Uses Sequelize.sync()
but generates ALTER statements instead of DROP+CREATE.
Only works for postgres right now. Please open an issue if you need support for other databases.
sequelize-sync-diff
syncs your schema with your database using sequelize.sync({ force: false })
and you need to provide a dummy database where it will recreate the same schema you have and will call .sync()
but this time with { force: true }
. So at this point you have your working database not fully up to date since sync()
has its limitations but you have a dummy empty database but with the full schema your data models need. Then, internally, https://github.com/gimenete/dbdiff is used to compare both databases and you get the SQL required to execute in your working database to be up to date with your data model changes.
npm install sequelize-sync-diff
// Before defining anything in your data model you need this code:
var Sequelize = require('sequelize')
require('sequelize-sync-diff')(Sequelize)
// At this point Sequelize has the new syncDiff() method injected
// and some code has been injected for introspecting the data model
// Use Sequelize as you always do
var sequelize = new Sequelize('postgres://user:pass@localhost/database', {})
var User = sequelize.define('User', {/* ... */})
var Project = sequelize.define('Project', {/* ... */})
Project.hasOne(User)
// Now is time to use .syncDiff()
sequelize.sync()
.then(() => sequelize.syncDiff('postgres://aro:aro@localhost/database_dummy'))
.then(function(sql) {
// Here we have the ALTER statemetns
console.log(sql)
})
The SQL statements are returned to you and they are not executed for a good reason: there are scenarios on which this will cause data loss (e.g. dropping a column). Also some statements could fail depending on the data you have in your working database (e.g. changing the data type of a column or setting it to NOT NULL). For more information please check dbdiff caveats.