liteorm

A simple wrapper for sqlite; with typings based on TypeScript decorators and reflect-metadata.

• Async eventemitter (emittery)
  • I make sure that you can intercept query objects and raw SQL (as well as their parameters) in an async way
• ~~Auto-define _id as PRIMARY KEY INTEGER AUTOINCREMENT (Use _id as default name for primary key)~~
  • I use ROWID, instead.
• Auto-append createdAt, updatedAt if @Table({ timestamp: true })
• JSON, Date, Boolean, and MongoDB interop
• Additional type StringArray, inspired by Anki schema
• Query with JSON, and tested with https://q2search.herokuapp.com/LiteORM, using MongoDB-like languages, with some differences (for example, $regex is currently not supported, use $like, $nlike, $substr, $nsubstr instead.)
• JSON querying is supported via JSON1 extension. I made it easy to query using dot notation, just like MongoDB.
  • So, you can use data.a
• Multiple SQLite databases, with cloned schemas or different schemas. Strongly-typed in the IDE.

Usage

Please see /tests/suites and ankisync.js

Installation

npm i liteorm
# or yarn add liteorm

Caveats

• Type Number by default is associated with REAL. To change it to INTEGER, use

@prop({type: 'int'}) count!: number;

• BLOB is associated with Type ArrayBuffer.

@prop() data!: ArrayBuffer;

• To get a strongly-typed default / onUpdate, you might have to declare typing twice.

@prop<Record<string, string>>({ default: () => ({}) }) data!: Record<string, string>;
@prop<number, EntryClass>({ default: 1, onUpdate: (ent) => parseToInt(ent) }) order!: number;

• You might have to declare your own interface to get keys for createdAt, updatedAt, because typing is based directly on Class.