rasa-webchat-josh v0.0.9

webchat

A simple webchat widget to connect with a chatbot. Forked from react-chat-widget Further forked from rasa-webchat

Features

• Text Messages
• Quick Replies
• Images and Videos
• Snippet style for links (only as responses for now)
• Markdown support
• Easy to import in a script tag or as a React Component
• Persistent sessions
• Event Observer

Setup

In a ` tag

In your <body/>:

<div id="webchat"/>
<script src="https://storage.googleapis.com/mrbot-cdn/webchat-0.5.0.js"></script>
<script>
  WebChat.default.init({
    selector: "#webchat",
    initPayload: "/get_started",
    interval: 1000, // 1000 ms between each message
    customData: {"userId": "123"}, // arbitrary custom data. Stay minimal as this will be added to the socket
    socketUrl: "http://localhost:5500",
    socketPath: "/socket.io/",
    title: "Title",
    subtitle: "Subtitle",
    inputTextFieldHint: "Type a message...",
    connectingText: "Waiting for server...",
    fullScreenMode: false,
    profileAvatar: "http://to.avat.ar",
    openLauncherImage: 'myCustomOpenImage.png',
    closeLauncherImage: 'myCustomCloseImage.png',
    params: {
      images: {
        dims: {
          width: 300,
          height: 200,
        }
      },
      storage: "local"
    }
  })
</script>

About images: width and height define the size in pixels that images in messages are crop-scaled to. If not present, the image will scale to the maximum width of the container and the image.

It is recommended to use a particular version (i.e. "webchat-.js") however the file "webchat-latest.js" is also available and is updated continuously with the latest version.

As a React component

Install the package from GitHub by running:

npm install --save rasa-webchat-josh
npm install --save react-event-observer

Then once it is installed it can be implemented as follows.

import React, { Component } from 'react';
import ReactObserver from 'react-event-observer';
import {
  Widget,
  BOT_MESSAGE_RECEIVED,
  CHAT_OPENED,
  CHAT_CLOSED
} from 'rasa-webchat-josh';

class CustomWidget extends Component {

  constructor(props){
    super(props);
    this.observer = ReactObserver();
    this.initializeListeners();
  }

  //Event listener
  initializeListeners(){
    this.msgAddListener = this.observer.subscribe(BOT_MESSAGE_RECEIVED,(data) =>{
      console.log("Message added **; size "+data);
    });
    this.chatOpened = this.observer.subscribe(CHAT_OPENED,(data) =>{
      console.log("Chat opened; isChatOpen "+data);
    });
    this.chatClosed = this.observer.subscribe(CHAT_CLOSED,(data) =>{
      console.log("Chat closed; isChatOpen "+data);
    });    
  }

  render(){
    return(
      <div>
      <Widget
        interval={2000}
        initPayload={"/get_started"}
        socketUrl={"http://localhost:5500"}
        socketPath={"/socket.io/"}
        title={"Title"}
        inputTextFieldHint={"Type a message..."}
        connectingText={"Waiting for server..."}
        embedded={true}
        openLauncherImage="myCustomOpenImage.png"
        closeLauncherImage="myCustomCloseImage.png"
        observer={this.observer}
        params={{
          images: {
            dims: {
              width: 300,
              height: 200
            }
          },
          storage: "local"
        }}
      />
      </div>
    )
  }
}

• Make sure to have the prop embedded set to true if you don't want to see the launcher.

Backend

Rasa Core

Use the SocketIOInput channel: See instructions in the Rasa Core documentation

Others

Your backend must expose a socket with socket.io

Receiving messages from the chat

@socketio.on('user_uttered')
    def handle_message(message):
        # do something

Sending messages from the backend to the chat widget

sending plain text

emit('bot_uttered', {"text": "hello"}, room=session_id)

sending quick replies

message = {
  "text": "Happy?",
  "quick_replies":[
    {"title":"Yes", "payload":"/affirm"},
    {"title":"No", "payload":"/deny"}
  ]}
emit('bot_uttered', message, room=socket_id)

sending a link Snippet

Admittedly a bit far fetched, thinking that Snippets would evolve to carousels of generic templates :)

message = {
  "attachment":{
    "type":"template",
    "payload":{
      "template_type":"generic",
      "elements":[
        {
          "title":"Title",
          "buttons":[ {
            "title":"Link name",
            "url": "http://link.url"
          }
        ]
      }
    ]
  }
}
}
emit('bot_uttered', message, room=socket_id)

sending a Video Message

message = {
  "attachment":{
    "type":"video",
    "payload":{
      "title":"Link name",
      "src": "https://www.youtube.com/watch?v=f3EbDbm8XqY"
    }
  }
}
emit('bot_uttered', message, room=socket_id)

sending an Image Message

message = {
      "attachment":{
        "type":"image",
        "payload":{
          "title":"Link name",
          "src": "https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcT_IX5FSDQLrwm9qvuXu_g7R9t_-3yBSycZ8OxpRXvMDaTAeBEW"
        }
      }
    }
emit('bot_uttered', message, room=socket_id)

Usage

Session Persistence

storage specifies the location where the the conversation and state of the WebChat is stored in the browser's storage.

storage: "session" defines the state to be stored in the session storage. The session storage persists on reload of the page, and is cleared after the browser or tab is closed, or when sessionStorage.clear()is called.

storage: "local" defines the state to be stored in the local stoage. The local storage persists after the the browser is closed, and is cleared when the cookies of the browser are cleared, or when localStorage.clear()is called.

API

Styles

hierarchy:

.conversation-container
  |-- .header
        |-- .title
        |-- .close-function
        |-- .loading
  |-- .messages-container
        |-- .message
              |-- .client
              |-- .response
        |-- .replies
              |-- .reply
              |-- .response
        |-- .snippet
              |-- .snippet-title
              |-- .snippet-details
              |-- .link
        |-- .imageFrame
        |-- .videoFrame
  |-- .sender
        |-- .new-message
        |-- .send

Usage with Docker

Since you have to install the package from GitHub, npm will clone the repo to the global .npm directory before building the module in your node_modules directory. For this reason docker will have trouble installing the package, of course the global .npm directory doesn't exist in the container. To solve this simply add the following line in your Dockerfile before the RUN npm install command

RUN mkdir -p /root/.npm

Contributors

@PHLF @znat @TheoTomalty @Hub4IT @dliuproduction