Skip to content

m-onyshchuk/stugna

BranchesTags

Repository files navigation

Stugna

Workflow Status Coverage Status

Stugna is a simple engine for expert systems building. Rules and facts can be described in JSON format, so all your knowledge base can be serialized into/from one JSON file.

Usage example

To illustrate how a small expert system can function with rules and facts, we will build one in this example. First of all, we need to create an expert system instance:

const {StugnaES} = require("stugna-es");
let es = new StugnaES();

Our expert system will try to guess transport type by 2 facts - wheels count and motor presence. Next we have to add some rules to our system:

let rules = [
  {
    condition: "wheels = 4 AND motor = 'present'",
    factName: "transport",
    factValue: "car"
  },
  {
    condition: "wheels = 2 AND motor = 'present'",
    factName: "transport",
    factValue: "motorcycle"
  },
  {
    condition: "wheels = 4 AND motor = 'missing'",
    factName: "transport",
    factValue: "skateboard"
  },
  {
    condition: "wheels = 2 AND motor = 'missing'",
    factName: "transport",
    factValue: "bike"
  }
];
es.rulesImport(rules);

Each of these 4 rules contain mandatory fields:

  • condition - a simple logic expression with operators >, <, =, <>, AND, OR, NOT and parentheses
  • factName - a fact's name that will be added to the system if the condition is met. Fact names can be compared to variable names.
  • factValue - fact's value to add to system in success condition case, i.e. value of variable.

Being that simple, our tiny expert system is somewhat useful already. We can ask it what knowledge (facts set) does it need to get info about wanted transport fact:

let factsUnknown = es.factGetPredecessorsUnknown("transport");
console.log(factsUnknown);
// [ 'wheels', 'motor' ]

Next we can add missing facts to the system:

let facts = [
  {
    name: "wheels",
    value: 4,
    description: "This transport has 4 wheels"
  },
  {
    name: "motor",
    value: "missing",
    description: "This transport does`t have motor"
  }
];
es.factsImport(facts);

Now, we can query original parent fact:

let factWanted = es.factGet("transport");
console.log(factWanted);
// {
//   name: 'transport',
//   value: 'skateboard',
//   history: [ "rule: wheels = 4 AND motor = 'missing' / transport / skateboard" ],
//   changed: true
// }

Object factWanted has the following fields:

  • name - fact name
  • value - fact value
  • history - fact history or changes reason list
  • changed - whether the fact was changed by the rules

StugnaES methods

The methods of the StugnaES class are described below.

constructor

Instance creating

const {StugnaES} = require("stugna-es");
let options = {
  toSaveEvents: true,
  toExplainMore: false,
  passCountMax: 16
};
let es = new StugnaES(options);
  • options - object with optional fields:
    • toSaveEvents - parameter to save various events about facts and rules. Default value is true.
    • toExplainMore - parameter for extended explanation of rule processing. Explanations are saved in the events log. Default value is false.
    • passCountMax - parameter to limit passing by rules list. Read more in periodic rules. Default value is 16.

ruleAdd

Add one rule to the system.

let rule = {
  precondition: "is_water = TRUE",
  condition: "is_underwater = TRUE OR is_yellow = TRUE",
  missing: "TRUE",
  factName: "transport",
  factValue: "submarine",
  factNameElse: "transport",
  factValueElse: "ship",
  priority: 5,
  final: 3,
  description: "Is this water transport?"  
};
let isTrigger = true;
es.ruleAdd(rule, isTrigger);

  • precondition - option string field with the same syntax as condition. It can be useful to check rules condition or not. Details in example7.js

  • condition - mandatory string field to describe condition for adding new fact to system. Condition is a logic expression which contains:

    • fact names. Each not reserved word without single quotes is labeled as fact name.
    • boolean values, reserved words TRUE or FALSE
    • numbers, integer or float
    • strings. Each string must be in single quotes. All words without single quotes will be labeled as fact names.
    • operators:
      • > - greater than operator returns true if the left operand is greater than the right operand, and false otherwise, example: wheels > 4
      • >= - greater than or equal operator returns true if the left operand is greater or equal than the right operand, and false otherwise, example: wheels >= 4
      • < - less than operator returns true if the left operand is less than the right operand, and false otherwise, example: wheels < 4
      • <= - less than or equal operator returns true if the left operand is less than or equal the right operand, and false otherwise, example: wheels <= 4
      • = - equality operator checks whether its two operands are equal, returning a boolean result, equality is strict, example: motor = 'present'
      • <> - strict inequality operator checks whether its two operands are not equal, returning a boolean result, example: motor <> 'present'
      • LIKE - operator is similar SQL LIKE operator but without wildcard support (% and _). It returns true if left operand contains substring from right operand. Examples: 'lazy dog' LIKE 'dog' return true and 'lazy dog' LIKE 'fox' returns false.
      • AND - logical AND operator (logical conjunction) for a set of boolean operands will be true if and only if all the operands are true. Otherwise, it will be false.
      • OR - logical OR operator (logical disjunction) for a set of operands is true if and only if one or more of its operands is true.
      • NOT - logical NOT operator (logical complement, negation) takes truth to falsity and vice versa, example: NOT FALSE
      • + - arithmetic addition operator for numbers and concatenation operator for strings, example with numbers: cats > dogs + 2, example with strings: 'lazy_dog' = 'lazy' + '_' + 'dog'
      • - - arithmetic subtraction operator, example: cats > dogs - 2
      • * - multiplication arithmetic operator, example: cats > dogs * 2
      • / - division arithmetic operator, example: cats > dogs / 2
    • parentheses to group operators, example: (wheels = 4 AND motor = 'present') OR weight > 1000

    All reserved words (TRUE, FALSE, AND, OR, NOT, LIKE) are case sensitive. Operator precedence by group, high to low:

    • ( )
    • + - * /
    • < <= > >= = <> LIKE
    • AND OR NOT

  • missing - optional field with default values for missing facts. It works with condition only (not with precondition). Details in example8.js

  • factName - name of new fact, which will be added if condition is met. Field must be a string without spaces, mandatory.

  • factValue - value of new fact, which will be added if condition is met. Field can be boolean (TRUE or FALSE), numerical (integer or float) or string, mandatory.

  • priority - rule priority, number, optional, default value is 1. All rules are processing by order with priority. Rules with small priority are processing first, with big priority - last. Order of rule processing with same priority is undetermined.

  • description - short fact description for logging, string, optional

  • factNameElse - name of new fact, which will be added if condition is not met. Field must be a string without spaces, optional. Details in example5.js

  • factValueElse - value of new fact, which will be added if condition is not met. Field can be boolean (TRUE or FALSE), numerical (integer or float) or string, optional.

  • final - This is an optional numerical field in a rule that can have one of three values: 1, 2, or 3. Each value dictates when to halt the evaluation of subsequent rules:

    • 1 - stop evaluating the next rules if the current rule's condition is met.
    • 2 - stop evaluating the next rules if the current rule's condition is not met and the field factNameElse exists.
    • 3 - stop evaluating the next rules, regardless of any conditions.

    Details in example6.js

  • isTrigger - if true, after new rule adding, rules check procedure starts automatically to generate new possible facts due to given rules in the system. Parameter is boolean, optional, default value - true Fields factNameElse and factValueElse have to be both present or both absent in rule object.

rulesImport

Import array of rules to the system.

let rules = [
  {
    condition: "wheels = 4 AND motor = 'present'",
    factName: "transport",
    factValue: "car",
    priority: 10,
    description: "Transport with engine and 4 wheels is a car"
  },
  {
    condition: "wheels = 2 AND motor = 'present'",
    factName: "transport",
    factValue: "motorcycle",
    priority: 10,
    description: "Transport with engine and 2 wheels is a motorcycle"
  },
  {
    condition: "wheels = 4 AND motor = 'missing'",
    factName: "transport",
    factValue: "skateboard",
    priority: 10,
    description: "Transport with 4 wheels and without engine is a skateboard"
  },
  {
    condition: "wheels = 2 AND motor = 'missing'",
    factName: "transport",
    factValue: "bike",
    priority: 10,
    description: "Transport with 2 wheels and without engine is a bike"
  }
];
let isTrigger = true;
es.rulesImport(rules, isTrigger);
  • rules - array of objects with fields like in ruleAdd method
  • isTrigger - if true, after import, rules check procedure starts automatically to generate new possible facts due to given rules in the system. Parameter is boolean, optional, default value - true

rulesAll

Returns all known rules.

let rulesAll = es.rulesAll();
console.log (rulesAll);
/*
[
  {
    condition: 'is_underwater = TRUE OR is_yellow = TRUE',
    factName: 'transport',
    factValue: 'submarine',
    factNameElse: 'transport',
    factValueElse: 'ship',
    priority: 5,
    description: 'Is this water transport?',
    final: 1
  },
  {
    condition: "wheels = 4 AND motor = 'present'",
    factName: 'transport',
    factValue: 'car',
    priority: 10,
    description: 'Transport with engine and 4 wheels is a car'
  },
  {
    condition: "wheels = 2 AND motor = 'present'",
    factName: 'transport',
    factValue: 'motorcycle',
    priority: 10,
    description: 'Transport with engine and 2 wheels is a motorcycle'
  },
  {
    condition: "wheels = 4 AND motor = 'missing'",
    factName: 'transport',
    factValue: 'skateboard',
    priority: 10,
    description: 'Transport with 4 wheels and without engine is a skateboard'
  },
  {
    condition: "wheels = 2 AND motor = 'missing'",
    factName: 'transport',
    factValue: 'bike',
    priority: 10,
    description: 'Transport with 2 wheels and without engine is a bike'
  }
]
*/

rulesClear

Cleans all rules in the system.

es.rulesClear();

factAdd

Add one fact to the system. If there is a fact with the same name in system it will be overwritten.

let fact = { 
  name: "wheels",
  value: 2, 
  description: "Transport has 2 wheels" 
};
let isTrigger = true;
es.factAdd(fact, isTrigger);
  • name - fact name, string, mandatory. Fact name can`t contains spaces.
  • value - fact value, numerical or string, mandatory
  • description - short fact description for logging, string, optional
  • isTrigger - if true, after new fact adding, rules check procedure starts automatically to generate new possible facts due to given rules in the system. Parameter is boolean, optional, default value - true

factsImport

Import array of facts to the system. Previous facts with same names will be overwritten.

let facts = [
  {
    name: "wheels",
    value: 4,
    description: "This transport has 4 wheels"
  },
  {
    name: "motor",
    value: "missing",
    description: "This transport does`t have motor"
  }
];
let isTrigger = true;
es.factsImport(facts, isTrigger);
  • facts - array of objects with fields like in factAdd method
  • isTrigger - if true, after import, rules check procedure starts automatically to generate new possible facts due to given rules in the system. Parameter is boolean, optional, default value - true

factsAllAsArray

Returns all known facts as array.

let factsArray = es.factsAllAsArray();
console.log (factsArray);
/*
[
  {
    name: 'wheels',
    value: 4,
    history: [
      'init: Transport has 2 wheels',
      'init: This transport has 4 wheels'
    ],
    changed: false
  },
  {
    name: 'motor',
    value: 'missing',
    history: [ 
      'init: This transport does`t have motor' 
    ],
    changed: false
  },
  {
    name: 'transport',
    value: 'skateboard',
    history: [ 
      'rule: Transport with 4 wheels and without engine is a skateboard' 
    ],
    changed: true
  }
]
*/

factsAllAsMap

Returns all known facts as map.

let factsMap = es.factsAllAsMap();
console.log (factsMap);
// { wheels: 4, motor: 'missing', transport: 'skateboard' }

factIsKnown

Is fact already known?

let name = "wheels"; 
let isKnown = es.factIsKnown(name);
console.log (isKnown);
// true
  • name - fact name, string, mandatory
  • return value - boolean

factGet

Returns fact by the name.

let name = "wheels"; 
let fact = es.factGet(name);
console.log (fact);
/*
{
  name: 'wheels',
  value: 4,
  history: [
    'init: Transport has 2 wheels',
    'init: This transport has 4 wheels'
  ],
  changed: false
}
*/
  • name - fact name, string, mandatory
  • value - fact value, number or string
  • history - fact changes history by rules
  • changed - whether the fact was changed by the rules

factGetPredecessorsWanted

Returns all fact names which may be needed to determine asked fact.

let name = "transport"; 
let wanted = es.factGetPredecessorsWanted(name);
console.log (wanted);
// [ 'is_underwater', 'is_yellow', 'wheels', 'motor' ]
  • name - fact name for which it is necessary to find predecessors, string, mandatory
  • return value - array of fact names

factGetPredecessorsUnknown

Returns fact names which are still unknown to determine asked fact. Facts that can be obtained using rules are not included in the returned list.

let name = "transport"; 
let unknown = es.factGetPredecessorsUnknown(name);
console.log (unknown);
// [ 'is_underwater', 'is_yellow' ]
  • name - fact name for which it is necessary to find predecessors, string, mandatory
  • return value - array of fact names

factsAreOrdered

Returns boolean flag - are facts ordered? Facts can be unordered only if periodic rules detected and system can`t order all facts.

let ordered = es.factsAreOrdered();
console.log (ordered);

factsClear

Cleans all facts in the system.

es.factsClear();

eventsAll

Returns all events if logging was turned on in constructor.

let events = es.eventsAll();
console.log (events);
/*
[
  { brief: 'rule add', subject: 'Is this water transport?' },
  { brief: 'rules passed', more: 'Rules pass count is 1' },
  {
    brief: 'rule add',
    subject: 'Transport with engine and 4 wheels is a car'
  },
  {
    brief: 'rule add',
    subject: 'Transport with engine and 2 wheels is a motorcycle'
  },
  {
    brief: 'rule add',
    subject: 'Transport with 4 wheels and without engine is a skateboard'
  },
  {
    brief: 'rule add',
    subject: 'Transport with 2 wheels and without engine is a bike'
  },
  { brief: 'rules passed', more: 'Rules pass count is 1' },
  { brief: 'fact add', subject: 'Transport has 2 wheels' },
  { brief: 'rules passed', more: 'Rules pass count is 1' },
  { brief: 'fact add', subject: 'This transport has 4 wheels' },
  { brief: 'fact add', subject: 'This transport does`t have motor' },
  {
    brief: 'rule ok',
    subject: 'Transport with 4 wheels and without engine is a skateboard'
  },
  { brief: 'rules passed', more: 'Rules pass count is 1' },
  { brief: 'rules passed', more: 'Rules pass count is 2' }
]
*/

eventsClear

Cleans all events in the system.

es.eventsClear();

Periodic rules

New facts in the system appear after passing through the list of rules. The passages stop when no new fact appears. Sometimes a situation may arise when the system has several rules that create a periodic change in facts during one pass when through the list of rules. In such a situation, an endless cycle of changing facts arises, which can only be interrupted artificially by setting the maximum number of passes through the list of rules. This parameter is called passCountMax and is set in the constructor. Below is an example of a periodic set of rules and the message that appears in the logs when the number of passes exceeds passCountMax.

const {StugnaES} = require("stugna-es");
let options = {
  toSaveEvents: true,
  passCountMax: 2
};
let es = new StugnaES(options);

let facts = [
  {
    name: "season",
    value: "winter",
    description: "Initial value of season fact"
  }
];
es.factsImport(facts);

let rules = [
  {
    condition: "season = 'winter'",
    factName: "season",
    factValue: "spring",
    description: "After winter comes spring"
  },
  {
    condition: "season = 'spring'",
    factName: "season",
    factValue: "summer",
    description: "After spring comes summer"
  },
  {
    condition: "season = 'summer'",
    factName: "season",
    factValue: "autumn",
    description: "After summer comes autumn"
  },
  {
    condition: "season = 'autumn'",
    factName: "season",
    factValue: "winter",
    description: "After autumn comes winter"
  },
];
let isTrigger = true;
es.rulesImport(rules, isTrigger);

let ordered = es.factsAreOrdered();
console.log (ordered);

let events = es.eventsAll();
console.log (events);

facts = es.factsAllAsArray();
console.log (facts);

/*
false
[
  { brief: 'fact add', subject: 'Initial value of season fact' },
  { brief: 'rule add', subject: 'After winter comes spring' },
  { brief: 'rule add', subject: 'After spring comes summer' },
  { brief: 'rule add', subject: 'After summer comes autumn' },
  { brief: 'rule add', subject: 'After autumn comes winter' },
  { brief: 'rule ok', subject: 'After winter comes spring' },
  { brief: 'rule ok', subject: 'After spring comes summer' },
  { brief: 'rule ok', subject: 'After summer comes autumn' },
  { brief: 'rule ok', subject: 'After autumn comes winter' },
  { brief: 'rules passed', more: 'Rules pass count is 1' },
  { brief: 'rule ok', subject: 'After winter comes spring' },
  { brief: 'rule ok', subject: 'After spring comes summer' },
  { brief: 'rule ok', subject: 'After summer comes autumn' },
  { brief: 'rule ok', subject: 'After autumn comes winter' },
  { brief: 'rules passed', more: 'Rules pass count is 2' },
  { brief: 'rules error', more: 'Periodic rules detected' }
]
[
  {
    name: 'season',
    value: 'winter',
    history: [
      'init: Initial value of season fact',
      'rule: After winter comes spring',
      'rule: After spring comes summer',
      'rule: After summer comes autumn',
      'rule: After autumn comes winter',
      'rule: After winter comes spring',
      'rule: After spring comes summer',
      'rule: After summer comes autumn',
      'rule: After autumn comes winter'
    ],
    changed: true
  }
]
*/

ruleApply

Standalone function to check one rule with facts array. No need to create StugnaES instance.

First example:

const {ruleApply} = require("./stugna-es");
const condition = "animal = 'cat' OR animal = 'dog'";
const facts = [{name:"animal", value:"cat"}];
let [result, error] = ruleApply(condition, facts);
console.log(result);
console.log(error);
// true
// null

Second example:

const {ruleApply} = require("./stugna-es");
const condition = "animal = 'monkey')";
const facts = [{name:"animal", value:"cat"}];
let [result, error] = ruleApply(condition, facts);
console.log(result);
console.log(error);
// false
// Parentheses mismatched (1)

All examples

Project repository contains set of examples:

About

Stugna is a simple engine for expert systems building.

Topics

javascript json system logic expert

Resources

Readme

License

MIT license
Activity

Stars

6 stars

Watchers

2 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages