bid-tool/scrape-data-keyword/models/langtons-scrap-model.js

56 lines
1.5 KiB
JavaScript

import { extractModelId, extractNumber } from "../system/ultils.js";
import { ScrapModel } from "./scrap-model.js";
export class LangtonsScrapModel extends ScrapModel {
action = async () => {
const urlsData = this.extractUrls();
for (let item of urlsData) {
await this.page.goto(item.url);
const data = await this.getItemsInHtml(item);
const results = this.filterItemByKeyword(item.keyword, data);
this.results[item.keyword] = results;
console.log({ results: this.results });
}
};
getItemsInHtml = async (data) => {
const elements = await this.page.$$(".row.product-grid.grid-view");
const results = [];
for (const el of elements) {
const url = await el
.$eval("a.js-pdp-link.pdp-link-anchor", (el) => el.getAttribute("href"))
.catch(() => null);
const image_url = await el
.$eval("img.tile-image.loaded", (img) => img.getAttribute("src"))
.catch(() => null);
const name = await el
.$eval(".link.js-pdp-link", (el) => el.textContent.trim())
.catch(() => null);
const current_price = await el
.$eval("div.max-bid-price.price", (el) => el.textContent.trim())
.catch(() => null);
results.push({
url: `${this.web_bid.origin_url}${url}`,
image_url,
name,
keyword: data.keyword,
model: extractModelId(`${this.web_bid.origin_url}${url}`),
current_price: extractNumber(current_price),
scrap_config_id: this.scrap_config_id,
});
}
return results;
};
}