Convey is a JavaScript Framework for connecting to processes created with AIOS.
The framework ist compatible with all major Browsers and can be used standalone as well as in conjunction with React, Angular or Vue.
In order to add Convey to your project, follow these steps:
- MQTT endpoint: e.g. wss://gaia.local/mqtt
- Identifier of the so called Identity to connect to.
Using npm:
$ npm i @leftshiftone/convey
Using in browser:
<script src="convey-std.js"/>
Create the following DOM structure in your project
<div class="lto-gaia">
<div class="lto-content"/>
<div class="lto-suggest"/>
<input class="lto-textbox"/>
<button class="lto-invoker"/>
</div>
Import Convey
<head>
<meta charset="UTF-8"/>
<link rel="stylesheet" href="convey-std.css"/>
<script src="convey-std.js"/>
</head>
Connect to AIOS
<script type="text/javascript">
const header = new GaiaConvey.QueueHeader("anyIdentityId", "optionalChannelId")
new GaiaConvey.Gaia(new GaiaConvey.DefaultListener())
.connect(new GaiaConvey.QueueOptions('wss://URL/mqtt', 443, "MQTT_USERNAME", "MQTT_PASSWORD"))
.then(connection => {
const interactionSubscription = connection.subscribeInteraction(header, payload => console.debug(`Interaction:`, payload), new GaiaConvey.ContentCentricRenderer());
connection.subscribeNotification(header, payload => console.debug('Notification:', payload));
connection.subscribeLogging(header, payload => console.debug('Logging:', payload))
connection.subscribeContext(header, payload => console.debug('context:', payload))
interactionSubscription.reception({initialMessage: "value"});
});
</script>
Import Convey
import {Gaia, ContentCentricRenderer, OffSwitchListener, ConversationQueueType, QueueOptions} from "@leftshiftone/convey";
Connect to AIOS
const header = new QueueHeader("anyIdentityId", "optionalChannelId")
new Gaia(new OffSwitchListener())
.connect(new QueueOptions('wss://URL/', 8080, "USERNAME", "PASSWORD"))
.then(connection => {
const interactionSubscription = connection.subscribeInteraction(header, payload => console.debug(`Interaction:`, payload), new ContentCentricRenderer());
connection.subscribeNotification(header, payload => console.debug('Notification:', payload));
connection.subscribeLogging(header, payload => console.debug('Logging:', payload))
connection.subscribeContext(header, payload => console.debug('context:', payload))
interactionSubscription.reception({initialMessage: "value"});
});
To use the default styling you can import it like the following:
import '@leftshiftone/convey/dist/convey-all.css';
The communication with AIOS contains several queueTypes where each one has its own purpose.
The Interaction queue is the main queue and is responsible for exchanging the elements configured in
AIOS. Convey automatically subscribes to this queue by calling interactionSubscription.reception()
.
The messages in this queue are rendered to HTML elements.
The Context consists of attributes defined in the AIOS BPMN process. It can be received by subscribing to this queue.
Each notification configured in the AIOS BPMN process can be received if subscribed to this queue.
AIOS sends logs for certain process executions which can be received by subscribing to this queue.
A Renderer defines how elements, which were received by the 'Interaction' queue, are rendered in
the HTML DOM tree. Furthermore, a renderer specifies the layout of an integration project.
You can define a renderer by passing it to connection.subscribeInteraction
.
const header = new QueueHeader("anyIdentityId", "optionalChannelId")
new Gaia().connect( ... )
.then(connection => {
const interactionSubscription = connection.subscribeInteraction(header, payload => console.debug(`Interaction:`, payload), new ContentCentricRenderer());
});
The classic renderer renders the messages in a classic top-down manner.
The content centric renderer tries to maximize the time a content is visible by updating the content if possible or displaying interrupting actions like intent cascading by overlaying the content.
Renderer implementation which is based on the reveal.js library. This renderer supports horizontal as well as vertical navigation.
No-operation dummy renderer. Mainly used for audio only use cases.
The Rating Decorator cannot be used as standalone renderer, but instead wraps any of the previous renderers. It adds a rating element with thumbs up / thumbs down and comment field to every incoming message. On click the rating is sent and stored in the backend.
const header = new QueueHeader("anyIdentityId", "optionalChannelId")
new Gaia().connect( ... )
.then(connection => {
const interactionSubscription = connection.subscribeInteraction(header, payload => console.debug(`Interaction:`, payload), new RatingDecorator(new ContentCentricRenderer()));
});
The rating decorator takes the RatingRenderStrategy as an optional argument. Possible options are
- ALL_EXCEPT_DISABLED_RATINGS (default): Render ratings for all interactions except the ones with rating markup where enabled is set to false.
- ONLY_ENABLED_RATINGS: Only render ratings for interactions for rating markup with enabled set to true.
Example usage:
new RatingDecorator(new ContentCentricRenderer(), RatingRenderStrategy.ONLY_ENABLED_RATINGS)
A listener provides the functionality to react to certain events. Events can be:
- Connected
- ConnectionLost
- PacketSend
- Disconnected
- Error
- Message
Acts as the base listener.
If an input text area should only be visible when an input is required, this is the listener to be used.
A behaviour adds event listeners to UI elements for a specific subscription.
The autocomplete behaviour adds autocomplete suggestions for a given textarea (or the default textarea with class "lto-textbox") to a given div (or the default div with class "lto-autocomplete"). Provide the autocomplete elements through the add-method.
Required html structure:
<div style="position: absolute">
<textarea title="" class="lto-textbox"></textarea>
<div class="lto-autocomplete"></div>
<button class="lto-invoker"></button>
</div>
const autocompleteBehaviour = new GaiaConvey.AutocompleteBehaviour({
maxNumberOfResults : 3
})
["hello", "world", "hello world", "hello1", "hello 123"].forEach(e => autocompleteBehaviour.add(e));
let gaia = new GaiaConvey.Gaia(renderer, listener);
gaia.connect(...)
.then(connection => {
const subscription = connection.subscribe(...);
autocompleteBehaviour.bind(subscription);
});
A language code must be present in all convey interactions in order for AIOS to be able to reply in the proper language.
- ISO 639-1 (e.g. de => german, en => english)
- Locale (e.g. de_DE, de_AT, en_US)
The languages can be provided in 3 different ways:
- By setting the environment variable CONVERSATION_LANGUAGE
- By defining a HTML element with the class "lto-language" containing the language code as its value
- By the browser. The default language browser or rather its locale will be sent to AIOS.
Please note that the language providers will be executed in the described order.
In other words, if the variable CONVERSATION_LANGUAGE is set, the language code defined in the HTML element or the browser will be ignored.
NOTE: Programmatically it is also possible to hardcode the language. See example below (LanguageInteractionInterceptor):
const header = new QueueHeader("anyIdentityId", "optionalChannelId")
new Gaia(new OffSwitchListener())
.connect(new QueueOptions('wss://URL/', 8080, "USERNAME", "PASSWORD"))
.then(connection => {
const interactionSubscription = connection.subscribeInteraction(header, payload => console.debug(`Interaction:`, payload), new ContentCentricRenderer(), new LanguageInteractionInterceptor("en"));
connection.subscribeNotification(header, payload => console.debug('Notification:', payload));
connection.subscribeLogging(header, payload => console.debug('Logging:', payload))
connection.subscribeContext(header, payload => console.debug('context:', payload))
interactionSubscription.reception({initialMessage: "value"});
});
The following modules are available:
- std: Contains default modules
- aud: Contains the audio module
- cod: Contains the code reader module (e.g. QRCode)
- map: Contains modules for Open Street Map, Google Maps and Here Maps
- vis: Contains modules for rendering data as charts
- all: Contains all modules
Convey provides very basic WAI-ARIA features. This includes corresponding roles
.
The main chat area is defined as a aria-role: log
with aria-live: polite
enabled. Chatbot outputs and inputs can be navigated
using the keyboard (<TAB> && <S-TAB>
).
Accessbility features are currently available when using the ContentCentricRenderer
(default renderer implementation) and
can NOT be disabled as we see accessibility as an important and necessary concept.
Releases are triggered locally. Just a tag will be pushed to trigger the CI release pipeline.
Run yarn trigger-release:major
locally.
Run yarn trigger-candidate:major
locally.
Run yarn trigger-release:minor
locally.
Run yarn trigger-candidate:minor
locally.
Switch to the minor version branch, before performing the release (e.g. release/1.3.x
) then run yarn trigger-release:patch
locally.
Visit conveyjs-starter to create a new project including conveyjs
For a short guide check the conveyjs-starter readme
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
In general, we follow the "fork-and-pull" Git workflow:
- Fork the repo on GitHub
- Clone the project to your own machine
- Commit changes to your own branch
- Push your work back up to your fork
- Submit a Pull request so that we can review your changes
Please refer to the project's guidelines for submitting patches and additions and make sure to update tests as appropriate.
NOTE: Be sure to merge the latest from "upstream" before making a pull request!
MIT Licence