noslaver/neuron

A python Brain-Computer interface

neuron

A python Brain-Computer interface

Installation

1. Clone the repository and enter it:

$ git clone git@github:noslaver/neuron.git
...
$ cd neuron/

2. Run the installation script and activate the virtual environment:

$ ./scripts/install.sh
...
$ source .env/bin/activate
[neuron] $ # you're good to go!

3. To check that everything is working as expected, run the tests:

$ pytest tests
...

Deployment

neuron can be fully deployed in a docker environment, using the run-pipeline.sh script, available in the scripts directory.

After deployment, the server is available at localhost:8000, the API is available at locahost:5000 and the GUI is available at localhost:8080.

Usage

Client

neuron's client allows publishing a mind sample to a server. The client is available as both an API and a CLI.

API usage:

from neuron.client import upload_sample

upload_sample(host='127.0.0.1', port=8000, path='sample.mind.gz')

CLI usage:

$ python -m neuron.client upload-sample \
        -H/--host '127.0.0.1' \
        -p/--port 8000 \
        'snapshot.mind.gz'

Server

neuron's server receives mind samples uploaded by clients and publishes them. Clients should send a POST request to /users/<user_id>/snapshots. The request body should contain a Protobuf message of type SnapshotInfo, as described in neuron/protobuf/neuron.proto. The exact publish method is configurable. By default, received samples are published to a RabbitMQ message bus.

Note that large binary data is saved to local storage, and only its location is published.

The server is available as both an API and a CLI.

API usage:

from neuron.server import run_server

def handle_sample(sample):
    print(sample)

upload_sample(host='127.0.0.1', port=8000, publish=handle_sample)

CLI usage:

$ python -m neuron.client run-server \
        -H/--host '127.0.0.1' \
        -p/--port 8000 \
        'rabbitmq://127.0.0.1:5672'

Parsers

neuron's parsers process incoming mind samples. Each parser specifies which part of the sample it is interested and returns a processed result, which based on the interface used, can be published to a message queue or output to sdtout.

API usage:

from cortex.parsers import parse
data = ...
result = parse('pose', data)

CLI usage: There are two usage, one to parse a file containing raw data:

$ python -m cortex.parsers parse 'pose' 'snapshot.raw' > 'pose.result'

The other subscribes to a message queue and publishes the processed result back to the broker:

$ python -m cortex.parsers run_parser 'pose' 'rabbitmq://127.0.0.1:5672'

Adding a new parser

To add a new parsers, simply add a new .py file into neuron/parsers. You can expose your parser either with a class or with a function. To expose as a class, your file should contain a class whose name ends with "Parser". The class should have a parse(self, context, snapshot) method. The class should also have a attribute named field which is a string indicating which parts of the snapshot it should be passed (e.g. pose).

To expose as a function, your file should contain a function whose name begins with "parse". The function signature should be parse_xxx(context, snapshot). You should add an attribute to the function named field which is a string indicating which parts of the snapshot it should be passed (e.g. pose).

Saver

neuron's saver allows saving processed mind data, published by the parsers, into databases.

API usage:

from cortex.saver import Saver
saver = Saver(database_url)
data = ...
saver.save('pose', data)

CLI usage: There are two usage, one to save a file containing processed data:

$ python -m cortex.saver save \
    -d/--database 'mongodb://127.0.0.1:27017' \
    'pose' \
    'pose.result'

The other subscribes to a message queue and saves published processed data to the database:

$ python -m cortex.saver run_saver \
    'mongodb://127.0.0.1:27017' \
    'rabbitmq://127.0.0.1:5672'

API

neuron's API allows a runs a server which exposes the processed data to an endpoint user via HTTP.

API usage:

from cortex.api import run_api_server
run_api_server(
    host = '127.0.0.1',
    port = 8000,
    database_url = 'mongodb://127.0.0.1:27017' 
)

CLI usage:

$ python -m cortex.api run-server \
    -H/--host '127.0.0.1' \
    -p/--port 5000 \
    -d/--database 'mongodb://127.0.0.1:27017'

The API exposes the following endpoints:

• GET /users

  Returns the list of all the supported users.

• GET /user/{user_id}

  Returns the specified user's info.

• GET /user/{user_id}/snapshots

  Returns the list of the specified user's snapshots.

• GET /user/{user_id}/snapshots/{snapshot_id}

  Returns the specified snapshot's info. Note that a snapshot is identified by its timestamp.

• GET /user/{user_id}/snapshots/{snapshot_id}/{results-name}

  Returns the specified snapshot's result.

CLI

neuron's CLI consumes neuron's API.

CLI usage:

$ python -m cortex.cli get-users

$ python -m cortex.cli get-user 42

$ python -m cortex.cli get-snapshot 42 1575446890088

$ python -m cortex.cli get-snapshot 42 1575446890088 pose

GUI

neuron's GUI visualizes the API.

API usage:

from cortex.gui import run_server
run_api_server(
    host = '127.0.0.1',
    port = 8080,
    api_host = '127.0.0.1',
    api_port = 5000,
)

CLI usage:

$ python -m cortex.gui run-server \
    -h/--host '127.0.0.1' \
    -p/--port 8080 \
    -H/--api-host '127.0.0.1' \
    -P/--api-port 5000

neuron's Issues

Complete documentation

Add __repr__ methods to every exported object