matheusmatos/botpress-rasa_nlu

The Rasa NLU module for Botpress

botpress-rasa_nlu

A module to use Rasa.ai with your Botpress bot.

Getting started

botpress install rasa_nlu

The Rasa NLU module should now be available in your dashboard.

You'll need a Rasa NLU server (local or in-cloud) to use this module, that will inject understanding metadata inside incoming messages through the Rasa API.

Events will have an rasa_nlu property populated with the extracted metadata from Rasa server.

Take a look:

bp.hear({'rasa_nlu.intent.name': 'greet'}, (event) => {
	bp.messenger.sendText(event.user.id, 'Hi Human! Nice to meet you!')
})

Enjoy Botpress!

Community

Pull requests are welcomed! We believe that it takes all of us to create something big and impactful.

There's a Slack community where you are welcome to join us, ask any question and even help others.

Get an invite and join us now! 👉https://slack.botpress.io

License

botpress-rasa_nlu is licensed under AGPL-3.0

botpress-rasa_nlu's Issues

Using with Rasa Core

I'm assuming to use this with Rasa core I just need to create a version of this that adjusts the server URL unless you want me to add something maybe to this project so you can use NLU or Core or both:

http://localhost:5005/conversations/default/parse?q="hello"

This is the normal format for Rasa Core.

Encoding problem

rasa_nlu expects the request params to be utf-8
When I send a message with german umlauts (e.g. ä ü ö), the message is encoded with cp1252 instead of utf-8 and rasa_nlu fails to parse it.

2017-12-07 10:37:05+0100 [_GenericHTTPChannelProtocol,1,127.0.0.1] Unhandled Error
	Traceback (most recent call last):
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/klein/app.py", line 50, in _call
	    result = f(*args, **kwargs)
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/rasa_nlu/server.py", line 92, in decorated
	    return f(*args, **kwargs)
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/rasa_nlu/server.py", line 77, in decorated
	    return f(*args, **kwargs)
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/twisted/internet/defer.py", line 1532, in unwindGenerator
	    return _inlineCallbacks(None, gen, Deferred())
	--- <exception caught here> ---
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/twisted/internet/defer.py", line 1386, in _inlineCallbacks
	    result = g.send(result)
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/rasa_nlu/server.py", line 133, in parse_get
	    for key, value in request.args.items()}
	  File "/Users/morth/src/enaptfix-bot/venv/lib/python3.6/site-packages/rasa_nlu/server.py", line 133, in <dictcomp>
	    for key, value in request.args.items()}
	builtins.UnicodeDecodeError: 'utf-8' codec can't decode byte 0xfc in position 5: invalid start byte

2017-12-07 10:37:05+0100 [-] "127.0.0.1" - - [07/Dec/2017:09:37:04 +0000] "GET /parse?q=Das%20T\xfcrschloss%20an%20meiner%20Eingangst\xfcr%20ist%20defekt&project=default HTTP/1.1" 500 4214 "-" "axios/0.15.3"

Confidence Scoring

By chance is there any way to only trigger if the confidence score is above a certain %? Wasn't sure if you had that setup yet or not.

Adding the module parameter

see the similar issue for the plugin for botkit

howdyai/botkit-rasa#3

Not loading module

The module only works after changing the order of the middleware and then changing back to it's original position.

I see on initial load of botpress:

info: Loaded botpress-rasa_nlu, version 1.0.0

in the console.

When I change the positions on in the middleware area of botpress I see in the console:

debug: Loading middleware: rasa_nlu.incoming

which I don't see on the initial load of the modules.

Please advise! Love to help

botpress 10.20.1 and rasa_nlu 0.12.3 how to integrate them?

Hi..

I am running botpress 10.20.1 and rasa_nlu 0.12.3 on centOS 7

my botpress nlu config as follow:

vi config/nlu.json

{
"intentsDir": "./generated/intents",
"entitiesDir": "./generated/entities",
"provider": "rasa",
"debugModeEnabled": false,
"minimumConfidence": "0.3"
}
~

and my rasa_nlu already configured and run using tensorflow
how to integrate it?

11:52:44 - info: Starting botpress version 10.20.1
11:52:44 - info: [Ghost Content Manager] (transparent) Initialized
11:52:44 - debug: [Ghost Content Manager] (transparent) Added root folder generated/media, doing nothing.
11:52:44 - debug: [Ghost Content Manager] (transparent) Added root folder generated/flows, doing nothing.
11:52:44 - info: [Skills] Initiated
11:52:44 - info: Loaded @botpress/audience, version 10.20.1
11:52:44 - verbose: [Renderers] Enabled for webchat.
11:52:44 - info: Loaded @botpress/channel-web, version 10.20.1
11:52:44 - info: Loaded @botpress/hitl, version 10.20.1
11:52:44 - debug: [Ghost Content Manager] (transparent) Added root folder generated/intents, doing nothing.
11:52:44 - debug: [Ghost Content Manager] (transparent) Added root folder generated/entities, doing nothing.
11:52:44 - info: Loaded @botpress/nlu, version 10.20.1
11:52:45 - info: Loaded @botpress/skill-choice, version 10.20.1
11:52:45 - info: Loaded @botpress/terminal, version 10.20.1
11:52:45 - info: Loaded 6 modules
11:52:45 - info: [Skills] Loaded 1 skills
11:52:45 - debug: [Ghost Content Manager] (transparent) Added root folder generated/content, doing nothing.
11:52:45 - debug: Loading middleware: rendering.instrumentation
11:52:45 - debug: Loading middleware: hitl.captureInMessages
11:52:45 - debug: Loading middleware: nlu.incoming
11:52:45 - debug: Loading middleware: hear
11:52:45 - debug: Loading middleware: hitl.captureOutMessages
11:52:45 - debug: Loading middleware: webchat.sendMessages
11:52:45 - debug: Loading middleware: fallback
11:52:45 - debug: Loading data for builtin_text from builtin_text.json
11:52:45 - info: Read 8 item(s) from builtin_text.json
11:52:45 - info: Bot launched. Visit: http://localhost:3000
11:52:45 - debug: Loading data for builtin_image from builtin_image.json
11:52:45 - debug: Loading data for builtin_single-choice from builtin_single-choice.json
11:52:45 - debug: Loading data for builtin_card from builtin_card.json
11:52:45 - debug: Loading data for builtin_action-button from builtin_action-button.json
11:52:45 - debug: Loading data for builtin_carousel from builtin_carousel.json
11:52:45 - debug: Loading data for builtin_raw from builtin_raw.json
11:52:45 - info: ------------
11:52:45 - info: Webchat available at http://localhost:3000/s/chat
11:52:45 - info: ------------
11:52:48 - debug: [NLU::Rasa] Could not fetch model versions
11:52:56 - debug: Dialog: [ N/A ] ( N/A ) ** LOAD
11:52:56 - debug: Dialog: [main] (entry) <> NTHG
11:52:56 - debug: Dialog: [ N/A ] ( N/A ) -- ENDF
11:52:58 - debug: [NLU::Rasa] Could not fetch model versions
11:53:08 - debug: [NLU::Rasa] Could not fetch model versions
11:53:18 - debug: [NLU::Rasa] Could not fetch model versions
11:53:28 - debug: [NLU::Rasa] Could not fetch model versions
11:53:38 - debug: [NLU::Rasa] Could not fetch model versions
11:53:48 - debug: [NLU::Rasa] Could not fetch model versions
11:53:58 - debug: [NLU::Rasa] Could not fetch model versions
11:54:08 - debug: [NLU::Rasa] Could not fetch model versions
11:54:18 - debug: [NLU::Rasa] Could not fetch model versions
11:54:28 - debug: [NLU::Rasa] Could not fetch model versions
11:54:38 - debug: [NLU::Rasa] Could not fetch model versions
11:54:48 - debug: [NLU::Rasa] Could not fetch model versions
11:54:58 - debug: [NLU::Rasa] Could not fetch model versions
11:55:08 - debug: [NLU::Rasa] Could not fetch model versions
11:55:18 - debug: [NLU::Rasa] Could not fetch model versions
11:55:28 - debug: [NLU::Rasa] Could not fetch model versions
11:55:38 - debug: [NLU::Rasa] Could not fetch model versions
11:55:48 - debug: [NLU::Rasa] Could not fetch model versions
11:55:58 - debug: [NLU::Rasa] Could not fetch model versions
11:56:08 - debug: [NLU::Rasa] Could not fetch model versions
11:56:18 - debug: [NLU::Rasa] Could not fetch model versions
11:56:28 - debug: [NLU::Rasa] Could not fetch model versions