node-ari-read

Replicates the asterisk READ Dialplan for nodejs. Is a helper for the official nodejs ARI client package ari-client

Install

$ npm i ari-read

Getting Started

Step 1:

After ari begins a new call use the readerFactory to create a read function for the channel, the factory requires 2 params exactly, the channel and ari variable. You need one read per channel.

Params

• incoming the incoming channel
• ari the ari application

Step 2:

To read one or more files and get a DTMF Response, await your read function. Pass a single object to this function, the object properties are below

The read function can be passed as a parameter to other parts of your application.

Object properties

• soundFiles - array of filenames to read (required)
• digitsInResponse - int number of digits requested, 0 will just read and return on completion, default 0
• attempts - int required, number of invalid attempts to allow before terminating call, default 3
• timeout - int required, number of seconds of inactivity to allow before terminating call, default 10
• responseValidCallback - callback function to test whether the response is valid or not, default is that all responses are valid
• debug - if set to true it will read out some debug commands, default false

Example

const client = require('ari-client');
const readerFactory = require('ari-read');

client.connect('http://localhost:8088', 'user', 'secret')
  .then((ari)=> {
    //handle new call
    ari.once('StasisStart',  async (event, incoming)=> {
        incoming.answer().then(async () => {
            //create reader
            const read = readerFactory(incoming, ari);

            //read without response
            await read(
                audioFiles: ['welcome'],
                digitsInResponse: 0,
            )
    
            //read with response, must be * or a digit between 1 and 3
            const selection = await read({
                audioFiles: [file1, file2], 
                digitsInResponse: 1, 
                responseValidCallback:  (num) => {
                    return num=='*' || (num > 0 && num <= 3)
                }
            }).catch(
                (err) => { console.log(err) }
            );

            //now you can do an action with your selection
            doSomething(selection); 
        })
    });
    //connect to the statis app
    ari.start('test');
  })
  .done(); // program will crash if it fails to connect

Example using TTS (Text-to-Speech)

This example replaces the readerFactory in the above example with one that will perform TTS before returning the file. Instead of sending soundFiles to this helper, send an array of text to read.

const readerFactory = require("ari-read");

const tts await (string) => {
    //use your favorite library and return the filename you generate.
}

const readerFactoryWithTTS = (channel, ari) => {
    const read = readerFactory(channel, ari);

    //the reader returns an async function for us to await
    return async (props) => {
        const { text} = props;

        let soundFiles = await Promise.all((typeof text === 'string' ? [text] : text).map(s => tts(s)));
        return await read({soundFiles, ...props})
    }
}
module.exports = readerFactoryWithTTS;

Concurrency Notes

This code has been optimized to run at high call load (Author has tested with 300 on concurrent on 4 CPUs)

• Asterisk ships with low version of nodejs, nodejs 10 has improved performance with async/promises and should be faster when at high load.
• Transcoding can use significant CPU, ensure that the source files use codecs that use the least CPU for best performance.
• If you setup clustering on the ARI App, you can give each process a different statis app to connect to. The dialplan can load balance between the statis apps.