RefinerDB

A browser database used to support advanced search scenarios. An engine that could power an Amazon Refiner-like search experience.

• Install
• Quick Start
• Create a database
• Add items to the database
• Creating indexes
• Querying the database
  • Exact Equals
  • String contains
  • Min Max
  • Sorting
  • Paging
• Displaying query results
• Further Docs

Install

npm install refinerdb

Quick Start

import { RefinerDB, IndexType } from "refinerdb";

// Create an instance of RefinerDB and register the indexes
let refinerDB = new RefinerDB("movies-db");
refinerDB.setIndexes([
  { key: "title", type: IndexType.String },
  // An index where the key doesn't match the property name on the item
  { key: "genre", type: IndexType.String, path: "genres" },
  // A date index
  { key: "released", type: IndexType.Date, path: "released_date" },
  // A number index
  { key: "score", type: IndexType.Number, path: "vote_average" },
  // A nested 'path' is supported as well
  { key: "director", type: IndexType.String, path: "director.name" },
]);

// Add data
let movies = await fetchMovies();
refinerDB.setItems(movies);

// Query Data
// A 7+ rated 'Action' or 'Comedy' with 'day' in the title
let filter = {
  title: "day*",
  genre: ["Action", "Comedy"],
  score: { min: 7 },
};
let { items, refiners, totalCount } = await refinerDB.query({
  filter,
  sort: "released",
  sortDir: "desc",
});

// items will be an array of movies matching the criteria
// refiners will be an object of like
// {
//     "genre": [
//         { key: "Action", count: 13 },
//         { key: "Comedy", count: 21 },
//         { key: "Drama", count: 42 }
//     ],
//     ...
// }

Create a database

import { RefinerDB } from "refinerdb";

// OPTION 1: Just pass a database name
let refinerDB = new RefinerDB("my-db");

// OPTION 2: Pass a config with any or all of the following options
let dbConfig = {
    // How long to wait to reindex after data has changed
    indexDelay: 500,
    // A handler to fire when the State Machine transitions state
    onTransition: (newState) => console.log("Refiner DB State: " + newState);
}
let refinerDB = new RefinerDB("my-db", dbConfig);

Add items to the database

Once you have a database instance you can seed it with an array of items.

let refinerDB = new RefinerDB("movies");
let movies = await getMovies();
refinerDB.setItems(movies);

• The items can be in any shape. But each item needs to have a property that represents the primary key.
  • The primary key can be an string or a number
• By default RefinerDB assumes that property is id
• When creating a database, you can explicitely tell RefinerDB what the primary key is using the idProperty setting.

let refinerDB = new RefinerDB("bookmarks", { idProperty: "__id" });

Creating indexes

What is the shape of your data? For any properties you want to filter/refine, you need to define an Index.

Imagine we had data that looked like this:

let items = [
  { title: "Watch the new Matrix", id: 1, tags: ["fun"] },
  { title: "Cut the grass", id: 2, ["chore", "lawn"] },
  { title: "Buy fetilizer", id: 4, tags: ["lawn", "shopping"] },
  { title: "Build RefinerDB", id: 3, ["fun", "code] },
  { title: "Document RefinerDB", id: 11, ["chore", "code"] },
];

We would create a RefinerDB instance then define two indexes, one for title, and one for tags. We won't need to filter or sort by id, so we dont' need an Index for it.

import { RefinerDB, IndexType } from "refinerdb";

let refinerDB = new RefinerDB("my-db");
let indexDefinitions = [
  { key: "tags", type: IndexType.String },
  // Because all of the titles are generally unique, generating a list of
  // refiner options and their counts is not that useful. We can improve
  // performance by skipping that step.
  { key: "title", type: IndexType.String, skipRefinerOptions: true },
];
refinerDB.setIndexes(indexDefinitions);

Here are the available IndexType options, as well as the full type definition of IndexConfig.

enum IndexType {
  String = "string",
  Number = "number",
  Date = "Date",
}

interface IndexConfig {
  /**
   * Unique identifier for the index.
   * If no "path" is provided, it is assumed the "key" matches the
   * property name on the item.
   */
  key: string;

  /**
   * String, Number, Date, etc...
   */
  type: IndexType;

  /**
   * If the key doesn't match the item's property name, use path.
   * Allows nested paths like "author.name"
   */
  path?: string;

  /**
   * For things like dates with timestamps, or really long strings,
   * do you really need to calculate refiner option?
   */
  skipRefinerOptions?: boolean;

  /**
   * A convenience property so you can build dynammic controls like
   * a "Sort Dropwdown"
   */
  label?: string;
}

Querying the database

To query, create a filter object, where each each property on the object maps to a registered Index key.

In the following example, we are looking for movies where:

• genre is "Action" or "Comedy"
• title includes "day"
• score is greater or equal to 6

// Each property on the filter should match a registered index key
let filter = {
  genre: ["Action", "Comedy"],
  title: "day*",
  score: { min: 6 },
};

// Pass the filter to the `query(criteria)` method
let { items, refiners } = await refinerDB.query({ filter });

Exact Equals

// Find Action movies
let filter = { genre: "Action" };

Pass multiple values and they will be OR'd together.

// Find movies with a genre of Action or Comedy
let filter = { genre: ["Action", "Comedy"] };

If you add multiple indexes to the filter, they will be AND'd together.

// Find all Action or Horror movies rated PG-13
let filter = { genre: ["Action", "Horror"], mpaa: "PG-13" };

Number Indexes work the same way

// Find anything with a score of exactly 7.1
let filter = { score: 7.1 };

See below for the min/max filter that may be more useful for Number indexes

String contains

You can query IndexType.String indexes using a 'contains' by including an * (asterisk) in your filter value.

// Put an asterisk at the end to do a "contains" query

// Find any movies with 'day' anywhere in the title
let filter = { title: "day*" };

// Find any movies with 'day' or 'night' anywhere in the title
let filter = { title: ["day*", "night*"] };

Min Max

You can query IndexType.Number and IndexType.Date indexes using a min/max range.

// Find all movies with a score greater or equal to 7
let filter = {
  score: { min: 7 },
};

// Find all movies released in 2012 with score less than 5
let filter = {
  released: {
    min: new Date("1/1/2012"),
    max: new Date("12/31/2012"),
  },
  score: { max: 5 },
};

Sorting

TODO: describe sort and sortDir

Paging

TODO: describe limit and skip

Displaying query results

TODO: describe shape of QueryResult

Further Docs

• Persisted Stores