신일섭

옥션 크롤링 기능 업데이트

const puppeteer = require('puppeteer');
const readline = require('readline');
// Gmarket Cart Crawling
(async () => {
// launching headless browser
const browser = await puppeteer.launch();
// making a new page
const page = await browser.newPage();
//console input id & password
var args = process.argv;
var a_id = args[2];
var a_pw = args[3];
console.log(args[2] + " " + args[3]);
// Gmarket login page
await page.goto('https://memberssl.auction.co.kr/Authenticate/?url=http%3a%2f%2fwww.auction.co.kr%2f&return_value=0');
// Ading user information
await page.evaluate((id, pwd) => {
document.querySelector('#id').value = id;
document.querySelector('#password').value = pwd;
}, a_id, a_pw);
await console.log(a_id,a_pw);
// try login
// await page.keyboard.press(String.fromCharCode(13));
await page.click('.btn_login');
await page.waitForNavigation();
// goto cart page
await page.goto('https://cart.auction.co.kr/ko/cart');
await page.screenshot({ path: 'auction.png', fullPage:true });
// container which will hold crawled data [{},{}...]
let data = [];
// crawling start! (using getOne and getAll function)
data = await getAll(page);
//data.push(await getAll(page)); // data[0][n]
//logging the result
for(let index = 0; index < data.length; index++){
console.log(data[index]);
}
await browser.close();
})();
// Crawling cart informations(object) to data(array)
async function getAll(page) {
var data = [];
//const number = await page.$$eval("#cart_list > ol > li:nth-child(1) > div.cart--basket_body > div > ul > li", (data) => data.length);
const number = await page.$$eval("#cart__body > div > div.box__cart-list > ul > li", (data) => data.length);
// counting the number of the box
for (let index = 0; index < number; index++) {
data.push(await getOne(page, index + 1));
// pushing to the array
}
return Promise.resolve(data);
}
// Crawling cart information to data(object)
async function getOne(page, index) {
var data = {};
// this is example code
//data.programPeriod = await page.$eval("#iph_content > div > div.list_type_h1.web_view.mt3 > table > tbody > tr:nth-child(" + index + ") > td:nth-child(5)", (data) => data.textContent);
// product name
data.prd_name = await page.$eval("#cart__body > div > div.box__cart-list > ul > li:nth-child(" + index + ") > div > div.box__group-seller-body > div > div.box__group-item > ul > li > div > div.box__item-info > dl > dd > div.box__title > a > span", data => data.textContent);
// product price
data.prd_price = await page.$eval("#cart__body > div > div.box__cart-list > ul > li:nth-child(" + index + ") > div > div.box__group-seller-body > div > div.box__group-item > ul > li > div > div.box__item-info > dl > dd > div.box__unit-price > div > span > strong", data => data.textContent);
// product link
data.prd_link = await page.$eval("#cart__body > div > div.box__cart-list > ul > li:nth-child(" + index + ") > div > div.box__group-seller-body > div > div.box__group-item > ul > li > div > div.box__item-info > dl > dd > div.box__title > a", data => data.href);
// product image source
data.prd_img = await page.$eval("#cart__body > div > div.box__cart-list > ul > li:nth-child(" + index + ") > div > div.box__group-seller-body > div > div.box__group-item > ul > li > div > div.box__item-img > a > img", data => data.src);
// not yet
return Promise.resolve(data);
}
\ No newline at end of file