Simple chatbot UI for the Web with JSON scripting 👋🤖🤙
- Quick set-up & implementation.
- Works with or without Natural Language Classifiers.
- 1KB GZipped. No dependencies. Written with ES5 (compatible with IE11+).
yarn add chat-bubble or npm install chat-bubble
Get the .ZIP file here.
This method assumes you've got a development environment running that's capable of transpiling ES6 JavaScript. There's a short guide on how to get one working here. Otherwise see "I have no ES6 dev environment." This guide will show you how to build this.
/************************************************************************/
/******* CONVENIENCE METHODS AVAILABLE FOR ES6 BUILD ENVIRONMENTS *******/
/************************************************************************/
// the URL of where you've installed the component; you may need to change this:
import {
Bubbles,
prepHTML
} from "../node_modules/chat-bubble/component/Bubbles.js";
// this is a convenience script that builds all necessary HTML,
// imports all scripts and stylesheets; your container DIV will
// have a default `id="chat"`;
// you can specify a different ID with:
// `container: "my_chatbox_id"` option
prepHTML({ relative_path: "../node_modules/chat-bubble/" });
/************************************************************************/
/************************ SAMPLE IMPLEMENTATION *************************/
/************************************************************************/
// initialize by constructing a named function...
const chatWindow = new Bubbles(
document.getElementById("chat"), // ...passing HTML container element...
"chatWindow" // ...and name of the function as a parameter
);
// `.talk()` will get your bot to begin the conversation
chatWindow.talk(
// pass your JSON/JavaScript object to `.talk()` function where
// you define how the conversation between the bot and user will go
{
// "ice" (as in "breaking the ice") is a required conversation object
// that maps the first thing the bot will say to the user
ice: {
// "says" defines an array of sequential bubbles
// that the bot will produce
says: ["Hey!", "Can I have a banana?"],
// "reply" is an array of possible options the user can pick from
// as a reply
reply: [
{
question: "🍌", // label for the reply option
answer: "banana" // key for the next conversation object
}
]
}, // end required "ice" conversation object
// another conversation object that can be queued from within
// any other conversation object, including itself
banana: {
says: ["Thank you!", "Can I have another banana?"],
reply: [
{
question: "🍌🍌",
answer: "banana"
}
]
} // end conversation object
} // end conversation object
);If you don't want to bother with setting up a development server and transpiler for ES6 code, I get it. Simply unzip the package and create index.html inside of that directory. Then add all the JavaScript that you see below the /*SAMPLE IMPLEMENTATION*/ comment in the code example above. Replace const with var.
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<title>My chat-bubble Project</title>
<!-- stylesheets are conveniently separated into components -->
<link rel="stylesheet" media="all" href="../styles/setup.css" />
<link rel="stylesheet" media="all" href="../styles/says.css" />
<link rel="stylesheet" media="all" href="../styles/reply.css" />
<link rel="stylesheet" media="all" href="../styles/typing.css" />
<link rel="stylesheet" media="all" href="../styles/input.css" />
</head>
<body>
<!-- container element for chat window -->
<div id="chat"></div>
<!-- import the JavaScript file -->
<script src="./component/Bubbles.js"></script>
<script>
/************************************************************************/
/**************** add "SAMPLE IMPLEMENTATION" code here *****************/
/************************************************************************/
</script>
</body>
</html>Now open this file in your browser. Done!
- Basic example: see how the code above looks in browser.
- Custom starting point: what if you wanted to resume conversation from somewhere other than required
ice:{}starting point? This is how you'd do it. - Keyboard input: a basic plugin-like structure that lets you implement your own keyboard input and text recognition (though it does not process natural language).
- Run scripts: your bot's replies can do things. Not only it could say something, but it could point your user towards an action, or perform it by running JavaScript.
- Natural Language Classifier implementation is possible with additional effort by intercepting the response message and keyboard input. Example using RASA (docs) can be found here.
Check out /examples folder for the source code and more ideas.
- Can I add images and HTML code to my bot?
- Yes! custom graphics, YouTube videos - whatever you want!
- How can I contribute?
- See the contribution guide here.
- Buy me a coffee: use the GH "Sponsor" button or do it via https://ko-fi.com/dmitrizzle
- You may need to add in polyfills for
Object.assign()andString.includes()
