Opencog services to SingularityNET

This repository contains a number of SingularityNET services which uses Opencog resources.

Actually, we have only one C++ gRPC service which is used to call any of the implemented services passing a command name and a list of arguments.

C++ and Scheme services are implemented in src/cpp-services and src/scm-services respectively.

• Echo - a simple 'echo' which just outputs its input, used to test the architecture and to serve as example.
• EchoScheme - the same, but written in Scheme.
• Miner - load a knowledge base and run Opencog's PatternMiner
• Ghost - handle several OpenCog servers locally, to operate MindAgents, and allow to handle GHOST agents in exclusive sessions.

HTML User's Guide Hub

Getting Started

In addition to the services themselves, this repository contains all the basic code and scripts required to build and deploy the Opencog services.

Building and testing services locally (without blockchain)

The below setup requires docker and will build docker images with all the required packages. Having a docker container is useful because the Opencog Services uses C++ gRPC bindings which requires a lot of pre-requisite packages to be installed.

git clone [email protected]:singnet/opencog-services.git
cd opencog-services
# make sure you have the latest base image
docker pull opencog/opencog-dev:cli
docker image build -t opencog_services_basic -f basic-dockerfile .
docker image build -t opencog_services -f Dockerfile .
docker run --name OPENCOG_SERVICE_DEV -ti opencog_services /bin/bash

Find server and client executables in bin/

You can run tests locally. Start a server

$ ./bin/server &

and make any number of requests using client

$ ./bin/client sync Echo foo bar
foo bar

The command line above is requesting the execution of the service Echo, passing two arguments foo and bar. The service just echoes the passed input.

The keyword sync before the service name states that the call should be synchronous. So the client command will wait for the service to finish before returning. You can make an asynchronous call using

$ ./bin/client async Echo foo bar
Task "Echo" started. See results in: http://54.203.198.53:7000/ServiceAsynchronousOutput/opencog/Echo-53171

In this case the service return an URL which can be acessed asynchronously to get the results (when they are ready). This is specially useful when running services that take too long to execute.

Important note: async calls are setup to work only in the deployment container. To test asynchronous commands locally you need to change src/server.cc:

#define ASYNCHRONOUS_API_OUTPUT_URL 
#define ASYNCHRONOUS_API_OUTPUT_DIR 

In the deployment environment, results remain available for 24 hours after being delivered. Then they are deleted.

Writing a new Opencog service

This tutorial explains how to write a new Opencog-based service.

Contributing and Reporting Issues

Please read our guidelines before submitting issues or pull requests.

Authors

• Andre Senna - Maintainer - SingularityNET

License

This project is licensed under the MIT License - see the LICENSE file for details