RasaHost

A user interfaces for Rasa NLU and Rasa Core, that simplify bot development.

Features

Editor for MD files (intents, stories, domain) in Rasa format
- Getting started is beyond easy, you just have to specify paths to your MD files.
- No migration is needed. The tool uses standard Rasa format.
- It does not have a dependency on the Rasa version.
Logs conversations
- All logs and conversations are saved in SQLite.
- You just have to create an agent and bind it to the host.
- The logging can be mixed with standard logging to files and console-like Rasa does by default.
- It does not have a dependency on the Rasa version.
Testing
- Chat control
- Memoization Policy visualization

Tech

python
flask
vuejs

Installation

pip install rasa-host

Running

Rasa-Host does not have a dependency on Rasa(in fact does not have Rasa packages dependency) and will work with all versions.

from rasa_core.interpreter import RasaNLUInterpreter
from rasa_core.agent import Agent
interpreter = RasaNLUInterpreter('models/current/nlu')
agent = Agent.load("models/current/dialogue", interpreter=interpreter)

from RasaHost import host
host.nlu_path = os.path.join(current_dir, "data/nlu/")
host.stories_path = os.path.join(current_dir, "data/stories/")
host.domain_path = os.path.join(current_dir, "data/domain.yml")
host.agent = agent
if __name__ == '__main__':    
    host.run()
    # werkzeug  -  * Running on http://0.0.0.0:5005/ (Press CTRL+C to quit)
    
    # GET http://localhost:5005/conversations/daniel/respond?q={{message}}

NLU - md files

from RasaHost import host
host.nlu_path = os.path.join(current_dir, "data/nlu/")
if __name__ == '__main__':    
    host.run()

Core - stories files

from RasaHost import host
host.stories_path = os.path.join(current_dir, "data/stories/")
if __name__ == '__main__':    
    host.run()

Core - domain file

from RasaHost import host
host.domain_path = os.path.join(current_dir, "data/domain.yml")
if __name__ == '__main__':    
    host.run()

Core - chat

Chat interface for testing Rasa. By default, does GET http://{{host}}/conversations/{{sender_id}}/respond?q={{message}}

from RasaHost import host
host.agent = agent
if __name__ == '__main__':    
    host.run()

Core - conversations

Conversations are saved in SQLite.

from RasaHost import host
host.agent = agent
if __name__ == '__main__':    
    host.run()

Core - logs

Logs are saved in SQLite.

from RasaHost import host
host.agent = agent
if __name__ == '__main__':    
    host.run()

Core - analyze

Analyze intents, stories, and domains. Shows warnings and suggestions.

from RasaHost import host
host.nlu_path = os.path.join(current_dir, "data/nlu/")
host.stories_path = os.path.join(current_dir, "data/stories/")
host.domain_path = os.path.join(current_dir, "data/domain.yml")
if __name__ == '__main__':    
    host.run()

Core - memoization policy

Decode memoization policy data.

from RasaHost import host
host.memoization_policy_path = os.path.join(current_dir, "models\current\dialogue\policy_1_MemoizationPolicy")
if __name__ == '__main__':    
    host.run()

Actions

You can also host actions, with or without the agent.

from rasa_core_sdk.executor import ActionExecutor
actionExecutor = ActionExecutor()
actionExecutor.register_package('actions')

from RasaHost import host
host.actionExecutor = actionExecutor
if __name__ == '__main__':    
    host.run()
    # werkzeug  -  * Running on http://0.0.0.0:5005/ (Press CTRL+C to quit)
    # POST http://localhost:5005/actions

Agent with actions

Example of running the app with agent and custom actions.

from rasa_core.interpreter import RasaNLUInterpreter
from rasa_core.agent import Agent
from rasa_core import utils, server
from rasa_core_sdk.executor import ActionExecutor
#  #rasa-host.endpoints.yml
#  action_endpoint:
#  url: "http://localhost:5005/actions"
action_endpoint_conf = utils.read_endpoint_config("rasa-host.endpoints.yml", endpoint_type="action_endpoint")
interpreter = RasaNLUInterpreter('models/current/nlu')
agent = Agent.load("models/current/dialogue", interpreter=interpreter, action_endpoint=action_endpoint_conf)

actionExecutor = ActionExecutor()
actionExecutor.register_package('actions')

from RasaHost import host
host.nlu_path = os.path.join(current_dir, "data/nlu/")
host.stories_path = os.path.join(current_dir, "data/stories/")
host.domain_path = os.path.join(current_dir, "data/domain.yml")
host.agent = agent
host.actionExecutor = actionExecutor
if __name__ == '__main__':    
    host.run()
    # werkzeug  -  * Running on http://0.0.0.0:5005/ (Press CTRL+C to quit)
    
    # GET http://localhost:5005/conversations/daniel/respond?q={{message}}
    # POST http://localhost:5005/actions

Channels

Supports channels. All the conversations will be logged.

from rasa_core.channels.botframework import BotFrameworkInput
input_channel = BotFrameworkInput(
        app_id="",
        app_password=""
)

from RasaHost import host
host.agent = agent
host.channels = [input_channel]
if __name__ == '__main__':    
    host.run()

danielgolabek/RasaHost