Parse SRT from sonix transcription

README.md

How does it work?

I was too lazy to create an interface so all interaction with the script takes place through the console.

1. Inject script. (for example using Tampermonkey)
2. Open the console in browser (F12)
3. Write sonixParser.getSRT();
4. Copy text from the console and save as .srt in text editor

Options

sonixParser.getSRT({
	lines: 3, // the number of strings in the subtitles. DEFAULT: 1
	lineLength: 50, // max length of strings in the subtitles. DEFAULT: 36
	named: true, // show speaker name. DEFAULT: false
	nameOnAnotherLine: true // DEFAULT: false
})

https://sonix.ai/

SonixParser.js

// ==UserScript==
// @name         Sonix parser
// @namespace    
// @version      0.2
// @description  Parse SRT from sonix transcription
// @author       Cabbage Storm
// @match        https://sonix.ai/resources/*
// @grant        none
// @copyright    2019, Newt300 (https://openuserjs.org//users/Newt300)
// @license      MIT
// ==/UserScript==

let sonixParser = {};

sonixParser.parse = function() {
   	let arr = [];
	document.querySelectorAll(".sonix--transcript-exchange").forEach(function(e) {
		let curr = {
			time: e.querySelector('.sonix--transcript-exchange-timestamp').getAttribute('data-ts'),
			speakerName: e.querySelector('.sonix--transcript-exchange-speaker-name').innerText,
			text: e.querySelector('.sonix--transcript-exchange-text').innerText,
			speach: []
		}
		e.querySelectorAll('.sonix--transcript-exchange-text span').forEach(function(el) {
			curr.speach.push({
				time: el.getAttribute('data-ts'),
				text: el.innerText
			});
		});
		arr.push(curr);
	})
	sonixParser.parsedJSON = arr;
	console.log(arr);
}

function prepare(options) {
	options = options ? options : {};
	options.lines = options.lines ? options.lines : 1;
	options.lineLength = options.lineLength ? options.lineLength : 36;
	let data = sonixParser.parsedJSON;
	let tempStrData = [];
	let sentences = [];
	
	data.forEach(function (e) {
		let currSent = '';
		let curSentObj = {};
		e.speach.forEach(function (t) {
			if (!curSentObj.from) curSentObj.from = secondsToHms(t.time);
			currSent += ' ' + t.text;
			curSentObj.lines = curSentObj.lines ? curSentObj.lines : 1;

			let currLine = currSent.split('\n')[curSentObj.lines - 1];
			if (t.text.match(/(\.|\?|!)/) && currLine.length > 10 || currLine.length > options.lineLength) {
				if (curSentObj.lines >= options.lines) {
					curSentObj.to = secondsToHms(t.time);
					curSentObj.text = currSent.replace(/ +/g,' ').trim();
					
					if (options.named) {
						curSentObj.text = e.speakerName.toUpperCase() 
										+ (options.nameOnAnotherLine ? ':\n' : ': ') 
										+ curSentObj.text;
					}

					sentences.push(curSentObj);
					currSent = '';
			 		curSentObj = {};
				} else {
					currSent += '\n';
					curSentObj.lines++;
				}
			}
		})
	});
	// console.log(sentences);
	return sentences;
}

sonixParser.getSRT = function(options) {
	sonixParser.parse();
	let data = prepare(options);
	let SRTstring = '';

	data.forEach(function (e, i, arr) {
		SRTstring += (i + 1) + '\n';
		SRTstring += e.from + ' --> ' + (arr[i+1] ? arr[i+1].from : e.to)  + '\n';
		SRTstring += e.text + '\n\n';

	});

	return SRTstring;
}

sonixParser.getSRTByReplies = function(options) {
	sonixParser.parse();
	let data = sonixParser.parsedJSON;
    let SRTstring = '';

	data.forEach(function (e, i, arr) {
		SRTstring += (i + 1) + '\n';
		SRTstring += secondsToHms(e.time) + ' --> ' + (arr[i+1] ? secondsToHms(arr[i+1].time) : secondsToHms(e.time + 10))  + '\n';
		if (options.named) {
			SRTstring += e.speakerName.toUpperCase() 
					  + (options.nameOnAnotherLine ? ':\n' : ': ');
		}
		SRTstring += e.text + '\n\n';
	})
	return SRTstring;
}



function secondsToHms(d) {
	var mill = d.split('.')[1]
    d = Number(d);
    var h = Math.floor(d / 3600);
    var m = Math.floor(d % 3600 / 60);
    var s = Math.floor(d % 3600 % 60);

    return ('0' + h).slice(-2) + ':' + ('0' + m).slice(-2) + ':' + ('0' + s).slice(-2) + ',' + ((mill ? mill : '000') + '000').slice(0,3); 
}

window.sonixParser = sonixParser;