Simple persistent map and array datastructures built on top of better-sqlite3.
Sometimes we just need a simple way to persist data. Maybe we are performing expensive operations and want to cache the results while exploring a problem. With better-sqlite3 we can store everything in a single file while working with just Map/Array type datastructures.
pnpm i better-sqlite3 sqlite-map
pnpm i -D @types/better-sqlite3
import Database from "better-sqlite3";
import { createBetterSqlite3Factory } from "sqlite-map";
const database = new Database("db.sqlite");
const factory = createBetterSqlite3Factory({
database,
});
type Message = {
id: number;
text: string;
};
const messages = factory.getMap<Message>("messages");
messages.set("message1", {
id: 1,
text: "Hello, World!",
});
messages.set("message2", {
id: 2,
text: "Hello, World!",
});
console.log([...messages.entries()]);
Simplest way to store object that may contain UInt8Array data is to use a serializer such as msgpack.
import Database from "better-sqlite3";
import { createBetterSqlite3Factory } from "sqlite-map";
import { encode, decode } from "@msgpack/msgpack";
const database = new Database("mydb.sqlite");
const factory = createBetterSqlite3Factory({
database,
});
type BinaryData = {
id: number;
bytes: Uint8Array;
};
const data = factory.getMap<BinaryData>("data", {
format: "binary",
serializer: {
encode,
decode,
},
});
data.set("message1", {
id: 1,
bytes: new Uint8Array([1, 2, 3]),
});
console.log([...data.entries()]);
const factory = createBetterSqlite3Factory({ database });
The factory instance can be used to create multiple maps/arrays. Each map/array has a different name and represents a sqlite table in the database.
import Database from "better-sqlite3";
import { createBetterSqlite3Factory } from "sqlite-map";
const database = new Database("db.sqlite");
const factory = createBetterSqlite3Factory({
database,
});
const messages = factory.getArray<string>(
"messages", // name of the sqlite table
"", // default value that will be used when performing set(index, value) operations that could resize the array creating holes
);
messages.push("value1");
console.log([...messages]);
The library also provides simple ways to work with related data (however these require full table scan to find/delete related data). We can pass a third parameter to the set function to associate related ids. The set()
call will create the corresponding columns in sqlite table to hold these related values (internally stored as comma separated values).
Later we can get all rows that have any of the related id's by using getRelated()
. getRelated/deleteRelated
are only convenience features and have performance issue because they perform a complete table scan to filter rows.
const database = new Database(":memory:");
const factory = createBetterSqlite3Factory({
database,
});
type User = {
id: string;
};
type Post = {
content: string;
};
const users = factory.getMap<User>("users");
/**
* By providing the second generic paramter ["user_id"] we can get
* typesafety/suggestions when trying to associate related ids
*/
const posts = factory.getMap<Post, ["user_id"]>("posts");
users.set("user1", { id: "user1" });
// we can associate one or more user_ids with the post
posts.set("post1", { content: "post1" }, { user_id: ["user1"] });
posts.set("post2", { content: "post2" }, { user_id: ["user1"] });
// we can get all posts that match any of the user_ids
console.log([...posts.getRelated({ user_id: ["user1"] })]);