def create_agent(model: Text, endpoints: Text = None) -> 'Agent':
from rasa_core.interpreter import RasaNLUInterpreter
from rasa_core.tracker_store import TrackerStore
from rasa_core import broker
from rasa_core.utils import AvailableEndpoints
core_path, nlu_path = get_model_subdirectories(model)
_endpoints = AvailableEndpoints.read_endpoints(endpoints)
_interpreter = None
if os.path.exists(nlu_path):
_interpreter = RasaNLUInterpreter(model_directory=nlu_path)
else:
_interpreter = None
logging.info("No NLU model found. Running without NLU.")
_broker = broker.from_endpoint_config(_endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(None,
_endpoints.tracker_store,
_broker)
return Agent.load(core_path,
generator=_endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=_endpoints.action)
async def load_agent_on_start(core_model, endpoints, nlu_model, app, loop):
"""Load an agent.
Used to be scheduled on server start
(hence the `app` and `loop` arguments)."""
from rasa_core import broker
from rasa_core.agent import Agent
_interpreter = NaturalLanguageInterpreter.create(nlu_model, endpoints.nlu)
_broker = broker.from_endpoint_config(endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(None,
endpoints.tracker_store,
_broker)
if endpoints and endpoints.model:
from rasa_core import agent
app.agent = Agent(interpreter=_interpreter,
generator=endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=endpoints.action)
await agent.load_from_server(app.agent, model_server=endpoints.model)
else:
app.agent = Agent.load(core_model,
interpreter=_interpreter,
generator=endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=endpoints.action)
return app.agent
def test_tracker_store_from_string(default_domain):
endpoints_path = "data/test_endpoints/custom_tracker_endpoints.yml"
store_config = utils.read_endpoint_config(endpoints_path, "tracker_store")
tracker_store = TrackerStore.find_tracker_store(default_domain,
store_config)
assert isinstance(tracker_store, ExampleTrackerStore)
def test_tracker_store_from_invalid_string(default_domain):
endpoints_path = "data/test_endpoints/custom_tracker_endpoints.yml"
store_config = utils.read_endpoint_config(endpoints_path, "tracker_store")
store_config.type = "any string"
tracker_store = TrackerStore.find_tracker_store(default_domain,
store_config)
assert isinstance(tracker_store, InMemoryTrackerStore)
def test_tracker_store_from_invalid_module(default_domain):
endpoints_path = "data/test_endpoints/custom_tracker_endpoints.yml"
store_config = utils.read_endpoint_config(endpoints_path, "tracker_store")
store_config.type = "a.module.which.cannot.be.found"
tracker_store = TrackerStore.find_tracker_store(default_domain,
store_config)
assert isinstance(tracker_store, InMemoryTrackerStore)
def test_find_tracker_store(default_domain):
store = utils.read_endpoint_config(DEFAULT_ENDPOINTS_FILE, "tracker_store")
tracker_store = RedisTrackerStore(domain=default_domain,
host="localhost",
port=6379,
db=0,
password="******",
record_exp=3000)
assert isinstance(
tracker_store,
type(TrackerStore.find_tracker_store(default_domain, store)))
def do_interactive_learning(cmdline_args, stories, additional_arguments):
_endpoints = AvailableEndpoints.read_endpoints(cmdline_args.endpoints)
_interpreter = NaturalLanguageInterpreter.create(cmdline_args.nlu,
_endpoints.nlu)
if cmdline_args.core:
if cmdline_args.finetune:
raise ValueError("--core can only be used without "
"--finetune flag.")
logger.info("Loading a pre-trained model. This means that "
"all training-related parameters will be ignored.")
_broker = PikaProducer.from_endpoint_config(_endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(
None, _endpoints.tracker_store, _broker)
_agent = Agent.load(cmdline_args.core,
interpreter=_interpreter,
generator=_endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=_endpoints.action)
else:
if cmdline_args.out:
model_directory = cmdline_args.out
else:
model_directory = tempfile.mkdtemp(suffix="_core_model")
_agent = train_dialogue_model(cmdline_args.domain, stories,
model_directory, _interpreter,
_endpoints, cmdline_args.dump_stories,
cmdline_args.config[0], None,
additional_arguments)
interactive.run_interactive_learning(
_agent,
stories,
finetune=cmdline_args.finetune,
skip_visualization=cmdline_args.skip_visualization)
def do_interactive_learning(cmdline_args, stories, additional_arguments):
_endpoints = AvailableEndpoints.read_endpoints(cmdline_args.endpoints)
_interpreter = NaturalLanguageInterpreter.create(cmdline_args.nlu,
_endpoints.nlu)
if (isinstance(cmdline_args.config, list)
and len(cmdline_args.config) > 1):
raise ValueError("You can only pass one config file at a time")
if cmdline_args.core and cmdline_args.finetune:
raise ValueError("--core can only be used without --finetune flag.")
elif cmdline_args.core:
logger.info("loading a pre-trained model. "
"all training-related parameters will be ignored")
_broker = PikaProducer.from_endpoint_config(_endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(
None, _endpoints.tracker_store, _broker)
_agent = Agent.load(cmdline_args.core,
interpreter=_interpreter,
generator=_endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=_endpoints.action)
else:
if not cmdline_args.out:
raise ValueError("you must provide a path where the model "
"will be saved using -o / --out")
_agent = train_dialogue_model(cmdline_args.domain, stories,
cmdline_args.out, _interpreter,
_endpoints, cmdline_args.dump_stories,
cmdline_args.config[0], None,
additional_arguments)
interactive.run_interactive_learning(
_agent,
stories,
finetune=cmdline_args.finetune,
skip_visualization=cmdline_args.skip_visualization)
def start_core(platform_token):
from rasa_core.utils import AvailableEndpoints
_endpoints = AvailableEndpoints(
# TODO: make endpoints more configurable, esp ports
model=EndpointConfig(
"http://localhost:5002"
"/api/projects/default/models/tags/production",
token=platform_token,
wait_time_between_pulls=1),
event_broker=EndpointConfig(**{"type": "file"}),
nlg=EndpointConfig("http://localhost:5002"
"/api/nlg",
token=platform_token))
from rasa_core import broker
_broker = broker.from_endpoint_config(_endpoints.event_broker)
from rasa_core.tracker_store import TrackerStore
_tracker_store = TrackerStore.find_tracker_store(None,
_endpoints.tracker_store,
_broker)
from rasa_core.run import load_agent
_agent = load_agent("models",
interpreter=None,
tracker_store=_tracker_store,
endpoints=_endpoints)
from rasa_core.run import serve_application
print_success("About to start core")
serve_application(
_agent,
"rasa",
5005,
"credentials.yml",
"*",
None, # TODO: configure auth token
True)
logging.getLogger('werkzeug').setLevel(logging.WARN)
logging.getLogger('matplotlib').setLevel(logging.WARN)
utils.configure_colored_logging(cmdline_args.loglevel)
utils.configure_file_logging(cmdline_args.loglevel,
cmdline_args.log_file)
logger.info("Rasa process starting")
_endpoints = AvailableEndpoints.read_endpoints(cmdline_args.endpoints)
_interpreter = NaturalLanguageInterpreter.create(cmdline_args.nlu,
_endpoints.nlu)
_broker = PikaProducer.from_endpoint_config(_endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(
None, _endpoints.tracker_store, _broker)
_agent = load_agent(cmdline_args.core,
interpreter=_interpreter,
tracker_store=_tracker_store,
endpoints=_endpoints)
serve_application(_agent,
cmdline_args.connector,
cmdline_args.port,
cmdline_args.credentials,
cmdline_args.cors,
cmdline_args.auth_token,
cmdline_args.enable_api,
cmdline_args.jwt_secret,
cmdline_args.jwt_method)
