train_extract.js

let rdx = document.getElementById('rdx');
let c = rdx.children;
let arr = [];

for (i of c) {
	let x = i.children;
	if (x.length < 4)
		continue;
	if (!x[0].innerText || !x[1].innerText || !x[2].innerText)
		continue;

	let code = x[0].innerText;
	let name = x[1].innerText;
	let kanji = x[2].innerText;

	let tmp = code.split('\n');
	if (tmp.length == 3)
		tmp = [tmp[1], tmp[2]];

	code = tmp.join(' ');
	if (code.substr(0, 3) != 'KK ') {
		code = x[1].innerText;
		name = x[2].innerText;
		kanji = x[3].innerText;

		tmp = code.split('\n');
		if (tmp.length == 3)
			tmp = [tmp[1], tmp[2]];

		code = tmp.join(' ');
		if (code.substr(0, 3) != 'KK ')
			continue;
	}

	let obj = {
		n: code.replace(' ', ''),
		romaji: name.toLowerCase(),
		kanji: kanji
	};

	arr.push(obj);
	console.log(obj);
}

// Sort by code
arr.sort((a, b) => {
	let x = a.code;
	let y = b.code;
	return x < y ? -1 : x > y ? 1 : 0;
});

// Clear duplicates
let tmp = {};
arr = arr.filter((obj) => {
	if (tmp[obj.code])
		return false;
	tmp[obj.code] = true;
	return true;
});

console.log(JSON.stringify(arr, null, 2));