Skip to content

Write DB migrations with SQL and run them with a CLI

License

Notifications You must be signed in to change notification settings

igorkamyshev/trona

Repository files navigation

trona

This library allows you to write migration scenarios using SQL and run them with a simple CLI.

  • Simple: only plain SQL
  • Lightweight: only 5.6 kB in node_modules
  • Modern: ESM support out of the box

Usage example

foo@bar:~$ yarn trona

Running evolve script

--- 1.sql ---

CREATE TABLE Customers (
    id   INTEGER     NOT NULL,
    name VARCHAR(32) NOT NULL,
    primary key (id)
);

Evolution is successful!

foo@bar:~$

Installation

First you need to install trona via package manager:

yarn add trona

Configuration

Then you need to setup simple configuration file named .trona-config.js and containing script that exports async function runQuery. The function should be rejected in case of failure of said query and in case of SELECT query successfully executed returns array of selected rows in form of an object {[field]: value}.

PostgreSQL example

import PG from 'pg';

const client = new PG.Client({
  // ...
});

await client.connect();
console.log(`Connected to database`);

export function runQuery(query) {
  return client.query(query).then((result) => result.rows);
}

MySQL example

import mysql from 'mysql';
import { promisify } from 'util';

const connection = mysql.createConnection({
  // ...
});

const connect = promisify(connection.connect).bind(connection);
const runQuery = promisify(connection.query).bind(connection);

await connect();
console.log(`Connected to database`);

export { runQuery };

Write evolutions

Create a folder evolutions for your evolutions script and add your first evolution to it. Note the rules which you should follow writing said evolutions:

  1. Name file {evolution-number}-{text}.sql. Text part -{text} is optional and can be omitted (e.g. both 1.sql and 1-create-table.sql are correct). Note: file name should always start with number. Evolution script file with any other symbol will be ignored (e.g. *1-incorrect.sql or -1-worng.sql) and warning will be shown during execution.
  2. Complement evolution file with fallback scripts. Separate evolution and fallback scripts with "#DOWN" comment as it show in example below.
  3. Execution will not be started in case of any conflict of numbering is found (e.g. existing any of two: 1-abba.sql, 1-baab.sql, 1.sql, 1-baba.sql)

Folder content example:

- evolutions
    - 1.sql
    - 2.sql
    - 3.sql
    ...

Evolution contents example:

CREATE TABLE Customers (
    id   INTEGER     NOT NULL,
    name VARCHAR(32) NOT NULL,
    primary key (id)
);

#DOWN

DROP TABLE Customers;

Run command

yarn trona

This command will create table with information about evolutions, if it doesn't exist. After it will execute all your evolutions.

Usage

After you managed to successfully setup trona you can run yarn trona command. This command will automatically detect any changed or new files in your evolutions folder, run respected fallback scripts if needed and than evolve your databae schema (e. g. if you have 1.sql, 2.sql, and 3.sql evolutions already in your database, you have changed 2.sql and added 4.sql it will run fallback for 3.sql and 2.sql and then run 2.sql, 3.sql, and 4.sql scripts)

Options

Interactivity

By default evolution script will ask for a confirmation to run a degrade script. You can disable this feature by -y or --no-interactive flag.

yarn trona -y

Config

You can change path to trona config file by providing -c or --config-path option (by default trona will try to find config file .trona-config.js in a root directory).

yarn trona -c config/.trona-my-config.js

Custom evolutions folder

Custom evolutions folder can be choosen by providing -d or --evolutions-dir option. By default evolutions folder is being used.

yarn trona -d migrations

Maintenance

Release flow

  1. Bump version in package.json
  2. Fill CHANGELOG.md
  3. Commit changes by git commin -m "Release X.X.X"
  4. Create git tag for release by git tag -a vX.X.X -m "vX.X.X"
  5. Push changes to remote by git push --follow-tags
  6. Release package to registry by yarn clean-publish
  7. Fill release page with changelog on GitHub