barcodebot/parse-foodbook.js

"use strict";

const cheerio = require("cheerio");

let onclickRegex = /^loadDetailPage\('([^']+)'\);/;
let urlRegex = /url\('([^']+)'\)/;

function getOnclickUrl(onclickText) {
	let match = onclickRegex.exec(onclickText);

	if (match == null) {
		throw new Error("Expected URL not found in onclick handler");
	} else {
		return match[1];
	}
}

function getUrl(cssText) {
	let match = urlRegex.exec(cssText);

	if (match == null) {
		return null; /* Sometimes there is no picture */
	} else {
		return match[1];
	}
}

module.exports = {
	parseSearch: function (html) {
		let $ = cheerio.load(html);

		let products = $(".productresult .media").get().map((row) => {
			let $row = $(row);
			let onclickValue = $row.attr("onclick");

			return getOnclickUrl(onclickValue);
		});

		return products;
	},
	parseProduct: function (html) {
		let $ = cheerio.load(html);

		let brandTextElement = $(".section-data h1 a#brandlink");
		let brand = brandTextElement.text().trim();
		brandTextElement.remove();

		let product = $(".section-data h1").text().trim();

		return {
			name: `${brand}, ${product}`,
			image: getUrl($(".assetImageDefault").attr("style"))
		}
	}
};