Emotiv Cortex2 Python Client

Author: methylDragon

---

Image sources: Emotiv

Introduction

Unofficial Python client for the Emotiv EEG Cortex 2 API.

Features the entire JSON-RPC API communicated via asynchronous websockets for speed.

The Cortex 2 app is used to host a websocket web server gateway interface that takes JSON requests and returns JSON data.

This Python client is designed to be a wrapper client for said API. It streams multi-session sensor data using a separate asynchronous thread!

API Reference: https://emotiv.gitbook.io/cortex-api/

PyPI Link: https://pypi.org/manage/project/cortex2

Support my efforts!

Or leave a tip! ヾ(°∇°*)

Requirements

• Python 3.6 or above

Installation

pip install cortex2

Example Usage

Remember to set up your client ID and secret by registering

Also ensure you've started the EmotivApp, since it hosts the websocket server! (You'll probably have to use Windows or Mac.)

If the client fails to connect, you might have to restart the script using the client.

from cortex2 import EmotivCortex2Client

url = "wss://localhost:6868"

# Remember to start the Emotiv App before you start!
# Start client with authentication
client = EmotivCortex2Client(url,
                             client_id='CLIENT_ID',
                             client_secret="CLIENT_SECRET",
                             check_response=True,
                             authenticate=True,
                             debug=False)

# Test API connection by using the request access method
client.request_access()

# Explicit call to Authenticate (approve and get Cortex Token)
client.authenticate()

# Connect to headset, connect to the first one found, and start a session for it
client.query_headsets()
client.connect_headset(0)
client.create_session(0)

# Subscribe to the motion and mental command streams
# Spins up a separate subscription thread
client.subscribe(streams=["mot", "com"])

# Test message handling speed
a = client.subscriber_messages_handled
time.sleep(5)
b = client.subscriber_messages_handled
print((b - a) / 5)

# Grab a single instance of data
print(client.receive_data())

# Continously grab data, while making requests periodically
while True:
    counter += 1
    # time.sleep(0.1)

    if counter % 5000 == 0:
        print(client.request_access())

    # Try stopping the subscriber thread
    if counter == 50000:
        client.stop_subscriber()
        break

    try:
        # Check the latest data point from the motion stream, from the first session
        print(list(client.data_streams.values())[0]['mot'][0])
    except:
        pass

You can also connect by explicitly stating IDs!

client.connect_headset(headset_id="EPOCPLUS-3B9AXXXX")
client.create_session(headset_id="EPOCPLUS-3B9AXXXX")

You can do a lot more! The entire API is covered, and everything generally works the same way.

Additional Notable Features

• The websockets API is wrapped with helper functions to do request ID checking, asynchronous message handling, and some basic data sanitisation
• Threads are used to try to allow for subscription to multiple sessions and multiple headsets
  • The threads have a helper function that automatically splits the data streams so you may work on the streams independently as opposed to all lumped together as one object
• Automatic syncing helper methods for detecting existing sessions and headsets will fire off when the relevant methods are called. (Eg. when trying to update, connect, or disconnect headsets, the client will automatically query and update the dictionary of seen headsets.)

Notable Class Attributes

Name	Type	Description	Example
debug	bool	Debug flag. Slows down data subscription rate if set.
client_id	str	Client ID
client_secret	str	Client Secret
cortex_token	str	Cortex Token
approved	bool	True if access is granted
authorized	bool	True if cortex token is issued
headsets	OrderedDict()	OrderedDict of headset object dicts seen
connected_headsets	OrderedDict()	OrderedDict of connected headset object dicts
sessions	OrderedDict()	OrderedDict of seen session object dicts
data_deque_size	int	Maximum size of each subscriber data buffer
data_streams	dict	Sensor data streams keyed by session	{session_id: {data_stream_deques_types: [data]}, ...}
subscribed_streams	dict	Data stream names and descriptions that are subscribed, keyed by session	{session_id: {stream_names: info}, ...}
subscriber_spinning	bool	Tracks whether the subscriber thread is running
subscriber_messages_handled	int	Counts how many subscriber messages have been handled
subscriber_reading	threading.Event()	Event flag to pause the subscriber without killing it

Class Methods

Class Helpers

• send_authed_request(method=None, params={}, request=None)
• send_request(method=None, params={}, request=None)
• set_client_id(client_id)
• set_client_secret(client_secret)
• _verify_key(key, response, field=None)
• _exit_handler(\_signo, _stack_frame)

---

Connecting to Cortex API

• get_cortex_info()

---

Authentication

• get_user_login()

• request_access()

• has_access_right()

• authorize()

• generate_new_token()

• get_user_information()

• get_license_info()

• Helpers

  • authenticate()

---

Headsets

• query_headsets(id=None, sync=True)

• control_device(command, headset_id=None, mappings=None)

• refresh_headsets()

• connect_headset(headset_id_idx=0, headset_id=None, mappings=None)

• disconnect_headset(headset_id)

• update_headset(settings, headset_id_idx=0, headset_id=None)

• maximise_headset(headset_id_idx=0, headset_id=None)

• maximise_headset_no_motion(headset_id_idx=0, headset_id=None)

• Helpers

  • sync_headsets(response=None, query=True)
  • _get_headset_type(headset_id)

---

Sessions

• create_session(headset_id_idx=0, headset_id=None)

• create_activated_session(headset_id_idx=0, headset_id=None)

• update_session(status, session_id_idx=0, session_id=None)

• activate_session(session_id_idx=0, session_id=None)

• close_session(session_id_idx=0, session_id=None)

• query_sessions(sync=True)

• Helpers

  • sync_sessions(response=None, query=True)

---

Data Subscription

• subscribe(streams, session_id_idx=0, session_id=None)

• unsubscribe(streams, session_id_idx=0, session_id=None)

• Helpers

  • receive_data()
  • spin_subscriber()
  • stop_subscriber()
  • pause_subscriber()
  • resume_subscriber()
  • _subscriber_thread()
  • _create_data_streams(session_id)
  • _delete_data_streams(session_id)
  • _split_and_update_data_streams(data_sample)

---

Records

• create_record(title, session_id_idx=0, session_id=None, description=None, subject_name=None, tags=None)
• stop_record(title, session_id_idx=0, session_id=None)
• update_record(record_id, description=None, tags=None)
• delete_record(record_ids)
• export_record(record_ids, folder, format, stream_types, version=None)
• export_edf(record_ids, folder, stream_types)
• export_csv_v1(record_ids, folder, stream_types)
• export_csv_v2(record_ids, folder, stream_types)
• query_records(query, order_by={'startDatetime': "DESC"}, limit=0, offset=0, include_markers=False)
• get_record_infos(record_ids)

---

Markers

• inject_marker(time, value, label, session_id_idx=0, session_id=None, port=None, extras=None)
• update_marker(marker_id, time, session_id_idx=0, session_id=None, extras=None)

---

BCI

• query_profiles()
• get_current_profile(headset_id_idx=0, headset_id=None)
• setup_profile(status, profile, headset_id_idx=0, headset_id=None, new_profile_name=None)
• create_profile(profile)
• load_profile(profile, headset_id_idx=0, headset_id=None)
• unload_profile(profile, headset_id_idx=0, headset_id=None)
• save_profile(profile)
• rename_profile(profile, headset_id_idx=0, headset_id=None, new_profile_name=None)
• delete_profile(profile)
• load_guest_profile(headset_id_idx=0, headset_id=None)
• get_detection_info(detection)
• get_mental_command_info()
• get_facial_expression_info()
• training(detection, status, action, session_id_idx=0, session_id=None)
• start_training(detection, action, session_id_idx=0, session_id=None)
• accept_training(detection, action, session_id_idx=0, session_id=None)
• reject_training(detection, action, session_id_idx=0, session_id=None)
• reset_training(detection, action, session_id_idx=0, session_id=None)
• erase_training(detection, action, session_id_idx=0, session_id=None)

---

Advanced BCI

• get_trained_signature_actions(detection, profile=None, session=None)
• get_training_time(detection, session_id_idx=0, session_id=None)
• facial_expression_signature_type(status, profile=None, session=None, signature=None)
• get_facial_expression_signature_type(profile=None, session=None)
• set_facial_expression_signature_type(profile=None, session=None, signature=None)
• facial_expression_threshold(status, action, profile=None, session=None, value=None)
• get_facial_expression_threshold(action, profile=None, session=None)
• set_facial_expression_threshold(action, profile=None, session=None, value=None)
• mental_command_active_action(status, profile=None, session=None, actions=[])
• get_mental_command_active_action(profile=None, session=None)
• set_mental_command_active_action(profile=None, session=None, actions=[])
• mental_command_brain_map(profile=None, session=None)
• mental_command_get_skill_rating(profile=None, session=None, action=None)
• mental_command_training_threshold(profile=None, session=None)
• mental_command_action_level(status, profile=None, session=None, level=None)
• get_mental_command_action_level(profile=None, session=None)
• set_mental_command_action_level(profile=None, session=None, level=None)
• mental_command_action_sensitivity(status, profile=None, session=None, values=[])
• get_mental_command_action_sensitivity(status, profile=None, session=None)
• set_mental_command_action_sensitivity(status, profile=None, session=None, values=[])

EmotivCortex2Client Changelog

v1.0.1 - 2019_07_09

Changed

• Fixed submodule import bug

v1.0.0 - 2019_07_09

Added

• Initial release:
  • WebsocketClient with tested asynchronicity
  • EmotivCortex2Client with full API