Master Quest SQLite3

Master Quest is a brave new attempt at Data Mapping.

It kinda looks like an ORM, but it isn't. It's not SQL, it's NoSQL with benefits.

You get to choose which contraints to keep and which to forget.

Guiding Principles

• NoSQL (well, as far as you care)
• Migrations don't suck
• Harness the awesomeness of indexes
• deletedAt to delete a record
• null to delete a field
• unindexed fields stored as json
• ids should be one either
  • deterministic (duplicate records not okay, last updatedAt wins)
  • cryptographically random (duplicate records consolidated in application)
• avoid mutating data (CRDT-ish style)
• if it won't scale, don't fret it (no foreign key contraints)
• id is simply named id
• multi-key ids should be sha256sum(key1 + key2 + key3)
• JavaScript is camelCasey, databases are snake_casey. We can handle that.
• join tables are in alphabet order i.e. foo, bar, bar_foo

TODO / In Progress

• Multi-Master Replication
• Relationships
  • currently detaches before saving (most important)
• MongoDB / RethinkDB -ish queries
• RealTime

USAGE

npm install --save 'https://github.com/coolaj86/node-masterquest-sqlite3.git'

'use strict';

// works with sqlite3, sqlcipher, and sqlite3-cluster
var db = new (require('sqlite3').Database)('/tmp/data.sqlite3');

require('masterquest-sqlite3').wrap(db, {
  modelname: 'Persons'
, indices: [ 'firstName', 'lastName' ]
, hasMany: [ 'children' ]
}).then(function (mq) {

  // update (or create) deterministic record
  var john = {
    id: '[e-mail]'
  , firstName: 'john'
  , lastName: 'doe'
  , dog: { name: 'ralph', color: 'gold' }
  , children: [ '[e-mail]' ]
  };

  mq.Persons.upsert(john.id, john).then(function () {
    // note: if `dog` existed, it will be overwritten, not merged
    // note: `children` will be removed before save

    mq.Persons.get('[e-mail]').then(function (data) {
      // dog will be rehydrated from json
      // children will not be fetched and attached
      console.log(data);
    });

  });

});

API

It's kinda CRUDdy... but don't let that scare you.

• upsert(id, data[, oldId]) - creates or updates based on existence in DB (use this)
  • modifies createdAt and or updatedAt
• create(id, obj) - same as above, but fails if the object exists
• save(data[, oldId]) - (just don't use this, please) creates or updates based on presence of ID
• destroy(id) - mark a record as deletedAt from DB
• get(id) - grab one by id
• find(attrs, opts) - grab many by indexable attributes
  • attrs
    • explicit null will find all (and requires that limit be set)
    • { foo: 2, bar: 6 } will find records where foo is 2 and bar is 6
  • opts
    • orderBy
    • orderByDesc
    • limit

Schema

Anything that isn't in the schema

• indices specifies an array of strings
  • [ 'firstName', 'lastName' ]
• relationships are option and current only exclude during save
  • hasMany, belongsTo, hasOne, belongsToMany, hasAndBelongsToMany
• createdAt, updatedAt, deletedAt timestamps are always added
  • turn off with timestamps: false
• id is always id
  • change with idname: 'myId'

Migrations

You can only add indexes. You cannot rename or remove them.

To add an index, simply change the schema.

{ modelname: 'persons'
, indices: [ 'firstName', 'lastName' ]
, hasMany: [ 'children' ]
}

LICENSE

Dual-licensed MIT and Apache-2.0

See LICENSE