pyknow Documentation

Release 1.5.0

Roberto Abdelkader Martínez Pérez

Apr 05, 2018

 Contents

1 User Guide 3
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 Difference between CLIPS and PyKnow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 From PyPI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Getting the source code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 The Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.3 DefFacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.4 KnowledgeEngine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.1 Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4.2 Conditional Elements: Composing Patterns Together . . . . . . . . . . . . . . . . . . . . . 13
1.4.3 Field Constraints: FC for sort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.4 Composing FCs: &, | and ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.5 Variable Binding: The << Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4.6 MATCH object . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.4.7 AS object . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.4.8 Nested matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.4.9 Mutable objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 Cookbook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2 API Documentation 21
2.1 Modules documentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.1 pyknow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.2 pyknow.abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.3 pyknow.activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.4 pyknow.agenda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.5 pyknow.conditionalelement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.6 pyknow.engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.7 pyknow.factlist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.1.8 pyknow.fact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.1.9 pyknow.fieldconstraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

i
 2.1.10 pyknow.rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.11 pyknow.strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.12 pyknow.watchers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.13 pyknow.matchers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.1.14 pyknow.matchers.rete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.1.15 pyknow.matchers.rete.abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.1.16 pyknow.matchers.rete.check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.1.17 pyknow.matchers.rete.dnf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1.18 pyknow.matchers.rete.mixins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1.19 pyknow.matchers.rete.nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1.20 pyknow.matchers.rete.token . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.1.21 pyknow.matchers.rete.utils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.1.22 pyknow.operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3 Release History 33
3.1 1.5.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 1.4.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3 1.3.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4 1.2.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.5 1.1.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.6 1.1.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.7 1.0.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.8 1.0.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.9 <1.0.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

4 Source documentation 35

Python Module Index 37

ii
 pyknow Documentation, Release 1.5.0

PyKnow is a Python library for building expert systems strongly inspired by CLIPS.

from random import choice

from pyknow import *

class Light(Fact):
"""Info about the traffic light."""
pass

class RobotCrossStreet(KnowledgeEngine):
@Rule(Light(color='green'))
def green_light(self):
print("Walk")

@Rule(Light(color='red'))
def red_light(self):
print("Don't walk")

@Rule(AS.light << Light(color=L('yellow') | L('blinking-yellow')))

def cautious(self, light):
print("Be cautious because light is", light["color"])

>>> engine = RobotCrossStreet()

>>> engine.reset()
>>> engine.declare(Light(color=choice(['green', 'yellow', 'blinking-yellow', 'red'])))
>>> engine.run()
Be cautious because light is blinking-yellow

You can find some more examples on GitHub.

Contents 1
pyknow Documentation, Release 1.5.0

2 Contents
 CHAPTER 1

User Guide

1.1 Introduction

1.1.1 Philosophy

We aim to implement a Python alternative to CLIPS, as compatible as possible. With the goal of making it easy for
the CLIPS programmer to transfer all of his/her knowledge to this platform.

1.1.2 Features

• Python 3 compatible.
• Pure Python implementation.
• Matcher based on the RETE algorithm.

1.1.3 Difference between CLIPS and PyKnow

1. CLIPS is a programming language, PyKnow is a Python library. This imposes some limitations on the construc-
tions we can do (specially on the LHS of a rule).
2. CLIPS is written in C, PyKnow in Python. A noticeable impact in performance is to be expected.
3. In CLIPS you add facts using assert, in Python assert is a keyword, so we use declare instead.

1.2 Installation

1.2.1 From PyPI

To install PyKnow, run this command in your terminal:

3
pyknow Documentation, Release 1.5.0

$ pip install pyknow

1.2.2 Getting the source code

PyKnow is developed on Github.

You can clone the repository using the git command:
$ git clone https://github.com/buguroo/pyknow.git

Or you can download the releases in .zip or .tar.gz format.

Once you have a copy of the source, you can install it running this command:
$ python setup.py install

1.3 The Basics

An expert system is a program capable of pairing up a set of facts with a set of rules to those facts, and execute some
actions based on the matching rules.

1.3.1 Facts

Facts are the basic unit of information of PyKnow. They are used by the system to reason about the problem.
Let’s enumerate some facts about Facts, so. . . metafacts ;)
1. The class Fact is a subclass of dict.
>>> f = Fact(a=1, b=2)
>>> f['a']
1

2. Therefore a Fact does not mantain an internal order of items.

>>> Fact(a=1, b=2) # Order is arbirary :O
Fact(b=2, a=1)

3. In contrast to dict, you can create a Fact without keys (only values), and Fact will create a numeric index for
your values.
>>> f = Fact('x', 'y', 'z')
>>> f[0]
'x'

4. You can mix autonumeric values with key-values, but autonumeric must be declared first:
>>> f = Fact('x', 'y', 'z', a=1, b=2)
>>> f[1]
'y'
>>> f['b']
2

5. You can subclass Fact to express different kinds of data or extend it with your custom functionality.

4 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

class Alert(Fact):
"""The alert level."""
pass

class Status(Fact):
"""The system status."""
pass

f1 = Alert('red')
f2 = Status('critical')

from pyknow import Fact

from django.contrib.auth.models import User as DjangoUser

class User(Fact):
@classmethod
def from_django_model(cls, obj):
return cls(pk=obj.pk,
name=obj.name,
email=obj.email)

def save_to_db(self):
return DjangoUser.create(**self)

1.3.2 Rules

In PyKnow a rule is a callable, decorated with Rule.

Rules have two components, LHS (left-hand-side) and RHS (right-hand-side).
• The LHS describes (using patterns) the conditions on which the rule * should be executed (or fired).
• The RHS is the set of actions to perform when the rule is fired.
For a Fact to match a Pattern, all pattern restrictions must be True when the Fact is evaluated against it.

class MyFact(Fact):
pass

@Rule(MyFact()) # This is the LHS

def match_with_every_myfact():
"""This rule will match with every instance of `MyFact`."""
# This is the RHS
pass

@Rule(Fact('animal', family='felinae'))
def match_with_cats():
"""
Match with every `Fact` which:

* f[0] == 'animal'
* f['family'] == 'felinae'

"""
print("Meow!")

You can use logic operators to express complex LHS conditions.

1.3. The Basics 5

pyknow Documentation, Release 1.5.0

@Rule(
AND(
OR(User('admin'),
User('root')),
NOT(Fact('drop-privileges'))
)
)
def the_user_has_power():
"""
The user is a privileged one and we are not dropping privileges.

"""
enable_superpowers()

For a Rule to be useful, it must be a method of a KnowledgeEngine subclass.

Note: For a list of more complex operators you can check the pyknow.operator module.

Facts vs Patterns

The difference between Facts and Patterns is small. In fact, Patterns are just Facts containing Pattern Conditional
Elements instead of regular data. They are used only in the LHS of a rule.
If you don’t provide the content of a pattern as a PCE, PyKnow will enclose the value in a LiteralPCE automatically
for you.
Also, you can’t declare any Fact containing a PCE, if you do, you will receive a nice exception back.

>>> ke = KnowledgeEngine()
>>> ke.declare(Fact(L("hi")))
Traceback (most recent call last):
File "<ipython-input-4-b36cff89278d>", line 1, in <module>
ke.declare(Fact(L('hi')))
File "/home/pyknow/pyknow/engine.py", line 210, in declare
self.__declare(*facts)
File "/home/pyknow/pyknow/engine.py", line 191, in __declare
"Declared facts cannot contain conditional elements")
TypeError: Declared facts cannot contain conditional elements

1.3.3 DefFacts

Most of the time expert systems needs a set of facts to be present for the system to work. This is the purpose of the
DefFacts decorator.

@DefFacts()
def needed_data():
yield Fact(best_color="red")
yield Fact(best_body="medium")
yield Fact(best_sweetness="dry")

All DefFacts inside a KnowledgeEngine will be called every time the reset method is called.

6 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

Note: The decorated method MUST be generators.

1.3.4 KnowledgeEngine

This is the place where all the magic happens.

The first step is to make a subclass of it and use Rule to decorate its methods.
After that, you can instantiate it, populate it with facts, and finally run it.

Listing 1: greet.py
from pyknow import *

class Greetings(KnowledgeEngine):
@DefFacts()
def _initial_action(self):
yield Fact(action="greet")

@Rule(Fact(action='greet'),
NOT(Fact(name=W())))
def ask_name(self):
self.declare(Fact(name=input("What's your name? ")))

@Rule(Fact(action='greet'),
NOT(Fact(location=W())))
def ask_location(self):
self.declare(Fact(location=input("Where are you? ")))

@Rule(Fact(action='greet'),
Fact(name=MATCH.name),
Fact(location=MATCH.location))
def greet(self, name, location):
print("Hi %s! How is the weather in %s?" % (name, location))

engine = Greetings()
engine.reset() # Prepare the engine for the execution.
engine.run() # Run it!

$ python greet.py
What's your name? Roberto
Where are you? Madrid
Hi Roberto! How is the weather in Madrid?

Handling facts

The following methods are used to manipulate the set of facts the engine knows about.

declare

Adds a new fact to the factlist (the list of facts known by the engine).

1.3. The Basics 7

pyknow Documentation, Release 1.5.0

>>> engine = KnowledgeEngine()

>>> engine.reset()
>>> engine.declare(Fact(score=5))
<f-1>
>>> engine.facts
<f-0> InitialFact()
<f-1> Fact(score=5)

Note: The same fact can’t be declared twice unless facts.duplication is set to True.

retract

Removes an existing fact from the factlist.

Listing 2: Both, the index and the fact can be used with retract
>>> engine.facts
<f-0> InitialFact()
<f-1> Fact(score=5)
<f-2> Fact(color='red')
>>> engine.retract(1)
>>> engine.facts
<f-0> InitialFact()
<f-2> Fact(color='red')

modify

Retracts some fact from the factlist and declares a new one with some changes. Changes are passed as arguments.
>>> engine.facts
<f-0> InitialFact()
<f-1> Fact(color='red')
>>> engine.modify(engine.facts[1], color='yellow', blink=True)
<f-2>
>>> engine.facts
<f-0> InitialFact()
<f-2> Fact(color='yellow', blink=True)

duplicate

Adds a new fact to the factlist using an existing fact as a template and adding some modifications.
>>> engine.facts
<f-0> InitialFact()
<f-1> Fact(color='red')
>>> engine.duplicate(engine.facts[1], color='yellow', blink=True)
<f-2>
>>> engine.facts
<f-0> InitialFact()
<f-1> Fact(color='red')
<f-2> Fact(color='yellow', blink=True)

8 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

Cycle of execution: DefFacts, reset & run

Because this topic is often a direct cause of misunderstanding, it deserves a special mention here, in the basics.
For a KnowledgeEngine to run, this things must happen:
1. The class must be instantiated, of course.
2. The reset method must be called:
• This declares the special fact InitialFact. Necessary for some rules to work properly.
• Declare all facts yielded by the methods decorated with @DefFacts.
3. The run method must be called. This starts the cycle of execution.

Difference between DefFacts and declare

Both are used to declare facts on the engine instance, but:

• declare adds the facts directly to the working memory.
• Generators declared with DefFacts are called by the reset method, and all the yielded facts they are added to the
working memory using declare.

1.4 Reference

The following diagram shows all the system components and the relationships among them.

1.4. Reference 9
pyknow Documentation, Release 1.5.0

Fact

retract declare

KnowledgeEngine contains

has has

Strategy has FactList contains

organize

contains Agenda

contains which is a special kind of

Activation

contains contains

Rule Context

contains

Conditional Element (CE) contains contains

contains

Pattern

contains

Field Constraint (FC) some of them contains

10 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

1.4.1 Rule

Rule is the basic method of composing patterns. You can add as many patterns or conditional elements as you want to
a Rule and it will fire if every one of them matches. Therefore, it behaves like AND by default.

@Rule(<pattern_1>,
<pattern_2>,
...
<pattern_n>)
def _():
pass

The following diagram shows the rules of composition of a rule:

1.4. Reference 11
pyknow Documentation, Release 1.5.0

Rule

contains

Conditional Element (CE)

contains contains
AND, OR, NOT

contains

Pattern

contains

Composable FC
contains
&, |, ~

contains

Field Constraint (FC)

L(), W(), P()

salience

This value, by default 0, determines the priority of the rule in relation to the others. Rules with a higher salience will
be fired before rules with a lower one.

Listing 3: r1 has precedence over r2

@Rule(salience=1)
def r1():
pass
(continues on next page)

12 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

(continued from previous page)

@Rule(salience=0)
def r2():
pass

1.4.2 Conditional Elements: Composing Patterns Together

AND

AND creates a composed pattern containing all Facts passed as arguments. All of the passed patterns must match for
the composed pattern to match.

Listing 4: Match if two facts are declared, one matching Fact(1) and
other matching Fact(2)
@Rule(AND(Fact(1),
Fact(2)))
def _():
pass

OR

OR creates a composed pattern in which any of the given pattern will make the rule match.

Listing 5: Match if a fact matching Fact(1) exists and/or a fact matching

Fact(2) exists
@Rule(OR(Fact(1),
Fact(2)))
def _():
pass

Warning: If multiple facts match, the rule will be fired multiple times, one for each valid combination of matching
facts.

NOT

This element matches if the given pattern does not match with any fact or combination of facts. Therefore this element
matches the absence of the given pattern.

1.4. Reference 13
pyknow Documentation, Release 1.5.0

Listing 6: Match if no fact match with Fact(1)

@Rule(NOT(Fact(1)))
def _():
pass

TEST

Check the received callable against the current binded values. If the execution returns True the evaluation will continue
and stops otherwise.

Listing 7: Match for all numbers a, b, c where a > b > c

@Rule(Number(MATCH.a),
Number(MATCH.b),
TEST(lambda a, b: a > b),
Number(MATCH.c),
TEST(lambda b, c: b > c))
def _(a, b, c):
pass

EXISTS

This CE receives a pattern and matches if one or more facts matches this pattern. This will match only once while one
or more matching facts exists and will stop matching when there is no matching facts.

Listing 8: Match once when one or more Color exists

@Rule(EXISTS(Color()))
def _():
pass

FORALL

The FORALL conditional element provides a mechanism for determining if a group of specified CEs is satisfied for
every occurence of another specified CE.

Listing 9: Match when for every Student fact there is a Reading, Writing
and Arithmetic fact with the same name.
@Rule(FORALL(Student(MATCH.name),
Reading(MATCH.name),
Writing(MATCH.name),
Arithmetic(MATCH.name)))
def all_students_passed():
pass

Note: All binded variables captured inside a FORALL clause won’t be passed as context to the RHS of the rule.

14 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

Note: Any time the rule is activated the matching fact is the InitialFact.

1.4.3 Field Constraints: FC for sort

L (Literal Field Constraint)

This element performs an exact match with the given value. The matching is done using the equality operator ==.

Listing 10: Match if the first element is exactly 3

@Rule(Fact(L(3)))
def _():
pass

Note: This is the default FC used when no FC is given as a pattern value. pattern.

W (Wildcard Field Constraint)

This element matches with any value.

Listing 11: Match if some fact is declared with the key mykey.
@Rule(Fact(mykey=W()))
def _():
pass

Note: This element only matches if the element exist.

P (Predicate Field Constraint)

The match of this element is the result of applying the given callable to the fact-extracted value. If the callable returns
True the FC will match, in other case the FC will not match.

Listing 12: Match if some fact is declared whose first parameter is an

instance of int
@Rule(Fact(P(lambda x: isinstance(x, int))))
def _():
pass

1.4.4 Composing FCs: &, | and ~

All FC can be composed together using the composition operators &, | and ~.

1.4. Reference 15
pyknow Documentation, Release 1.5.0

ANDFC() a.k.a. &

The composed FC matches if all the given FC match.

Listing 13: Match if key x of Point is a value between 0 and 255.

@Rule(Fact(x=P(lambda x: x >= 0) & P(lambda x: x <= 255)))
def _():
pass

ORFC() a.k.a. |

The composed FC matches if any of the given FC matches.

Listing 14: Match if name is either Alice or Bob.

@Rule(Fact(name=L('Alice') | L('Bob')))
def _():
pass

NOTFC() a.k.a. ~

This composed FC negates the given FC, reversing the logic. If the given FC matches this will not and vice versa.

Listing 15: Match if name is not Charlie.

@Rule(Fact(name=~L('Charlie')))
def _():
pass

1.4.5 Variable Binding: The << Operator

Any pattern and some FCs can be binded to a name using the << operator.

Listing 16: The first value of the matching fact will be binded to the name
value and passed to the function when fired.
@Rule(Fact('value' << W()))
def _(value):
pass

Deprecated since version 1.2.0: Use MATCH object instead.

Listing 17: The whole matching fact will be binded to f1 and passed to
the function when fired.
@Rule('f1' << Fact())
def _(f1):
pass

Deprecated since version 1.2.0: Use AS object instead.

16 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

1.4.6 MATCH object

The MATCH objects helps generating more readable name bindings. Is syntactic sugar for a Wildcard Field Constraint
binded to a name. For example:

@Rule(Fact(MATCH.myvalue))
def _(myvalue):
pass

Is exactly the same as:

@Rule(Fact("myvalue" << W()))

def _(myvalue):
pass

1.4.7 AS object

The AS object like the MATCH object is syntactic sugar for generating bindable names. In this case any attribute
requested to the AS object will return a string with the same name.

@Rule(AS.myfact << Fact(W()))

def _(myfact):
pass

Is exactly the same as:

@Rule("myfact" << Fact(W()))

def _(myfact):
pass

Warning: This behavior will vary in future releases of PyKnow and the string flavour of the operator may
disappear.

1.4.8 Nested matching

New in version 1.3.0.

Nested matching is useful to match against Fact values which contains nested structures like dicts or lists.

>>> Fact(name="scissors", against={"scissors": 0, "rock": -1, "paper": 1})

>>> Fact(name="paper", against={"scissors": -1, "rock": 1, "paper": 0})
>>> Fact(name="rock", against={"scissors": 1, "rock": 0, "paper": -1})

Nested matching take the form field__subkey=value. (That’s a double-underscore). For example:

>>> @Rule(Fact(name=MATCH.name, against__scissors=1, against__paper=-1))

... def what_wins_to_scissors_and_losses_to_paper(self, name):
... print(name)

Is possible to match against an arbitrary deep structure following the same method.

1.4. Reference 17
pyknow Documentation, Release 1.5.0

>>> class Ship(Fact):

... pass
...
>>> Ship(data={
... "name": "SmallShip",
... "position": {
... "x": 300,
... "y": 200},
... "parent": {
... "name": "BigShip",
... "position": {
... "x": 150,
... "y": 300}}})

In this example we can check for collision between a ship and its parent with the following rule:
>>> @Rule(Ship(data__name=MATCH.name1,
... data__position__x=MATCH.x,
... data__position__y=MATCH.y,
... data__parent__name=MATCH.name2,
... data__parent__position__x=MATCH.x,
... data__parent__position__y=MATCH.y))
... def collision_detected(self, name1, name2, **_):
... print("COLLISION!", name1, name2)

If the nested data structure contains list, tuples or any other sequence you can use numeric indexes as needed.
>>> Ship(data={
... "name": "SmallShip",
... "position": {
... "x": 300,
... "y": 200},
... "enemies": [
... {"name": "Destroyer"},
... {"name": "BigShip"}]})
>>>
>>> @Rule(Ship(data__enemies__0__name="Destroyer"))
... def next_enemy_is_destroyer(self):
... print("Bye byee!")

1.4.9 Mutable objects

PyKnow’s matching algorithm depends on the values of the declared facts being immutable.
When a Fact is created, all its values are transformed to an immutable type if they are not. For this matter the method
pyknow.utils.freeze is used internally.
>>> class MutableTest(KnowledgeEngine):
... @Rule(Fact(v1=MATCH.v1, v2=MATCH.v2, v3=MATCH.v3))
... def is_immutable(self, v1, v2, v3):
... print(type(v1), "is Immutable!")
... print(type(v2), "is Immutable!")
... print(type(v3), "is Immutable!")
...
>>> ke = MutableTest()
>>> ke.reset()
(continues on next page)

18 Chapter 1. User Guide

 pyknow Documentation, Release 1.5.0

(continued from previous page)

>>> ke.declare(Fact(v1={"a": 1, "b": 2}, v2=[1, 2, 3], v3={1, 2, 3}))
>>> ke.run()
frozendict is Immutable
frozenlist is Immutable
frozenset is Immutable
>>>

Note: You can import frozendict and frozenlist from pyknow.utils module. However frozenset is a Python built-in
type.

Register your own mutable freezer

If you need to include your own custom mutable types as fact values you have to register a specialized type freezer for
your custom type.

>>> from pyknow.utils import freeze

>>> @freeze.register(MyType)
... def freeze_mytype(obj):
... return ... # My frozen version of my type

Unfreeze frozen objects

To easily unfreeze the frozen objects pyknow.utils contains an unfreeze method.

>>> class MutableTest(KnowledgeEngine):

... @Rule(Fact(v1=MATCH.v1, v2=MATCH.v2, v3=MATCH.v3))
... def is_immutable(self, v1, v2, v3):
... print(type(unfreeze(v1)), "is Mutable!")
... print(type(unfreeze(v2)), "is Mutable!")
... print(type(unfreeze(v3)), "is Mutable!")
...
>>> ke = MutableTest()
>>> ke.reset()
>>> ke.declare(Fact(v1={"a": 1, "b": 2}, v2=[1, 2, 3], v3={1, 2, 3}))
>>> ke.run()
dict is Mutable
list is Mutable
set is Mutable
>>>

Note: The same freeze registration procedure shown above also applies to unfreeze.

1.5 Cookbook

1.5. Cookbook 19
pyknow Documentation, Release 1.5.0

20 Chapter 1. User Guide

 CHAPTER 2

API Documentation

2.1 Modules documentation

2.1.1 pyknow

2.1.2 pyknow.abstract

class pyknow.abstract.Matcher(engine)
Bases: object
changes(adding=None, deleting=None)
Main interface with the matcher.
Called by the knowledge engine when changes are made in the working memory and return a set of
activations. ‫ﯾﺗم اﺳﺗدﻋﺎؤه ﺑواﺳطﺔ ﻣﺣرك اﻟﻣﻌرﻓﺔ ﻋﻧد إﺟراء اﻟﺗﻐﯾﯾرات ﻓﻲ اﻟذاﻛرة اﻟﻌﺎﻣﻠﺔ وإرﺟﺎع ﻣﺟﻣوﻋﺔ ﻣن ﻋﻣﻠﯾﺎت اﻟﺗﻧﺷﯾط‬.
reset()
Reset the matcher memory.
class pyknow.abstract.Strategy(*args, **kwargs)
Bases: object
update_agenda(agenda, added, removed)

2.1.3 pyknow.activation
‫ﺗﻣﺛل ﻋﻣﻠﯾﺎت اﻟﺗﻧﺷﯾط اﻟﻘواﻋد اﻟﺗﻲ ﺗﺗواﻓﻖ ﻣﻊ ﻗﺎﺋﻣﺔ ﺣﻘﺎﺋﻖ ﻣﺣددة‬.
Activations represent rules that matches against a specific factlist.
class pyknow.activation.Activation(rule, facts, context=None)
Bases: object
Activation object

21
pyknow Documentation, Release 1.5.0

2.1.4 pyknow.agenda

class pyknow.agenda.Agenda
Bases: object
‫ﻣﺟﻣوﻋﺔ ﻣن ﻋﻣﻠﯾﺎت اﻟﺗﻧﺷﯾط اﻟﺗﻲ ﺗﺗﻌﺎﻣل ﻣﻊ ﺣﺎﻟﺔ اﻟﺗﻧﻔﯾذ اﻟﺧﺎﺻﺔ ﺑﮭﺎ‬
Collection of activations that handles its execution state.

Note: Extracted from clips documentation: The agenda is a collection of activations

which are those rules which match pattern entities ‫ﺟدول اﻷﻋﻣﺎل ﻋﺑﺎرة ﻋن ﻣﺟﻣوﻋﺔ ﻣن ﻋﻣﻠﯾﺎت اﻟﺗﻧﺷﯾط اﻟﺗﻲ‬
‫ھﻲ ﺗﻠك اﻟﻘواﻋد اﻟﺗﻲ ﺗطﺎﺑﻖ ﻛﯾﺎﻧﺎت اﻟﻧﻣط‬

get_next()
Returns the next activation, removes it from activations list.

2.1.5 pyknow.conditionalelement

class pyknow.conditionalelement.AND
Bases: pyknow.conditionalelement.OperableCE, pyknow.conditionalelement.
ConditionalElement
class pyknow.conditionalelement.OR
Bases: pyknow.conditionalelement.OperableCE, pyknow.conditionalelement.
ConditionalElement
class pyknow.conditionalelement.NOT
Bases: pyknow.conditionalelement.OperableCE, pyknow.conditionalelement.
ConditionalElement
class pyknow.conditionalelement.TEST
Bases: pyknow.conditionalelement.OperableCE, pyknow.conditionalelement.
ConditionalElement
class pyknow.conditionalelement.EXISTS
Bases: pyknow.conditionalelement.OperableCE, pyknow.conditionalelement.
ConditionalElement
class pyknow.conditionalelement.FORALL
Bases: pyknow.conditionalelement.OperableCE, pyknow.conditionalelement.
ConditionalElement

2.1.6 pyknow.engine

pyknow engine represents CLIPS modules

class pyknow.engine.KnowledgeEngine
Bases: object
This represents a clips’ module, wich is an inference engine holding a set of rules (as pyknow.
rule.Rule objects), an agenda (as pyknow.agenda.Agenda object) and a fact-list (as pyknow.
factlist.FactList objects)
This could be considered, when inherited from, as the knowlege-base.
declare(*facts)
Declare from inside a fact, equivalent to assert in clips.

22 Chapter 2. API Documentation

 pyknow Documentation, Release 1.5.0

Note: This updates the agenda.

duplicate(template_fact, **modifiers)
Create a new fact from an existing one.
get_activations()
Return activations
get_deffacts()
Return the existing deffacts sorted by the internal order
get_rules()
Return the existing rules.
halt()
modify(declared_fact, **modifiers)
Modifies a fact.
Facts are inmutable in Clips, thus, as documented in clips reference manual, this retracts a fact and then
re-declares it
modifiers must be a Mapping object containing keys and values to be changed.
To allow modifying positional facts, the user can pass a string containing the symbol “_” followed by the
numeric index (starting with 0). Ex:

>>> ke.modify(my_fact, _0="hello", _1="world", other_key="!")

reset()
Performs a reset as per CLIPS behaviour (resets the agenda and factlist and declares InitialFact())

Note: If persistent facts have been added, they’ll be re-declared.

retract(idx_or_declared_fact)
Retracts a specific fact, using its index

Note: This updates the agenda

run(steps=inf )
Execute agenda activations

2.1.7 pyknow.factlist

fact-list implementation from CLIPS.

See Making a List section on the user guide. Also see retrieving the fact-list on the clips programming manual
class pyknow.factlist.FactList
Bases: collections.OrderedDict
Contains a list of facts (asserted data).
In clips, there is the concept of “modules” (pyknow.engine.KnowledgeEngine), wich have their own
pyknow.factlist.FactList and pyknow.agenda.Agenda

2.1. Modules documentation 23

pyknow Documentation, Release 1.5.0

A factlist acts as both the module’s factlist and a fact-set yet currently most methods from a fact-set
are not yet implemented
changes
Return a tuple with the removed and added facts since last run.
declare(fact)
Assert (in clips terminology) a fact.
This keeps insertion order.

Warning: This will reject any object that not descend from the Fact class.

Parameters fact – The fact to declare, must be derived from pyknow.fact.Fact.

Returns (int) The index of the fact in the list.
Throws ValueError If the fact providen is not a Fact object

retract(idx_or_fact)
Retract a previously asserted fact.
See “Retract that fact” in Clips User Guide.
Parameters idx – The index of the fact to retract in the factlist
Returns (int) The retracted fact’s index
Throws IndexError If the fact’s index providen does not exist

2.1.8 pyknow.fact

class pyknow.fact.Fact(*args, **kwargs)

Bases: pyknow.conditionalelement.OperableCE, pyknow.pattern.Bindable, dict
Base Fact class
as_dict()
Return a dictionary containing this Fact data.
copy()
Return a copy of this Fact.
classmethod from_iter(pairs)
has_field_constraints()
has_nested_accessor()
static is_special()
update([E ], **F) → None. Update D from dict/iterable E and F.
If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a
.keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
class pyknow.fact.InitialFact(*args, **kwargs)
Bases: pyknow.fact.Fact

24 Chapter 2. API Documentation

 pyknow Documentation, Release 1.5.0

2.1.9 pyknow.fieldconstraint

class pyknow.fieldconstraint.L
Bases: pyknow.pattern.Bindable, pyknow.fieldconstraint.FieldConstraint
Literal Field Constraint
value
class pyknow.fieldconstraint.W
Bases: pyknow.pattern.Bindable, pyknow.fieldconstraint.FieldConstraint
Wildcard Field Constraint
class pyknow.fieldconstraint.P
Bases: pyknow.pattern.Bindable, pyknow.fieldconstraint.FieldConstraint
Predicate Field Constraint
match

2.1.10 pyknow.rule

class pyknow.rule.Rule
Bases: pyknow.conditionalelement.ConditionalElement
Base CE, all CE are to derive from this class.
This class is used as a decorator, thus provoking __call__ to be called twice:
1. The first call is when the decorator is been created. At this point we assign the function decorated to
self._wrapped and return self to be called the second time.
2. The second call is to execute the decorated function, se we pass all the arguments along.
new_conditions(*args)
Generate a new rule with the same attributes but with the given conditions.

2.1.11 pyknow.strategies

class pyknow.strategies.DepthStrategy(*args, **kwargs)

Bases: pyknow.abstract.Strategy
pyknow.strategies.listdict()
Defaultdict of a list

2.1.12 pyknow.watchers

Watchers are loggers that log detailed information on CLIPS, disabled by default and that can be enabled by the
(watch) method.
Here, we expose a rule, fact and agenda watchers as well as a method to enable/disable them both individually and all
of them.
pyknow.watchers.watch(*what, level=10)
Enable watchers.
Defaults to enable all watchers, accepts a list names of watchers to enable.

2.1. Modules documentation 25

pyknow Documentation, Release 1.5.0

pyknow.watchers.unwatch(*what)
Disable watchers.
Defaults to enable all watchers, accepts a list names of watchers to enable.

2.1.13 pyknow.matchers

2.1.14 pyknow.matchers.rete

RETE algorithm implementation.

This is implemented as described by Charles L. Forgy in his original Ph.D thesis paper. With minor changes to allow
CLIPS like matching and a more pythonic approach.
class pyknow.matchers.rete.ReteMatcher(*args, **kwargs)
Bases: pyknow.abstract.Matcher
RETE algorithm with pyknow matcher interface.
static build_alpha_part(root_node)
Given a set of already adapted rules, build the alpha part of the RETE network starting at root_node.
static build_beta_part(alpha_terminals)
Given a set of already adapted rules, and a dictionary of patterns and alpha_nodes, wire up the beta part of
the RETE network.
build_network()
changes(adding=None, deleting=None)
Pass the given changes to the root_node.
static prepare_ruleset()
Given a KnowledgeEngine, generate a set of rules suitable for RETE network generation.
print_network()
Generate a graphviz compatible graph.
reset()
Reset the matcher memory.
show_network()

2.1.15 pyknow.matchers.rete.abstract

Abstract base classes for the RETE implementation.

class pyknow.matchers.rete.abstract.Check
Bases: object
class pyknow.matchers.rete.abstract.Node
Bases: object
Node interface.
add_child(child, callback)
Add a child to self.children if necessary.
reset()
Reset itself and recursively all its children.

26 Chapter 2. API Documentation

 pyknow Documentation, Release 1.5.0

class pyknow.matchers.rete.abstract.OneInputNode
Bases: pyknow.matchers.rete.abstract.Node
Nodes which only have one input port.
activate(token)
Make a copy of the received token and call self._activate.
class pyknow.matchers.rete.abstract.TwoInputNode
Bases: pyknow.matchers.rete.abstract.Node
Nodes which have two input ports: left and right.
activate_left(token)
Make a copy of the received token and call _activate_left.
activate_right(token)
Make a copy of the received token and call _activate_right.

2.1.16 pyknow.matchers.rete.check

class pyknow.matchers.rete.check.CheckFunction(key_a, key_b, expected, check)

Bases: tuple
check
Alias for field number 3
expected
Alias for field number 2
key_a
Alias for field number 0
key_b
Alias for field number 1
class pyknow.matchers.rete.check.FactCapture
Bases: pyknow.matchers.rete.abstract.Check, pyknow.matchers.rete.check.
_FactCapture
class pyknow.matchers.rete.check.FeatureCheck
Bases: pyknow.matchers.rete.abstract.Check, pyknow.matchers.rete.check.
_FeatureCheck
get_check_function()
staticmethod(function) -> method
Convert a function to be a static method.
A static method does not receive an implicit first argument. To declare a static method, use this idiom:
class C: def f(arg1, arg2, . . . ): . . . f = staticmethod(f)
It can be called either on the class (e.g. C.f()) or on an instance (e.g. C().f()). The instance is ignored
except for its class.
Static methods in Python are similar to those found in Java or C++. For a more advanced concept, see the
classmethod builtin.
class pyknow.matchers.rete.check.SameContextCheck
Bases: pyknow.matchers.rete.abstract.Check

2.1. Modules documentation 27

pyknow Documentation, Release 1.5.0

class pyknow.matchers.rete.check.TypeCheck
Bases: pyknow.matchers.rete.abstract.Check, pyknow.matchers.rete.check.
_TypeCheck
class pyknow.matchers.rete.check.WhereCheck
Bases: pyknow.matchers.rete.abstract.Check, pyknow.matchers.rete.check.
_WhereCheck

2.1.17 pyknow.matchers.rete.dnf

Rewrite engine to get disjuntive normal form of the rules

pyknow.matchers.rete.dnf.dnf(exp)
pyknow.matchers.rete.dnf.unpack_exp(exp, op)

2.1.18 pyknow.matchers.rete.mixins

Mixing classes for the RETE nodes.

class pyknow.matchers.rete.mixins.AnyChild
Bases: object
This node allow any kind of node as a child.
add_child(node, callback)
Add node and callback to the children set.
class pyknow.matchers.rete.mixins.ChildNode(node, callback)
Bases: tuple
Used to store node/callback pair in nodes children set.
callback
Alias for field number 1
node
Alias for field number 0
class pyknow.matchers.rete.mixins.HasMatcher(matcher)
Bases: object
This node need a match callable as parameter.
class pyknow.matchers.rete.mixins.NoMemory
Bases: object
The node has no memory so we have nothing to do.

2.1.19 pyknow.matchers.rete.nodes

RETE nodes implementation.

This are the node types needed by this RETE implementation. Some node types (like ‘The One-input Node for Testing
Variable Bindings) are not needed in this implementation.
class pyknow.matchers.rete.nodes.BusNode
Bases: pyknow.matchers.rete.mixins.AnyChild, pyknow.matchers.rete.mixins.
NoMemory, pyknow.matchers.rete.abstract.Node

28 Chapter 2. API Documentation

 pyknow Documentation, Release 1.5.0

The Bus Node.

The node that reports working memory changes to the rest of the network.
This node cannot be activated in the same manner as the other nodes. No tokens can be sent to it since this is
the node where the first tokens are built.
add(fact)
Create a VALID token and send it to all children.
remove(fact)
Create an INVALID token and send it to all children.
class pyknow.matchers.rete.nodes.ConflictSetNode(rule)
Bases: pyknow.matchers.rete.mixins.AnyChild, pyknow.matchers.rete.abstract.
OneInputNode
Conflict Set Change Node.
This node is the final step in the network. Any token activating this node will produce an activation (VALID
token) or deactivation (INVALID token) of the internal rule with the token context and facts.
get_activations()
Return a list of activations.
class pyknow.matchers.rete.nodes.FeatureTesterNode(matcher)
Bases: pyknow.matchers.rete.mixins.AnyChild, pyknow.matchers.rete.mixins.
HasMatcher, pyknow.matchers.rete.mixins.NoMemory, pyknow.matchers.rete.
abstract.OneInputNode
Feature Tester Node.
This node implementation represents two different nodes in the original paper: The One-input Node for Testing
Constant Features and The One-input Node for Testing Variable Bindings.
The trick here is this node receives a callable object at initilization time and uses it for testing the received tokens
on activation. The given callable can return one of the following things:
• Boolean:
– True: The test pass. The token will be sent to the children nodes.
– False: The test failed. Do nothing.
• Mapping (dict):
– With content: The test pass. In addition the pairs key-value will be added to the token context.
– Empty: The test failed. Do nothing.
The only exception here is when the callable returns a mapping with some key and some value, and the current
context of the token also have an entry for this key but with a different value. In this case the test do not pass.
class pyknow.matchers.rete.nodes.NotNode(matcher)
Bases: pyknow.matchers.rete.mixins.AnyChild, pyknow.matchers.rete.mixins.
HasMatcher, pyknow.matchers.rete.abstract.TwoInputNode
Not Node.
This is a special kind of node representing the absence of some fact/condition.
This node is similar to OrdinaryMatchNode in the sense it has two input ports and try to match tokens arriving
in both of them. But pass VALID tokens to the children when no matches are found and INVALID tokens when
they are.

2.1. Modules documentation 29

pyknow Documentation, Release 1.5.0

class pyknow.matchers.rete.nodes.OrdinaryMatchNode(matcher)
Bases: pyknow.matchers.rete.mixins.AnyChild, pyknow.matchers.rete.mixins.
HasMatcher, pyknow.matchers.rete.abstract.TwoInputNode
Ordinary Two-input Node.
This kind of node receive tokens at two ports (left and right) and try to match them.
The matching function is a callable given as a parameter to __init__ and stored internally. This functions will
receive two contexts, one from the left and other from the right, and decides if they match together (returning
True or False).
Matching pairs will be combined in one token containing facts from both and a combined context. This com-
bined tokens will be sent to all children.
class pyknow.matchers.rete.nodes.WhereNode(matcher)
Bases: pyknow.matchers.rete.mixins.AnyChild, pyknow.matchers.rete.mixins.
HasMatcher, pyknow.matchers.rete.mixins.NoMemory, pyknow.matchers.rete.
abstract.OneInputNode
Check some conditions over a token context.

2.1.20 pyknow.matchers.rete.token

Token object and related objects needed by the RETE algorithm.

class pyknow.matchers.rete.token.Token
Bases: pyknow.matchers.rete.token._Token
Token, as described by RETE but with context.
class TagType
Bases: enum.Enum
Types of Token TAG data.
INVALID = False
VALID = True
copy()
Make a new instance of this Token.
This method makes a copy of the mutable part of the token before making the instance.
classmethod invalid(data, context=None)
Shortcut to create an INVALID Token.
is_valid()
Test if this Token is VALID.
to_info()
Create and return an instance of TokenInfo with this token.
This is useful, for example, to use this token information as a dictionary key.
classmethod valid(data, context=None)
Shortcut to create a VALID Token.
class pyknow.matchers.rete.token.TokenInfo
Bases: pyknow.matchers.rete.token._TokenInfo
Tag agnostig version of Token with inmutable data.

30 Chapter 2. API Documentation

 pyknow Documentation, Release 1.5.0

to_invalid_token()
Create an INVALID token using this data.
to_valid_token()
Create a VALID token using this data.

2.1.21 pyknow.matchers.rete.utils

pyknow.matchers.rete.utils.extract_facts(rule)
Given a rule, return a set containing all rule LHS facts.
pyknow.matchers.rete.utils.generate_checks(fact)
Given a fact, generate a list of Check objects for checking it.
pyknow.matchers.rete.utils.prepare_rule(exp)
Given a rule, build a new one suitable for RETE network generation.
Meaning:
1. Rule is in disjuntive normal form (DNF).
2. If the rule is empty is filled with an InitialFact.
3. If the rule starts with a NOT, an InitialFact is prepended.
4. If any AND starts with a NOT, an InitialFact is prepended.
#. If the rule is an OR condition, each NOT inside will be converted to AND(InitialFact(), NOT(. . . ))
pyknow.matchers.rete.utils.wire_rule(rule, alpha_terminals, lhs=None)

2.1.22 pyknow.operator

The operator module contains a set of predicate functions constructors based on the P() field constraint.
These operators can be composed together and binded like normal Field Constraints.
Example:

>>> WIDTH = 640

>>> HEIGHT = 480
>>>
>>> class Player(Fact):
... pass
...
>>>
>>> @Rule(
... Player(
... x=MATCH.x & GE(0) & LE(WIDTH),
... y=MATCH.y & BETWEEN(0, HEIGHT),
... name=MATCH.name & (CALL.startswith("@") | CALL.endswith("_ADM"))
... )
... )
... def admin_in_visible_area(self, x, y, name):
... pass
...

pyknow.operator.TRUTH = P(<class 'bool'>,)

Return True if obj is true, and False otherwise. This is equivalent to using the bool constructor.

2.1. Modules documentation 31

pyknow Documentation, Release 1.5.0

pyknow.operator.LT(b)
Less than operator.
pyknow.operator.LE(b)
Less than or equal operator.
pyknow.operator.EQ(b)
Equal operator.
pyknow.operator.NE(b)
Not equal operator.
pyknow.operator.GE(b)
Greater than or equal operator.
pyknow.operator.GT(b)
Greater than operator.
pyknow.operator.IS(b)
Tests object identity.
pyknow.operator.IS_NOT(b)
Tests object identity.
pyknow.operator.CONTAINS(b)
Return the outcome of the test b in a
pyknow.operator.BETWEEN(a, b)
The BETWEEN operator selects values within a given range. The BETWEEN operator is inclusive: begin and
end values are included.
pyknow.operator.REGEX(pattern, flags=0)
Regular expression matching.
pyknow.operator.LIKE(pattern)
Unix shell-style wildcards. Case-sensitive
pyknow.operator.ILIKE(pattern)
Unix shell-style wildcards. Case-insensitive

32 Chapter 2. API Documentation

 CHAPTER 3

Release History

3.1 1.5.0

• Added Python version 3.7 to tox.ini.

• Monkey and bananas example.
• Fixed bug, numeric index args gets repeated in a weird way introduced in 1.4.0.
• Pass only the defined args in absence of kwargs.

3.2 1.4.0

• Order integer facts keys after making a copy.

• as_dict method for Fact.
• freeze and unfreeze method documentation.
• unfreeze method in pyknow.utils.
• Zebra example from Clips.

3.3 1.3.0

• pyknow.operator module.
• Nested matching.
• Added Talk ‘Sistemas Expertos en Python con PyKnow - PyConES 2017’ to docs folder.

33
pyknow Documentation, Release 1.5.0

3.4 1.2.0

• Freeze fact values as the default behavior to address Issue #9.

• Added pyknow.utils.anyof to mitigate Issue #7.
• Raise RuntimeError if a fact value is modified after declare().
• Added MATCH and AS objects.

3.5 1.1.1

• Removing the borg optimization for P field constraints.

• Use the hash of the check in the sorting of the nodes to always generate the same alpha part of the network.

3.6 1.1.0

• Allow any kind of callable in Predicate Field Constraints (P()).

3.7 1.0.1

• DNF of OR clause inside AND or Rule was implemented wrong.

3.8 1.0.0

• RETE matching algorithm.

• Better Rule decorator system.
• Facts are dictionaries.
• Documentation.

3.9 <1.0.0

• Unestable API.
• Wrong matching algorithm.
• Bad performance
• PLEASE DON’T USE THIS.

34 Chapter 3. Release History

 CHAPTER 4

Source documentation

• genindex
• modindex
• search

35
pyknow Documentation, Release 1.5.0

36 Chapter 4. Source documentation

 Python Module Index

p
pyknow, 21
pyknow.abstract, 21
pyknow.activation, 21
pyknow.agenda, 22
pyknow.conditionalelement, 22
pyknow.engine, 22
pyknow.fact, 24
pyknow.factlist, 23
pyknow.fieldconstraint, 25
pyknow.matchers, 26
pyknow.matchers.rete, 26
pyknow.matchers.rete.abstract, 26
pyknow.matchers.rete.check, 27
pyknow.matchers.rete.dnf, 28
pyknow.matchers.rete.mixins, 28
pyknow.matchers.rete.nodes, 28
pyknow.matchers.rete.token, 30
pyknow.matchers.rete.utils, 31
pyknow.operator, 31
pyknow.rule, 25
pyknow.strategies, 25
pyknow.watchers, 25

37
pyknow Documentation, Release 1.5.0

38 Python Module Index

 Index

A ChildNode (class in pyknow.matchers.rete.mixins), 28

activate() (pyknow.matchers.rete.abstract.OneInputNode ConflictSetNode (class in pyknow.matchers.rete.nodes),
method), 27 29
activate_left() (pyknow.matchers.rete.abstract.TwoInputNodeCONTAINS() (in module pyknow.operator), 32
method), 27 copy() (pyknow.fact.Fact method), 24
copy() (pyknow.matchers.rete.token.Token method), 30
activate_right() (pyknow.matchers.rete.abstract.TwoInputNode
method), 27
Activation (class in pyknow.activation), 21
D
add() (pyknow.matchers.rete.nodes.BusNode method), 29 declare() (pyknow.engine.KnowledgeEngine method), 22
add_child() (pyknow.matchers.rete.abstract.Node declare() (pyknow.factlist.FactList method), 24
method), 26 DepthStrategy (class in pyknow.strategies), 25
add_child() (pyknow.matchers.rete.mixins.AnyChild dnf() (in module pyknow.matchers.rete.dnf), 28
method), 28 duplicate() (pyknow.engine.KnowledgeEngine method),
Agenda (class in pyknow.agenda), 22 23
AND (class in pyknow.conditionalelement), 22
AnyChild (class in pyknow.matchers.rete.mixins), 28
E
as_dict() (pyknow.fact.Fact method), 24 EQ() (in module pyknow.operator), 32
EXISTS (class in pyknow.conditionalelement), 22
B expected (pyknow.matchers.rete.check.CheckFunction at-
tribute), 27
BETWEEN() (in module pyknow.operator), 32
extract_facts() (in module pyknow.matchers.rete.utils), 31
build_alpha_part() (pyknow.matchers.rete.ReteMatcher
static method), 26 F
build_beta_part() (pyknow.matchers.rete.ReteMatcher
Fact (class in pyknow.fact), 24
static method), 26
FactCapture (class in pyknow.matchers.rete.check), 27
build_network() (pyknow.matchers.rete.ReteMatcher
FactList (class in pyknow.factlist), 23
method), 26
FeatureCheck (class in pyknow.matchers.rete.check), 27
BusNode (class in pyknow.matchers.rete.nodes), 28
FeatureTesterNode (class in py-
know.matchers.rete.nodes), 29
C FORALL (class in pyknow.conditionalelement), 22
callback (pyknow.matchers.rete.mixins.ChildNode from_iter() (pyknow.fact.Fact class method), 24
attribute), 28
changes (pyknow.factlist.FactList attribute), 24 G
changes() (pyknow.abstract.Matcher method), 21 GE() (in module pyknow.operator), 32
changes() (pyknow.matchers.rete.ReteMatcher method), generate_checks() (in module py-
26 know.matchers.rete.utils), 31
Check (class in pyknow.matchers.rete.abstract), 26 get_activations() (pyknow.engine.KnowledgeEngine
check (pyknow.matchers.rete.check.CheckFunction at- method), 23
tribute), 27 get_activations() (pyknow.matchers.rete.nodes.ConflictSetNode
CheckFunction (class in pyknow.matchers.rete.check), 27 method), 29

39
pyknow Documentation, Release 1.5.0

get_check_function() (py- node (pyknow.matchers.rete.mixins.ChildNode attribute),

know.matchers.rete.check.FeatureCheck 28
method), 27 NoMemory (class in pyknow.matchers.rete.mixins), 28
get_deffacts() (pyknow.engine.KnowledgeEngine NOT (class in pyknow.conditionalelement), 22
method), 23 NotNode (class in pyknow.matchers.rete.nodes), 29
get_next() (pyknow.agenda.Agenda method), 22
get_rules() (pyknow.engine.KnowledgeEngine method), O
23 OneInputNode (class in pyknow.matchers.rete.abstract),
GT() (in module pyknow.operator), 32 26
OR (class in pyknow.conditionalelement), 22
H OrdinaryMatchNode (class in py-
halt() (pyknow.engine.KnowledgeEngine method), 23 know.matchers.rete.nodes), 29
has_field_constraints() (pyknow.fact.Fact method), 24
has_nested_accessor() (pyknow.fact.Fact method), 24 P
HasMatcher (class in pyknow.matchers.rete.mixins), 28 P (class in pyknow.fieldconstraint), 25
prepare_rule() (in module pyknow.matchers.rete.utils), 31
I prepare_ruleset() (pyknow.matchers.rete.ReteMatcher
ILIKE() (in module pyknow.operator), 32 static method), 26
InitialFact (class in pyknow.fact), 24 print_network() (pyknow.matchers.rete.ReteMatcher
INVALID (pyknow.matchers.rete.token.Token.TagType method), 26
attribute), 30 pyknow (module), 21
invalid() (pyknow.matchers.rete.token.Token class pyknow.abstract (module), 21
method), 30 pyknow.activation (module), 21
IS() (in module pyknow.operator), 32 pyknow.agenda (module), 22
IS_NOT() (in module pyknow.operator), 32 pyknow.conditionalelement (module), 22
is_special() (pyknow.fact.Fact static method), 24 pyknow.engine (module), 22
is_valid() (pyknow.matchers.rete.token.Token method), pyknow.fact (module), 24
30 pyknow.factlist (module), 23
pyknow.fieldconstraint (module), 25
K pyknow.matchers (module), 26
key_a (pyknow.matchers.rete.check.CheckFunction at- pyknow.matchers.rete (module), 26
tribute), 27 pyknow.matchers.rete.abstract (module), 26
key_b (pyknow.matchers.rete.check.CheckFunction at- pyknow.matchers.rete.check (module), 27
tribute), 27 pyknow.matchers.rete.dnf (module), 28
KnowledgeEngine (class in pyknow.engine), 22 pyknow.matchers.rete.mixins (module), 28
pyknow.matchers.rete.nodes (module), 28
L pyknow.matchers.rete.token (module), 30
L (class in pyknow.fieldconstraint), 25 pyknow.matchers.rete.utils (module), 31
LE() (in module pyknow.operator), 32 pyknow.operator (module), 31
LIKE() (in module pyknow.operator), 32 pyknow.rule (module), 25
listdict() (in module pyknow.strategies), 25 pyknow.strategies (module), 25
LT() (in module pyknow.operator), 31 pyknow.watchers (module), 25

M R
REGEX() (in module pyknow.operator), 32
match (pyknow.fieldconstraint.P attribute), 25
remove() (pyknow.matchers.rete.nodes.BusNode
Matcher (class in pyknow.abstract), 21
method), 29
modify() (pyknow.engine.KnowledgeEngine method), 23
reset() (pyknow.abstract.Matcher method), 21
N reset() (pyknow.engine.KnowledgeEngine method), 23
reset() (pyknow.matchers.rete.abstract.Node method), 26
NE() (in module pyknow.operator), 32 reset() (pyknow.matchers.rete.ReteMatcher method), 26
new_conditions() (pyknow.rule.Rule method), 25 ReteMatcher (class in pyknow.matchers.rete), 26
Node (class in pyknow.matchers.rete.abstract), 26 retract() (pyknow.engine.KnowledgeEngine method), 23
retract() (pyknow.factlist.FactList method), 24

40 Index
 pyknow Documentation, Release 1.5.0

Rule (class in pyknow.rule), 25

run() (pyknow.engine.KnowledgeEngine method), 23

S
SameContextCheck (class in py-
know.matchers.rete.check), 27
show_network() (pyknow.matchers.rete.ReteMatcher
method), 26
Strategy (class in pyknow.abstract), 21

T
TEST (class in pyknow.conditionalelement), 22
to_info() (pyknow.matchers.rete.token.Token method), 30
to_invalid_token() (pyknow.matchers.rete.token.TokenInfo
method), 30
to_valid_token() (pyknow.matchers.rete.token.TokenInfo
method), 31
Token (class in pyknow.matchers.rete.token), 30
Token.TagType (class in pyknow.matchers.rete.token), 30
TokenInfo (class in pyknow.matchers.rete.token), 30
TRUTH (in module pyknow.operator), 31
TwoInputNode (class in pyknow.matchers.rete.abstract),
27
TypeCheck (class in pyknow.matchers.rete.check), 27

U
unpack_exp() (in module pyknow.matchers.rete.dnf), 28
unwatch() (in module pyknow.watchers), 25
update() (pyknow.fact.Fact method), 24
update_agenda() (pyknow.abstract.Strategy method), 21

V
VALID (pyknow.matchers.rete.token.Token.TagType at-
tribute), 30
valid() (pyknow.matchers.rete.token.Token class method),
30
value (pyknow.fieldconstraint.L attribute), 25

W
W (class in pyknow.fieldconstraint), 25
watch() (in module pyknow.watchers), 25
WhereCheck (class in pyknow.matchers.rete.check), 28
WhereNode (class in pyknow.matchers.rete.nodes), 30
wire_rule() (in module pyknow.matchers.rete.utils), 31

Index 41