PixieDB

A tiny in-memory javascript database with indexing and SQL like filters.

Features

• Speed - PixieDb perform get in the O(1) and other all operations (insert, delete, select*) in log(n) time. Can perform get operation with unique index (18M ops/s) and binary-index (5M ops/s) which is 15-20 times faster than lokijs/lokidb.
• Quick load - loads data 20x faster than lokijs/lokidb.
• Realtime filtering - perform filtering and event calling in realtime.
• Memory efficient - use iterators and Binary indexes (red black tree) for indexes to perform filtering.
• Events - notifies you about load, insert, update, delete, and quit to sync your state with the database.
• Indexes = for fast filtering.
• Chaining = supports filter chaining.

[!WARNING] Please keep in mind that PixieDb is still in under active development.

Usage

1import { PixieDb } from "pixiedb";
2
3const products = [
4    { id: 1, name: "Apple", price: 5, category: "Fruit" },
5    { id: 2, name: "Banana", price: 10, category: "Fruit" },
6    { id: 3, name: "Grapes", price: 6, category: "Fruit" },
7    { id: 4, name: "Orange", price: 8, category: "Fruit" },
8    { id: 5, name: "Potato", price: 18, category: "Vegetable" },
9    { id: 6, name: "Milk", price: 7, category: "Dairy" },
10    // ...
11]
12
13// provide unique key, data and indexes for better performance
14// 3rd param data is optional, Can be loaded after using the load method
15const pd = new PixieDb('id', ["price", "category"], products) 
16// or
17const pd = new PixieDb<Product>('id', ["price", "category"]) // pass type if using typescript
18pd.load(products) // to load data later
19
20const byId = pd.select().eq("id", 2).single()
21console.log(byId); // { id: 2, name: "Banana", price: 10, category: "Fruit" }
22
23// can also pass an array of fields to select method to pick only those fields/properties
24const fruitBelow10 = pd.select(["id", "name", "price"]).eq("category", "Fruit").lte("price", 10).orderBy(["name", ["price", "desc"]]).range(2, 3).data()
25console.log(fruitBelow10); // [{ id: 3, name: "Grapes", price: 6 }, ...]
26
27const updatedBanana = pd.where().eq("name", "Banana").update({price: 100})
28// [{ id: 2, name: "Banana", price: 100, category: "Fruit" }, ...]
29
30// delete all docs where name equals "Apple"
31const deletedApples = pd.where().eq("name", "Apple").delete()
32// [{ id: 1, name: "Apple", price: 5, category: "Fruit"}, ...]

Installation

1# using npm
2npm install pixiedb
3
4# using pnpm
5pnpm add pixiedb
6
7# using yarn
8yarn add pixiedb
9
10# using bun
11bun add pixiedb

Docs

PixieDb

This is a class which creates an PixieDb instance to use.

1// pass type/interface if using typescript
2const pd = new PixieDb<Product>('id', ["price", "category"]) 
3
4// or with data
5const pd = new PixieDb<Product>('id', ["price", "category"], products)

Methods

load

Used to import data without cloning (so don't mutate the data or clone before load). Pass true as second parameter to clear the previous data and indexes state. (Default: false).

1pd.load(products)
2// or
3pd.load(products, true)
4// remove previous data and index state

get

Get single doc/row using key (primary key/unique id). Returns doc/row, if present else undefined.

1pd.get(2)
2// { id: 2, name: "Banana", price: 10, category: "Fruit" }

select

Get single doc/row using key (primary key/unique id). Returns doc/row, if present else undefined.

1pd.select().eq("category", "Fruit").gte("price", 6).data()
2// [{ id: 2, name: "Banana", price: 10, category: "Fruit" }, { id: 3, name: "Grapes", price: 6, category: "Fruit" }, ...]
3
4pd.select(["id", "name", "price"]).eq("category", "Fruit").lte("price", 6).data()
5// [{ id: 1, name: "Apple", price: 5 }, ...]
6
7pd.select().eq("category", "Fruit").between("price", [6, 10]).data()
8// [{ id: 2, name: "Banana", price: 10, category: "Fruit" }, { id: 3, name: "Grapes", price: 6, category: "Fruit" }, { id: 4, name: "Orange", price: 8, category: "Fruit" }, ...]

where

used to perform delete/update with complex filtering

1// this will delete and return all the docs according to the filters
2pd.where().eq("category", "Fruit").gte("price", 6).delete()
3// [{ id: 2, name: "Banana", price: 10, category: "Fruit" }, { id: 3, name: "Grapes", price: 6, category: "Fruit" }, ...]
4
5pd.where().eq("category", "Fruit").between("price", [6, 10]).update({price: 11})
6// [{ id: 2, name: "Banana", price: 11, category: "Fruit" }, { id: 3, name: "Grapes", price: 11, category: "Fruit" }, { id: 4, name: "Orange", price: 11, category: "Fruit" }, ...]

data

Get all docs/rows ordered respecting to primary key/unique id. Pass false to get all without clone (don't modify). Default: true

1pd.data()
2// [{ id: 1, name: "Apple", price: 5, category: "Fruit" }, ...]

count

Get all docs/rows ordered respecting to primary key/unique id. Pass false to get all without clone (don't modify). Default: true

1pd.select().count()
2// 6
3
4pd.select().eq("category", "Fruit").between("price", [6, 10]).count()
5// 4

close

to close/quit/terminate the database and remove all data/indexes and fire "Q" ("quit") event. Pass true to not emit events. Default: false

1pd.close()
2// or
3pd.close(true) // doesn't fire event

toJson

return JSON of all data (without cloning), key and index names.

1pd.toJSON()
2// { key: "id", indexes: ["price", "category", {name: "id", unique: true}], data: [{ id: 1, name: "Apple", price: 10, category: "Fruit" }, ...]
3
4// this will call the above toJSON method
5JSON.stringify(pd)

view more

Roadmap

• load docs
• get all docs
• get docs with key
• Events (load, change, insert, update, delete, quit)
• orderBy with multiple keys (sorting)
• single doc with filters
• count of docs with filters
• update of docs with filters
• delete of docs with filters
• Plugin support
• Unique indexes (currently override the previous)
• filters
  • eq (equal)
  • neq (not equal)
  • in (value in)
  • nin (value not in)
  • between - values within a given range (>= and <=). begin and end values are included.
  • nbetween - values not within a given range (< or >). begin and end values are not included.
  • gt (greater than)
  • gte (greater than or equal to)
  • lt (less than)
  • lte (less than or equal to)
  • custom query method
• range offset (from) and count (limit of docs to return)
• multiple tables
• joins
• changes api
• custom clone method
• custom compare method
• views
  • Basic views
  • Materialized views (persist)
• plugins
  • persist (localStorage, indexedb)
  • sync with other databases
  • sync with browser tabs
  • transaction

Other Details

query filters that use binary index (perform operation in log(n))

• eq
• in
• between log(n) + count of docs between
• gt log(n) + count of docs
• gte log(n) + count of docs
• lt log(n) + count of docs
• lte log(n) + count of docs

other query filters

• neq O(n)
• nin O(n)
• nbetween log(n) + count of docs (where value less than or equal to)