added report system

This commit is contained in:
Nick Leeman
2020-11-20 19:51:19 +01:00
parent 4fb8a19661
commit b617ac85fe
5 changed files with 142 additions and 10 deletions

View File

@@ -1,6 +1,7 @@
import puppeteer from "puppeteer";
import puppeteer, { product } from "puppeteer";
import jsonfile from "jsonfile";
import cheerio from "cheerio";
import fs from "fs-extra";
import { getDomain } from "tldts";
// Import website modules
@@ -13,6 +14,8 @@ start();
// Globals
let browser: puppeteer.Browser = null;
let allProducts: Array<any> = [];
let allBlogUrls: Array<any> = [];
async function start() {
try {
@@ -28,9 +31,12 @@ async function start() {
console.log(`------------------------------------- \n`);
for (let url of urls) {
allBlogUrls.push(url);
await crawlBlogPage(url);
}
await browser.close();
await generateReport();
} catch (error) {
console.error(`An Error Occured!`, error);
}
@@ -58,13 +64,13 @@ async function crawlBlogPage(url: string) {
console.log(`Detected ${$(".row-products").length} content egg products!`);
$(".row-products .cegg-list-logo-title a").each((index, element) => {
let url = $(element).attr("href");
let productUrl = $(element).attr("href");
let name = $(element).html().trim();
products.push({
name: name,
blogUrl: url,
url: "",
url: productUrl,
domain: "",
inStock: false,
});
@@ -73,7 +79,7 @@ async function crawlBlogPage(url: string) {
console.log("Checking product stocks...");
for (let index in products) {
let status = await crawlProductStock(products[index].blogUrl);
let status = await crawlProductStock(products[index].url);
products[index].domain = status[0];
products[index].url = status[1];
@@ -84,6 +90,8 @@ async function crawlBlogPage(url: string) {
} else {
console.log(`${products[index].name} - ${products[index].domain}`);
}
allProducts.push(products[index]);
}
} catch (error) {
@@ -125,5 +133,44 @@ async function crawlProductStock(url: string) {
console.log("Error while trying to crawl page! Skipping...");
}
}
async function generateReport() {
console.log("Generating report...");
let file = "./report.txt";
let totalProducts = 0;
let totalOutStock = 0;
for (let product of allProducts) {
totalProducts++;
if (product.inStock == false) {
totalOutStock++;
}
}
await fs.appendFile(file, `----------------------------------------\n`);
await fs.appendFile(file, `Content Egg Product Stock Crawler Report\n`);
await fs.appendFile(file, `Total Blog Urls: ${allBlogUrls.length}\n`);
await fs.appendFile(file, `Total Products: ${totalProducts}\n`);
await fs.appendFile(file, `Total Out of Stock: ${totalOutStock}\n`);
await fs.appendFile(file, `----------------------------------------\n`);
await fs.appendFile(file, ``);
let lastBlogUrl = "";
for (let product of allProducts) {
if (product.blogUrl != lastBlogUrl) {
await fs.appendFile(file, `\n\n`);
await fs.appendFile(file, `${product.blogUrl}\n`);
await fs.appendFile(file, `----------------------------------\n`);
}
if (product.inStock == false) {
await fs.appendFile(file, `${product.domain} - ${product.name} - ${product.url}\n`);
}
lastBlogUrl = product.blogUrl;
}
console.log("Report generated!");
}

View File

@@ -25,6 +25,7 @@ export namespace FitnessKoerier {
return true;
}
} catch (error) {
console.log(error);
console.error(`Error occured during stock check!`);
return false;
}
@@ -36,7 +37,11 @@ export namespace FitnessApparaat {
export async function check(html: string) {
try {
const $ = cheerio.load(html);
console.log($('[itemprop="offers"').find(".stock-red").length);
if ($('[itemprop="offers"]').find(".stock-red").length >= 1) {
return false;
} else {
return true;
}
} catch (error) {
console.log(error);
console.error(`Error occured during stock check!`);