --to-schema-datamodel ${schemaPath} --script, where ${schemaPath} is the path to the Prisma schema. Then it picks up the SQL sentences, and replaces the content of the migration file with them.