saurabh0719/py-rules-engine

py-rules-engine is a robust, Python-based rules engine that enables the creation of intricate logical conditions and actions.

pip install py-rules-engine

Key features:

Complex Logical Conditions: Define intricate conditions using logical operators.
Pythonic Rule Builder: Utilize a Pythonic interface for easy rule definition.
Flexible Rule Management: Store, configure, and share rules in JSON/YAML format, with seamless shifting between Python rule builder and other formats.
Nested Rules: Create multi-level rule structures.
Rule Evaluation: Evaluate rules in a given context with a built-in rule engine.
Zero Dependencies: Pure Python implementation for easy installation and use.

Example usage -

from py_rules.components import Condition, Result, Rule
from py_rules.engine import RuleEngine

# Create a condition
condition = Condition('temperature', '>=', 40) | Condition('wind_speed', '>', 50)

# Create a result
result = Result('message', 'str', 'Unfavourable weather conditions for work!')

# Create a rule
rule = Rule('Temperature Rule').If(condition).Then(result)

# initialise a new instance of RuleEngine with context
context = {'temperature': 45, 'wind_speed': 30}
engine = RuleEngine(context)

print(engine.evaluate(rule))
# 'Unfavourable weather conditions for work!'

# if a Rule is used without a Result, it simply returns True/False
rule = Rule('Bool Temperature Rule').If(condition)
print(engine.evaluate(rule))
# True

Installation

You can install py-rules-engine using pip:

pip install py-rules-engine

NOTE - this library is currently in Beta, and things may break between version bumps pre-1.0.0.

Rule structure

This section covers the basics of this library, including the structure of the Rule object.

Basics

The If-Then-Else structure is a fundamental part of the rule engine. It allows you to define a condition and specify the results based on whether the condition is met or not. Here's a breakdown of each part:

If: This is the condition that will be evaluated. It can be a simple condition (like "temperature > 30") or a complex condition involving multiple variables and logical operators (like "temperature > 30 AND humidity < 50"). The condition is evaluated against a given context, which is a dictionary of variables and their values.
Then: This part specifies the result or action that should be returned or performed if the IF condition is met (i.e., if it evaluates to True). It can be a simple value (like "It's hot") or a complex object. It can also be another Rule, allowing for nested rules.
Else: This part specifies the result or action that should be returned or performed if the IF condition is not met (i.e., if it evaluates to False). Like the THEN part, it can be a simple value, a complex object, or another Rule. The ELSE part is optional; if it's not provided and the IF condition is not met, the rule engine will return False.

Every Rule object's dict representation (rule.to_dict()) contains the following

metadata: This section contains metadata about the rule.
- version: The version of the rule.
- id: A unique identifier for the rule.
- parent_id: The id of the parent rule if this rule is nested, otherwise null.
- created: The UTC timestamp when the rule was created.
- name: The name of the rule.
- required_context_parameters: A list of context parameters that are required for this rule. These variables/parameters are passed to the RuleEngine via a context dictionary
if: This section contains the condition to be evaluated.
- condition or and or or: The condition to be evaluated. It can be a single condition or a logical combination (and, or) of multiple conditions. Each condition consists of a variable, an operator, and a value.
then: This section contains the result to be returned if the if condition evaluates to True.
else: This section contains the result to be returned if the if condition evaluates to False, or a nested rule with its own if, then, and else sections.

All the Rule(s) are evaluated against a context dictionary. The context is key-value dictionary of facts we use to evaluate all conditions and prepare return statements.

See the examples/ directory for more.

Go back to top

Representation

Every Rule object and it's conditions is represented as a dictionary structure. Ev

For example-

from py_rules.components import Condition, Result, Rule
from py_rules.engine import RuleEngine

# Create a condition
condition = Condition('temperature', '>', 40) | Condition('wind_speed', '>', 50)
no_work = Result('message', 'str', 'Unfavourable weather conditions for work!')
work = Result('message', 'str', 'Favourable weather conditions for work!')
rule = Rule('Temperature Rule').If(condition).Then(no_work).Else(work)

print(rule.to_dict())

"""
{
    "metadata": {
        "version": "0.3.0",
        "type": "Rule",
        "id": "a605f337-7d60-4a65-a361-96c282e3fc74",
        "created": "2023-12-15 12:53:48.492187",
        "required_context_parameters": ["wind_speed", "temperature"],
        "name": "Temperature Rule",
        "parent_id": None,
    },
    "if": {
        "or": [
            {
                "condition": {
                    "metadata": {
                        "version": "0.3.0",
                        "type": "Condition",
                        "id": "10346a0c-3934-4ed9-b393-053b871ab17d",
                        "created": "2023-12-15 12:53:48.492090",
                        "required_context_parameters": ["temperature"],
                    },
                    "variable": "temperature",
                    "operator": ">",
                    "value": {"type": "int", "value": 40},
                }
            },
            {
                "condition": {
                    "metadata": {
                        "version": "0.3.0",
                        "type": "Condition",
                        "id": "0f3e660e-01a6-41d1-849c-e5b27088ccb4",
                        "created": "2023-12-15 12:53:48.492140",
                        "required_context_parameters": ["wind_speed"],
                    },
                    "variable": "wind_speed",
                    "operator": ">",
                    "value": {"type": "int", "value": 50},
                }
            },
        ]
    },
    "then": {
        "result": {
            "message": {
                "type": "str",
                "value": "Unfavourable weather conditions for work!",
            }
        }
    },
    "else": {
        "result": {
            "message": {
                "type": "str",
                "value": "Favourable weather conditions for work!",
            }
        }
    },
}
"""

This dictionary is what is used for evaluation.

Go back to top

Rule builder & parser

Rule components

The following components can be used to build complex Rules.

Condition: This class represents a condition in a rule. It can be initialized with a variable, operator, and value, or with a condition dictionary. It supports logical and and or operations to combine conditions. The to_dict method returns a dictionary representation of the condition.
Result: This class represents a result in a Rule. It can be initialized with a key, type, and value, or with a result dictionary. It supports the and operation to combine results. The to_dict method returns a dictionary representation of the result.
Rule: This class represents a rule. It can be initialized with a name and optional keyword arguments. It supports the If, Then, and Else methods to set the condition and results of the rule. The to_dict method returns a dictionary representation of the rule.

from py_rules.components import Condition, Result, Rule

# Create a condition
condition = Condition('temperature', '>', 30)

# Create a result
result = Result('message', 'str', 'It is hot!')

# Create a rule
rule = Rule('Temperature Rule').If(condition).Then(result)

# Print the rule
print(rule.to_dict())

Output -

{
    "metadata": {
        "version": "0.3.0",
        "type": "Rule",
        "id": "5dbca846-5e59-4b6c-bdbf-9602d68c79ff",
        "created": "2023-12-15 04:49:45.183531",
        "required_context_parameters": ["temperature"],
        "name": "Temperature Rule",
        "parent_id": None,
    },
    "if": {
        "condition": {
            "metadata": {
                "version": "0.3.0",
                "type": "Condition",
                "id": "14950c65-3741-40da-ac13-e5cf1a0c49ce",
                "created": "2023-12-15 04:49:45.183438",
                "required_context_parameters": ["temperature"],
            },
            "variable": "temperature",
            "operator": ">",
            "value": {"type": "int", "value": 30},
        }
    },
    "then": {"result": {"message": {"type": "str", "value": "It is hot!"}}},
}

This will create a rule that checks if the temperature is greater than 30 and returns the message "It is hot!" if the condition is met.

Nested rules -

from py_rules.components import Condition, Result, Rule

# Create conditions
condition1 = Condition('temperature', '>', 30)
condition2 = Condition('humidity', '<', 50)

# Create results
result1 = Result('message', 'str', 'It is hot!')
result2 = Result('message', 'str', 'It is dry!')

# Create rules
rule1 = Rule('Temperature Rule').If(condition1).Then(result1)
rule2 = Rule('Humidity Rule').If(condition2).Then(result2)

# Create a nested rule
nested_rule = Rule('Nested Rule').If(condition1).Then(rule2).Else(result1)

# Print the nested rule
print(nested_rule.to_dict())

This will create a nested rule that checks if the temperature is greater than 30. If the condition is met, it evaluates another rule that checks if the humidity is less than 50. If the nested condition is met, it returns the message "It is dry!". If the nested condition is not met, it returns the message "It is hot!".

Here's the equivalent rule in JSON format:

{
    "metadata": {
        "version": "0.3.0",
        "type": "Rule",
        "id": "7f2a4893-322c-4973-8e41-c3930e37648b",
        "created": "2023-12-15 04:57:33.071026",
        "required_context_parameters": ["temperature", "humidity"],
        "name": "Nested Rule",
        "parent_id": null,
    },
    "if": {
        "condition": {
            "metadata": {
                "version": "0.3.0",
                "type": "Condition",
                "id": "515a3a0e-6a45-4848-9ea9-8202e66372cf",
                "created": "2023-12-15 04:57:33.070916",
                "required_context_parameters": ["temperature"],
            },
            "variable": "temperature",
            "operator": ">",
            "value": {"type": "int", "value": 30},
        }
    },
    "then": {
        "metadata": {
            "version": "0.3.0",
            "type": "Rule",
            "id": "7c045ec9-c262-4c6c-8b0f-34cc773d4f41",
            "created": "2023-12-15 04:57:33.071017",
            "required_context_parameters": ["humidity"],
            "name": "Humidity Rule",
            "parent_id": null,
        },
        "if": {
            "condition": {
                "metadata": {
                    "version": "0.3.0",
                    "type": "Condition",
                    "id": "ddc1e0f2-54a5-4c4f-aedd-2f6db69741dc",
                    "created": "2023-12-15 04:57:33.070964",
                    "required_context_parameters": ["humidity"],
                },
                "variable": "humidity",
                "operator": "<",
                "value": {"type": "int", "value": 50},
            }
        },
        "then": {"result": {"message": {"type": "str", "value": "It is dry!"}}},
    },
    "else": {"result": {"message": {"type": "str", "value": "It is hot!"}}},
}

Go back to top

Rule Parser

The RuleParser class is a utility class that is used to parse a rule from a python dictionary into a Rule object. This is useful when you want to define rules in a another format (JSON, YAML, etc.) and then load them into your Python code.

from py_rules.components import Rule
from py_rules.parser import RuleParser

# Define a rule as a dictionary
rule_dict = {
    "metadata": {
        "version": "0.3.0",
        "type": "Rule",
        "id": "9911bf90-b6a1-490f-ae5a-3fa9409529f8",
        "created": "2023-12-15 13:09:59.411292",
        "required_context_parameters": ["temperature"],
        "name": "Unnamed Rule 1",
        "parent_id": None,
    },
    "if": {
        "condition": {
            "metadata": {
                "version": "0.3.0",
                "type": "Condition",
                "id": "a4b678f7-44f3-4818-8a72-298880263703",
                "created": "2023-12-15 13:09:59.411350",
                "required_context_parameters": ["temperature"],
            },
            "variable": "temperature",
            "operator": ">",
            "value": {"type": "int", "value": 30},
        }
    },
    "then": {"result": {"message": {"type": "str", "value": "It's hot"}}},
    "else": {"result": {"message": {"type": "str", "value": "It's not hot"}}},
}


# Parse the rule
rule = RuleParser().parse(rule_dict)

print(type(rule))
# <class 'py_rules.components.Rule'>

assert isinstance(rule, Rule)

print(rule.metadata)
# {'version': '0.3.0', 'type': 'Rule', 'id': '9911bf90-b6a1-490f-ae5a-3fa9409529f8', 'created': '2023-12-15 13:09:59.411292', 'required_context_parameters': ['temperature'], 'name': 'Unnamed Rule 1', 'parent_id': None}

Go back to top

Rule engine and evaluation

The RuleEngine class is used to evaluate a rules. It takes a Rule object and a context (a dictionary) as input. The Rule is then evaluated against the context

Here's a breakdown of the methods in the RuleEngine class:

__init__: Initializes the RuleEngine with a context: dict; a dictionary of facts.
- The context dictionary MUST contain all the variable(s) being used, either in results or in conditions. The required_context_parameters set() of the Rule maintains a list of all unique context parameters required to evaluate the rule.
evaluate(rule: Rule): Evaluates the rule. It checks the 'if' condition and returns the result of the 'then' action if the condition is met, or the result of the 'else' action otherwise.

Here's an example of how to use the RuleEngine class:

from py_rules.builder import Condition, Result, Rule
from py_rules.engine import RuleEngine

# Create a condition
condition = Condition('temperature', '>', 30)

# Create a result
result = Result('message', 'str', 'It is hot!')

# Create a rule
rule = Rule('Temperature Rule').If(condition).Then(result)

# Create a context
context = {'temperature': 35}

# Create a rule engine
engine = RuleEngine(context)

# Evaluate the rule
print(engine.evaluate(rule))  # prints: {'message': 'It is hot!'}

In this example, the rule checks if the temperature is greater than 30. The context provides the actual temperature. The rule engine evaluates the rule in the given context and returns the result of the rule.

You can also use the RuleEngine class with complex rules that have nested conditions and multiple results. Here's an example:

# Create conditions
condition1 = Condition('temperature', '>', 30)
condition2 = Condition('humidity', '<', 50)

# Create results
result1 = Result('message', 'str', 'It is hot!')
result2 = Result('message', 'str', 'It is dry!')

# Create rules
rule1 = Rule('Temperature Rule').If(condition1).Then(result1)
rule2 = Rule('Humidity Rule').If(condition2).Then(result2)

# Create a context
context = {'temperature': 35, 'humidity': 45}

# Create a rule engine
engine = RuleEngine(context)

# Evaluate the rules
print(engine.evaluate(rule1))  # prints: {'message': 'It is hot!'}
print(engine.evaluate(rule2))  # prints: {'message': 'It is dry!'}

In this example, the first rule checks if the temperature is greater than 30, and the second rule checks if the humidity is less than 50. The context provides the actual temperature and humidity. The rule engines evaluate the rules in the given context and return the results of the rules.

Go back to top

Rule parsing and storage

Storing & loading rules from persistent storage enable reuse of rules across different sessions and sharing of rules between different systems.

The RuleStorage class is an abstract base class that defines the common interface for all storage classes. It has two abstract methods: load and store. Any class that inherits from RuleStorage must implement these methods. The RuleStorage class also has a RuleParser object that is used to parse a rule after it is loaded into a dictionary format.

The JSONRuleStorage and PickledRuleStorage classes are concrete classes that inherit from RuleStorage and implement the load and store methods. Each class is designed to work with a specific file format.

Each class also validates the file type in its constructor to ensure that it matches the expected file type. If the file type is not valid, it raises an InvalidRuleError.

You can also create your own RuleStorage class, as shown in the example below for yaml files -

import yaml

from py_rules.components import Condition, Result, Rule
from py_rules.storages import RuleStorage, JSONRuleStorage, PickledRuleStorage

yaml.Dumper.ignore_aliases = lambda *args: True

# CUSTOM Yaml Storage
class YAMLRuleStorage(RuleStorage):
    """
    RuleStorage class for YAML files.
    """

    format = 'yaml'

    def __init__(self, file_path):
        """
        Initialize the loader with a file_path.
        """
        super().__init__()
        self.file_path = file_path
        # validate that the file_path is valid and is a yaml file
        if not self.file_path.endswith('.yaml'):
            raise InvalidRuleError('Invalid file type. Only YAML files are supported.')

    def load(self):
        """
        Load a rule from a YAML file.
        """
        data = {}
        with open(self.file_path) as f:
            data = yaml.load(f, Loader=yaml.FullLoader)
        return self.parser.parse(data)

    def store(self, rule):
        """
        Store a rule in a YAML file.
        """
        data = rule.to_dict()
        with open(self.file_path, 'w') as f:
            yaml.dump(data, f, default_flow_style=False, indent=4, sort_keys=False)


# Define conditions
condition1 = Condition('number', 'in', [1, 2, 3])
condition2 = Condition('number', '=', 1)

# Combine conditions using logical 'and'
combined_condition = condition1 & condition2

# Define results
result1 = Result('xyz', 'str', 'Condition met')
result2 = Result('result', 'variable', 'xyz')

# Combine results using logical 'and'
combined_result = result1 & result2

# Define a nested rule
nested_rule = Rule('Nested rule').If(condition1).Then(result1)

# Define a complex rule with nested conditions and rules
complex_rule = Rule('Complex rule').If(combined_condition).Then(combined_result).Else(nested_rule)

# Store the complex rule in different formats
JSONRuleStorage('rule.json').store(complex_rule)
YAMLRuleStorage('rule.yaml').store(complex_rule)

# Load the complex rule from each format
json_rule = JSONRuleStorage('rule.json').load()
yaml_rule = YAMLRuleStorage('rule.yaml').load()

assert complex_rule == json_rule == yaml_rule

rule.json -

{
    "metadata": {
        "version": "0.3.0",
        "type": "Rule",
        "id": "8330dd39-a0a4-4f21-aab4-0f8e35924c74",
        "created": "2023-12-15 04:54:00.349206",
        "required_context_parameters": [
            "xyz",
            "number"
        ],
        "name": "Complex rule",
        "parent_id": null
    },
    "if": {
        "and": [
            {
                "condition": {
                    "metadata": {
                        "version": "0.3.0",
                        "type": "Condition",
                        "id": "16a74acf-3dfd-4c8f-a280-50dc3970455c",
                        "created": "2023-12-15 04:54:00.349063",
                        "required_context_parameters": [
                            "number"
                        ]
                    },
                    "variable": "number",
                    "operator": "in",
                    "value": {
                        "type": "list",
                        "value": [
                            {
                                "type": "int",
                                "value": 1
                            },
                            {
                                "type": "int",
                                "value": 2
                            },
                            {
                                "type": "int",
                                "value": 3
                            }
                        ]
                    }
                }
            },
            {
                "condition": {
                    "metadata": {
                        "version": "0.3.0",
                        "type": "Condition",
                        "id": "137af61f-16af-44b9-8c49-a1c61912704b",
                        "created": "2023-12-15 04:54:00.349134",
                        "required_context_parameters": [
                            "number"
                        ]
                    },
                    "variable": "number",
                    "operator": "=",
                    "value": {
                        "type": "int",
                        "value": 1
                    }
                }
            }
        ]
    },
    "then": {
        "result": {
            "xyz": {
                "type": "str",
                "value": "Condition met"
            },
            "result": {
                "type": "variable",
                "value": "xyz"
            }
        }
    },
    "else": {
        "metadata": {
            "version": "0.3.0",
            "type": "Rule",
            "id": "41ad9aa7-d072-49d1-9336-4186a3a6b69c",
            "created": "2023-12-15 04:54:00.349189",
            "required_context_parameters": [
                "number"
            ],
            "name": "Nested rule",
            "parent_id": null
        },
        "if": {
            "condition": {
                "metadata": {
                    "version": "0.3.0",
                    "type": "Condition",
                    "id": "16a74acf-3dfd-4c8f-a280-50dc3970455c",
                    "created": "2023-12-15 04:54:00.349063",
                    "required_context_parameters": [
                        "number"
                    ]
                },
                "variable": "number",
                "operator": "in",
                "value": {
                    "type": "list",
                    "value": [
                        {
                            "type": "int",
                            "value": 1
                        },
                        {
                            "type": "int",
                            "value": 2
                        },
                        {
                            "type": "int",
                            "value": 3
                        }
                    ]
                }
            }
        },
        "then": {
            "result": {
                "xyz": {
                    "type": "str",
                    "value": "Condition met"
                }
            }
        }
    }
}

rule.yaml

metadata:
    version: 0.3.0
    type: Rule
    id: 8330dd39-a0a4-4f21-aab4-0f8e35924c74
    created: '2023-12-15 04:54:00.349206'
    required_context_parameters:
    - xyz
    - number
    name: Complex rule
    parent_id: null
if:
    and:
    -   condition:
            metadata:
                version: 0.3.0
                type: Condition
                id: 16a74acf-3dfd-4c8f-a280-50dc3970455c
                created: '2023-12-15 04:54:00.349063'
                required_context_parameters:
                - number
            variable: number
            operator: in
            value:
                type: list
                value:
                -   type: int
                    value: 1
                -   type: int
                    value: 2
                -   type: int
                    value: 3
    -   condition:
            metadata:
                version: 0.3.0
                type: Condition
                id: 137af61f-16af-44b9-8c49-a1c61912704b
                created: '2023-12-15 04:54:00.349134'
                required_context_parameters:
                - number
            variable: number
            operator: '='
            value:
                type: int
                value: 1
then:
    result:
        xyz:
            type: str
            value: Condition met
        result:
            type: variable
            value: xyz
else:
    metadata:
        version: 0.3.0
        type: Rule
        id: 41ad9aa7-d072-49d1-9336-4186a3a6b69c
        created: '2023-12-15 04:54:00.349189'
        required_context_parameters:
        - number
        name: Nested rule
        parent_id: null
    if:
        condition:
            metadata:
                version: 0.3.0
                type: Condition
                id: 16a74acf-3dfd-4c8f-a280-50dc3970455c
                created: '2023-12-15 04:54:00.349063'
                required_context_parameters:
                - number
            variable: number
            operator: in
            value:
                type: list
                value:
                -   type: int
                    value: 1
                -   type: int
                    value: 2
                -   type: int
                    value: 3
    then:
        result:
            xyz:
                type: str
                value: Condition met

Go back to top

Tests

Install all dev dependencies

$ pip install dev-requirements.txt

Run tests -

python -m unittest -v

To do

Pass a global config dictionary from RuleEngine to control the following -
- Date parsing functions
- Rule metadata creation

Go back to top

saurabh0719 / py-rules-engine

Table of Contents :

Installation

Rule structure

Basics

Representation

Rule builder & parser

Rule components

Nested rules -

Rule Parser

Rule engine and evaluation

Rule parsing and storage

Tests

To do

About

Languages