A rules engine expressed in JSON

Synopsis

json-rules-engine is a powerful, lightweight rules engine. Rules are composed of simple json structures, making them human readable and easy to persist.

Features

• Rules expressed in simple, easy to read JSON
• Full support for ALL and ANY boolean operators, including recursive nesting
• Fast by default, faster with configuration; priority levels and cache settings for fine tuning performance
• Secure; no use of eval()
• Isomorphic; runs in node and browser
• Lightweight & extendable; 24kb gzipped w/few dependencies

Installation

1$ npm install json-rules-engine

Basic Example

This example demonstrates an engine for detecting whether a basketball player has fouled out (a player who commits five personal fouls over the course of a 40-minute game, or six in a 48-minute game, fouls out).

1import { Engine } from 'json-rules-engine'
2
3/**
4 * Setup a new engine
5 */
6let engine = new Engine()
7
8// define a rule for detecting the player has exceeded foul limits.  Foul out any player who:
9// (has committed 5 fouls AND game is 40 minutes) OR (has committed 6 fouls AND game is 48 minutes)
10engine.addRule({
11  conditions: {
12    any: [{
13      all: [{
14        fact: 'gameDuration',
15        operator: 'equal',
16        value: 40
17      }, {
18        fact: 'personalFoulCount',
19        operator: 'greaterThanInclusive',
20        value: 5
21      }]
22    }, {
23      all: [{
24        fact: 'gameDuration',
25        operator: 'equal',
26        value: 48
27      }, {
28        fact: 'personalFoulCount',
29        operator: 'greaterThanInclusive',
30        value: 6
31      }]
32    }]
33  },
34  event: {  // define the event to fire when the conditions evaluate truthy
35    type: 'fouledOut',
36    params: {
37      message: 'Player has fouled out!'
38    }
39  }
40})
41
42/**
43 * Define facts the engine will use to evaluate the conditions above.
44 * Facts may also be loaded asynchronously at runtime; see the advanced example below
45 */
46let facts = {
47  personalFoulCount: 6,
48  gameDuration: 40
49}
50
51// Run the engine to evaluate
52engine
53  .run(facts)
54  .then(results => {
55    // 'results' is an object containing successful events, and an Almanac instance containing facts
56    results.events.map(event => console.log(event.params.message))
57  })
58
59/*
60 * Output:
61 *
62 * Player has fouled out!
63 */

This is available in the examples

Advanced Example

This example demonstates an engine for identifying employees who work for Microsoft and are taking Christmas day off.

This demonstrates an engine which uses asynchronous fact data. Fact information is loaded via API call during runtime, and the results are cached and recycled for all 3 conditions. It also demonstates use of the condition path feature to reference properties of objects returned by facts.

1import { Engine } from 'json-rules-engine'
2
3// example client for making asynchronous requests to an api, database, etc
4import apiClient from './account-api-client'
5
6/**
7 * Setup a new engine
8 */
9let engine = new Engine()
10
11/**
12 * Rule for identifying microsoft employees taking pto on christmas
13 *
14 * the account-information fact returns:
15 *  { company: 'XYZ', status: 'ABC', ptoDaysTaken: ['YYYY-MM-DD', 'YYYY-MM-DD'] }
16 */
17let microsoftRule = {
18  conditions: {
19    all: [{
20      fact: 'account-information',
21      operator: 'equal',
22      value: 'microsoft',
23      path: '.company' // access the 'company' property of "account-information"
24    }, {
25      fact: 'account-information',
26      operator: 'in',
27      value: ['active', 'paid-leave'], // 'status' can be active or paid-leave
28      path: '.status' // access the 'status' property of "account-information"
29    }, {
30      fact: 'account-information',
31      operator: 'contains', // the 'ptoDaysTaken' property (an array) must contain '2016-12-25'
32      value: '2016-12-25',
33      path: '.ptoDaysTaken' // access the 'ptoDaysTaken' property of "account-information"
34    }]
35  },
36  event: {
37    type: 'microsoft-christmas-pto',
38    params: {
39      message: 'current microsoft employee taking christmas day off'
40    }
41  }
42}
43engine.addRule(microsoftRule)
44
45/**
46 * 'account-information' fact executes an api call and retrieves account data, feeding the results
47 * into the engine.  The major advantage of this technique is that although there are THREE conditions
48 * requiring this data, only ONE api call is made.  This results in much more efficient runtime performance
49 * and fewer network requests.
50 */
51engine.addFact('account-information', function (params, almanac) {
52  console.log('loading account information...')
53  return almanac.factValue('accountId')
54    .then((accountId) => {
55      return apiClient.getAccountInformation(accountId)
56    })
57})
58
59// define fact(s) known at runtime
60let facts = { accountId: 'lincoln' }
61engine
62  .run(facts)
63  .then((results) => {
64    console.log(facts.accountId + ' is a ' + results.events.map(event => event.params.message))
65  })
66  .catch(err => console.log(err.stack))
67
68/*
69 * OUTPUT:
70 *
71 * loading account information... // <-- API call is made ONCE and results recycled for all 3 conditions
72 * lincoln is a current microsoft employee taking christmas day off
73 */

This is available in the examples

Docs

The examples above provide a simple demonstrations of what json-rules-engine can do. To learn more about the advanced features and techniques, see the docs and read through the examples. There is also a walkthrough available.

Persisting Rules

Rules may be easily converted to JSON and persisted to a database, file system, or elsewhere. To convert a rule to JSON, simply call the rule.toJSON() method. Later, a rule may be restored by feeding the json into the Rule constructor.

1// save somewhere...
2let jsonString = rule.toJSON()
3
4// ...later:
5let rule = new Rule(jsonString)

Why aren't "fact" methods persistable? This is by design, for several reasons. Firstly, facts are by definition business logic bespoke to your application, and therefore lie outside the scope of this library. Secondly, many times this request indicates a design smell; try thinking of other ways to compose the rules and facts to accomplish the same objective. Finally, persisting fact methods would involve serializing javascript code, and restoring it later via eval(). If you have a strong desire for this feature, the node-rules project supports this (though be aware the capability is enabled via eval().

Debugging

To see what the engine is doing under the hood, debug output can be turned on via:

Node

1DEBUG=json-rules-engine

Browser

1// set debug flag in local storage & refresh page to see console output
2localStorage.debug = 'json-rules-engine'

License

ISC