Jason-style AgentSpeak for Python

Origin / Origen

English. This package is a Python interpreter for AgentSpeak, the agent-oriented programming language used by Jason, a well-known BDI (belief-desire-intention) multi-agent platform written in Java. It is a fork of niklasf/python-agentspeak (previously published on PyPI as agentspeak), maintained at farell14/python-agentspeak and published here as py-agentspeak while changes make their way back upstream.

Español. Este paquete es un intérprete en Python de AgentSpeak, el lenguaje de programación orientado a agentes que usa Jason, una conocida plataforma multi-agente BDI (creencias-deseos-intenciones) escrita en Java. Es un fork de niklasf/python-agentspeak (publicado antes en PyPI como agentspeak), mantenido en farell14/python-agentspeak y publicado aquí como py-agentspeak mientras los cambios llegan al repositorio original.

Key Features

• Jason-style AgentSpeak interpretation
• Easy integration with existing Python code
• Support for complex agent constructions

Setup / Instalación

English:

pip install py-agentspeak

Español:

pip install py-agentspeak

Requirements / Requisitos

• Python 3.8 or higher / Python 3.8 o superior
• Additional dependencies are automatically installed / Las dependencias adicionales se instalan automáticamente

Usage example / Ejemplo de uso

!hello_world.

+!hello_world <-
  .print("Hello world!").

Usage / Uso

English. Run a standalone agent program:

$ python -m agentspeak examples/hello_world.asl

Run an interactive console:

$ python -m agentspeak

See examples/embedded for an example that interfaces with custom Python code.

Español. Ejecutar un programa de agente independiente:

$ python -m agentspeak examples/hello_world.asl

Ejecutar una consola interactiva:

$ python -m agentspeak

Revisa examples/embedded para un ejemplo de integración con código Python propio.

Jason compability

python-agentspeak should be mostly equivalent to Jason.

• Plan annotations are parsed and used as plan labels (e.g. for .untellHow), but behavior annotations such as atomic or breakpoint are not interpreted.
• Standard library does not yet contain syntactic transformations with {begin ...} and {end}.
• Standard library contains some, but not all, introspective and plan-manipulation actions. Implemented: .current_intention, .intend, .drop_intention, .succeed_goal, .drop_all_intentions, .add_plan, .remove_plan, .relevant_plans, .plan_label. Not yet implemented: .fail_goal, event/desire introspection (.desire, .drop_event, .drop_all_events, .drop_desire, .drop_all_desires), and agent lifecycle actions (.create_agent, .kill_agent, .at, .perceive).
• .send(Receiver, askOne, Query, Answer) and .send(Receiver, askAll, Query, Answer) query the receiver's belief base directly and bind the answer in the same reasoning step, rather than actually dispatching an asynchronous message and suspending the sender until a reply arrives.
• Jason 2.0 fork join operators are tokenized by the lexer but not yet parsed/supported.
• Literals are only comparable if they have the same signature.