Update scrap
This commit is contained in:
nguyentrungthat
parent
3b0f4992a2
commit
72b463714d
|
|
@ -13,6 +13,7 @@
|
|||
"cheerio": "^1.1.0",
|
||||
"dayjs": "^1.11.18",
|
||||
"dotenv": "^17.2.3",
|
||||
"fs": "^0.0.1-security",
|
||||
"googleapis": "^166.0.0",
|
||||
"luxon": "^3.7.2",
|
||||
"mysql2": "^3.15.2",
|
||||
|
|
@ -1507,6 +1508,12 @@
|
|||
"node": ">=0.8"
|
||||
}
|
||||
},
|
||||
"node_modules/fs": {
|
||||
"version": "0.0.1-security",
|
||||
"resolved": "https://registry.npmjs.org/fs/-/fs-0.0.1-security.tgz",
|
||||
"integrity": "sha512-3XY9e1pP0CVEUCdj5BmfIZxRBTSDycnbqhIOGec9QYtmVH2fbLpj86CFWkrNOkt/Fvty4KZG5lTglL9j/gJ87w==",
|
||||
"license": "ISC"
|
||||
},
|
||||
"node_modules/fs-constants": {
|
||||
"version": "1.0.0",
|
||||
"resolved": "https://registry.npmjs.org/fs-constants/-/fs-constants-1.0.0.tgz",
|
||||
|
|
|
|||
|
|
@ -13,6 +13,7 @@
|
|||
"cheerio": "^1.1.0",
|
||||
"dayjs": "^1.11.18",
|
||||
"dotenv": "^17.2.3",
|
||||
"fs": "^0.0.1-security",
|
||||
"googleapis": "^166.0.0",
|
||||
"luxon": "^3.7.2",
|
||||
"mysql2": "^3.15.2",
|
||||
|
|
|
|||
|
|
@ -3,17 +3,19 @@ import puppeteer from "puppeteer";
|
|||
import axios from "axios";
|
||||
import mysql from "mysql2/promise";
|
||||
import { DateTime } from "luxon";
|
||||
import path from "node:path";
|
||||
import fs from "node:fs";
|
||||
|
||||
// Define function promise waiting for a given time
|
||||
async function wait(ms) {
|
||||
return new Promise((resolve) => setTimeout(resolve, ms));
|
||||
}
|
||||
const LOG_FILE = path.join(process.cwd(), "newitems.log");
|
||||
const MAX_LOG_SIZE = 20 * 1024 * 1024; // 20MB
|
||||
|
||||
// --- HELPERS ---
|
||||
const wait = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
|
||||
|
||||
async function safeGetContent(page) {
|
||||
try {
|
||||
return await page.content();
|
||||
} catch (err) {
|
||||
// Nếu navigation xảy ra → chờ page ổn định lại rồi đọc tiếp
|
||||
if (err.message.includes("Execution context was destroyed")) {
|
||||
await page.waitForNetworkIdle({ idleTime: 1000 }).catch(() => {});
|
||||
return await page.content();
|
||||
|
|
@ -24,71 +26,107 @@ async function safeGetContent(page) {
|
|||
|
||||
function parseAndConvertToSydney(dateString, fromTimezone, formatDate) {
|
||||
const currentYear = new Date().getFullYear();
|
||||
|
||||
// Detect format: "Dec-9 23:05" or "9-Dec 23:05"
|
||||
let format = "MMM-d HH:mm";
|
||||
if (/^\d/.test(dateString[0])) {
|
||||
format = "d-MMM HH:mm";
|
||||
}
|
||||
|
||||
let format = /^\d/.test(dateString[0]) ? "d-MMM HH:mm" : "MMM-d HH:mm";
|
||||
const fullDateString = `${dateString} ${currentYear}`;
|
||||
|
||||
const dt = DateTime.fromFormat(fullDateString, `${format} yyyy`, { zone: fromTimezone });
|
||||
|
||||
if (!dt.isValid) {
|
||||
// console.log("❌ Invalid Luxon parse:", dt.invalidReason, fullDateString, format);
|
||||
return null;
|
||||
}
|
||||
|
||||
if (!dt.isValid) return null;
|
||||
const systemTz = Intl.DateTimeFormat().resolvedOptions().timeZone;
|
||||
return dt.setZone(systemTz).toFormat(formatDate);
|
||||
}
|
||||
|
||||
async function scrapeWithPuppeteer(store) {
|
||||
// console.log(`Fetching with scrapeWithPuppeteer`);
|
||||
try {
|
||||
const browser = await puppeteer.launch({
|
||||
headless: true,
|
||||
args: ["--no-sandbox", "--disable-setuid-sandbox"],
|
||||
});
|
||||
const page = await browser.newPage();
|
||||
await page.goto(store.data, { waitUntil: "networkidle2" });
|
||||
// --- LOGIC LẤY CONFIG TỪ DB ---
|
||||
async function getConfigs(pool) {
|
||||
let configs = [];
|
||||
const [keywords] = await pool.execute("SELECT id, name FROM hot_item_keyword WHERE level = 1");
|
||||
const [keywordsCustom] = await pool.execute("SELECT id, name, url FROM newitems_config");
|
||||
const [markets] = await pool.execute("SELECT id, url, market_code, shipping_postcode, timezone, country_name FROM ebay_site WHERE flag = 1");
|
||||
|
||||
const keyWord = keywords?.map((el) => el.name)?.join("+");
|
||||
|
||||
markets.forEach((m) => {
|
||||
const postCode = m.shipping_postcode ? `&_stpos=${m.shipping_postcode}` : "";
|
||||
// URL FixedPrice
|
||||
configs.push({
|
||||
data: `${m.url}/sch/i.html?_from=R40&_nkw=${keyWord}&_sacat=0&_sop=10${postCode}`,
|
||||
type_custom: "cisco",
|
||||
config_id: null,
|
||||
from_site: m.market_code,
|
||||
timezone: m.timezone,
|
||||
listingType: "FixedPrice",
|
||||
});
|
||||
// URL Auction
|
||||
configs.push({
|
||||
data: `${m.url}/sch/i.html?_from=R40&_nkw=${keyWord}&_sacat=0&_sop=1${postCode}`,
|
||||
type_custom: "cisco",
|
||||
config_id: null,
|
||||
from_site: m.market_code,
|
||||
timezone: m.timezone,
|
||||
listingType: "Auction",
|
||||
});
|
||||
});
|
||||
|
||||
if (keywordsCustom.length > 0) {
|
||||
keywordsCustom.forEach((k) => {
|
||||
const matchedMarket = findMarketDataFromSearchUrl(k.url, markets);
|
||||
if (matchedMarket) {
|
||||
configs.push({
|
||||
data: k.url,
|
||||
type_custom: "custom",
|
||||
config_id: k.id,
|
||||
from_site: matchedMarket.market_code,
|
||||
timezone: matchedMarket.timezone,
|
||||
listingType: k.url?.includes("_sop=10") ? "FixedPrice" : "Auction",
|
||||
});
|
||||
}
|
||||
});
|
||||
}
|
||||
return configs;
|
||||
}
|
||||
|
||||
function findMarketDataFromSearchUrl(searchUrl, marketDatas) {
|
||||
if (!searchUrl?.trim()) return null;
|
||||
try {
|
||||
const searchHost = new URL(searchUrl.trim()).host;
|
||||
for (const data of marketDatas) {
|
||||
if (searchUrl.includes(data.url)) return data;
|
||||
const marketHost = new URL(data.url).host;
|
||||
if (searchHost === marketHost) return data;
|
||||
}
|
||||
} catch (e) {
|
||||
return null;
|
||||
}
|
||||
return null;
|
||||
}
|
||||
|
||||
// --- CORE SCRAPER ---
|
||||
async function scrapeWithPuppeteer(browser, store) {
|
||||
let page = null;
|
||||
try {
|
||||
page = await browser.newPage();
|
||||
|
||||
// Tối ưu RAM: Chặn các request không cần thiết
|
||||
await page.setRequestInterception(true);
|
||||
page.on("request", (req) => {
|
||||
if (["image", "stylesheet", "font", "media"].includes(req.resourceType())) req.abort();
|
||||
else req.continue();
|
||||
});
|
||||
|
||||
await page.goto(store.data, { waitUntil: "networkidle2", timeout: 60000 });
|
||||
|
||||
// ----- RETRY HANDLE (Cloudflare / Slow load) -----
|
||||
let retries = 0;
|
||||
while (retries < 10) {
|
||||
try {
|
||||
// console.log(`Retry ${retries + 1}`);
|
||||
const html = await safeGetContent(page);
|
||||
|
||||
// Detect Cloudflare or other blocking messages
|
||||
if (html.includes("Checking your browser")) {
|
||||
await wait(2000);
|
||||
retries++;
|
||||
continue;
|
||||
}
|
||||
|
||||
const hasItems = await page.$("li.s-card--horizontal");
|
||||
if (hasItems) break;
|
||||
|
||||
await wait(2000);
|
||||
retries++;
|
||||
} catch (err) {
|
||||
if (await page.$("li.s-card--horizontal")) break;
|
||||
await wait(2000);
|
||||
retries++;
|
||||
}
|
||||
}
|
||||
|
||||
// Wait for cards or detect Cloudflare
|
||||
const html = await safeGetContent(page);
|
||||
const needBrowserCheck = html.includes("Checking your browser");
|
||||
if (needBrowserCheck) {
|
||||
await page.waitForSelector("li.s-card--horizontal", { timeout: 15000 }).catch(() => null);
|
||||
}
|
||||
|
||||
// ================================================
|
||||
// MAIN SCRAPING LOGIC (FULL CONVERT FROM PHP)
|
||||
// ================================================
|
||||
const items = await page?.$$eval(
|
||||
"li.s-card--horizontal",
|
||||
(nodes, store) => {
|
||||
|
|
@ -117,11 +155,12 @@ async function scrapeWithPuppeteer(store) {
|
|||
nodes.forEach((node) => {
|
||||
const payload = {};
|
||||
// ---------------- LINK + ID ----------------
|
||||
const linkEl = node.querySelector("div.su-media__image a");
|
||||
const linkEl = node.querySelector("div.su-image a");
|
||||
// if (!linkEl) return;
|
||||
|
||||
payload.link_detail = linkEl?.href || "";
|
||||
const idMatch = linkEl.href.match(/\/itm\/(\d+)/);
|
||||
const linkDetail = linkEl && linkEl?.href ? linkEl?.href : "";
|
||||
if (!linkDetail) return;
|
||||
payload.link_detail = linkDetail;
|
||||
const idMatch = linkDetail.match(/\/itm\/(\d+)/);
|
||||
if (!idMatch) return;
|
||||
|
||||
payload.id = idMatch[1];
|
||||
|
|
@ -271,146 +310,87 @@ async function scrapeWithPuppeteer(store) {
|
|||
store
|
||||
);
|
||||
|
||||
const results = items.map((item) => {
|
||||
return items.map((item) => {
|
||||
if (!item.time) return item;
|
||||
|
||||
// PROCESS START TIME + END TIME
|
||||
const timeConvert = parseAndConvertToSydney(item.time, store.timezone, "yyyy/MM/dd HH:mm");
|
||||
const timestamp = new Date(timeConvert).getTime() / 1000;
|
||||
|
||||
item.timeConvert = timeConvert;
|
||||
item.start_time = timestamp;
|
||||
item.end_time = timestamp + 2592000;
|
||||
item.start_time_string = timeConvert;
|
||||
// delete item.time;
|
||||
return {
|
||||
...item,
|
||||
};
|
||||
const ts = new Date(timeConvert).getTime() / 1000;
|
||||
return { ...item, timeConvert, start_time: ts, end_time: item.end_time || ts + 2592000, start_time_string: timeConvert };
|
||||
});
|
||||
|
||||
await browser.close();
|
||||
return results;
|
||||
} catch (err) {
|
||||
console.log("Error scrapeWithPuppeteer:", err);
|
||||
console.error(`Lỗi tại ${store.data}:`, err.message);
|
||||
return [];
|
||||
} finally {
|
||||
if (page) await page.close();
|
||||
}
|
||||
}
|
||||
|
||||
function findMarketDataFromSearchUrl(searchUrl, marketDatas) {
|
||||
if (!searchUrl || !searchUrl.trim()) {
|
||||
return null;
|
||||
}
|
||||
|
||||
searchUrl = searchUrl.trim();
|
||||
|
||||
let searchHost;
|
||||
function trimLogFileIfNeeded() {
|
||||
try {
|
||||
searchHost = new URL(searchUrl).host;
|
||||
} catch (e) {
|
||||
searchHost = null;
|
||||
if (!fs.existsSync(LOG_FILE)) return;
|
||||
|
||||
const stats = fs.statSync(LOG_FILE);
|
||||
if (stats.size > MAX_LOG_SIZE) {
|
||||
const data = fs.readFileSync(LOG_FILE, "utf8");
|
||||
const lines = data.split("\n");
|
||||
// Giữ lại 70% dòng mới nhất
|
||||
const keepLines = Math.floor(lines.length * 0.7);
|
||||
const trimmed = lines.slice(-keepLines).join("\n");
|
||||
|
||||
fs.writeFileSync(LOG_FILE, trimmed + "\n", "utf8");
|
||||
console.log(`\n[SYSTEM] Log file trimmed (Size: ${(stats.size / 1024 / 1024).toFixed(2)}MB > 20MB)`);
|
||||
}
|
||||
|
||||
for (const data of marketDatas) {
|
||||
if (!data.url) continue;
|
||||
|
||||
// 1) direct substring match
|
||||
if (searchUrl.includes(data.url)) {
|
||||
return data;
|
||||
}
|
||||
|
||||
// 2) fallback: host comparison
|
||||
let marketHost = null;
|
||||
try {
|
||||
marketHost = new URL(data.url).host;
|
||||
} catch (e) {
|
||||
marketHost = null;
|
||||
}
|
||||
|
||||
if (searchHost && marketHost && searchHost === marketHost) {
|
||||
return data;
|
||||
} catch (err) {
|
||||
console.error("Error trimming log file:", err);
|
||||
}
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
// --- MAIN PROCESS ---
|
||||
async function main() {
|
||||
const db = await mysql.createConnection({
|
||||
host: process.env.MYSQL_HOST, // e.g. '127.0.0.1' or remote IP
|
||||
const pool = mysql.createPool({
|
||||
host: process.env.MYSQL_HOST,
|
||||
user: process.env.MYSQL_USER,
|
||||
password: process.env.MYSQL_PASSWORD,
|
||||
database: process.env.MYSQL_DB_NAME,
|
||||
waitForConnections: true,
|
||||
connectionLimit: 5,
|
||||
});
|
||||
|
||||
// console.log("✅ Connected to MySQL");
|
||||
|
||||
// const errors = [];
|
||||
let configs = [];
|
||||
const [keywords] = await db.execute("SELECT id, name FROM hot_item_keyword WHERE level = 1");
|
||||
const [keywordsCustom] = await db.execute("SELECT id, name, url FROM newitems_config");
|
||||
const [markets] = await db.execute("SELECT id, url, market_code, shipping_postcode, timezone, country_name FROM ebay_site WHERE flag = 1");
|
||||
const keyWord = keywords?.map((el) => el.name)?.join("+");
|
||||
markets.forEach((m) => {
|
||||
const postCode = m.shipping_postcode ? `&_stpos=${m.shipping_postcode}` : "";
|
||||
|
||||
// URL FixedPrice (_sop=10)
|
||||
configs.push({
|
||||
data: `${m.url}/sch/i.html?_from=R40&_nkw=${keyWord}&_sacat=0&_sop=10${postCode}`,
|
||||
type_custom: "cisco",
|
||||
config_id: null,
|
||||
from_site: m.market_code,
|
||||
timezone: m.timezone,
|
||||
listingType: "FixedPrice",
|
||||
const browser = await puppeteer.launch({
|
||||
headless: true,
|
||||
args: ["--no-sandbox", "--disable-setuid-sandbox", "--disable-dev-shm-usage", "--single-process"],
|
||||
});
|
||||
|
||||
// URL Auction (_sop=1)
|
||||
configs.push({
|
||||
data: `${m.url}/sch/i.html?_from=R40&_nkw=${keyWord}&_sacat=0&_sop=1${postCode}`,
|
||||
type_custom: "cisco",
|
||||
config_id: null,
|
||||
from_site: m.market_code,
|
||||
timezone: m.timezone,
|
||||
listingType: "Auction",
|
||||
});
|
||||
});
|
||||
console.log("🚀 Bắt đầu phiên làm việc mới...");
|
||||
|
||||
if (keywordsCustom.length > 0) {
|
||||
keywordsCustom.forEach((k) => {
|
||||
const matchedMarket = findMarketDataFromSearchUrl(k.url, markets);
|
||||
if (matchedMarket) {
|
||||
configs.push({
|
||||
data: k.url,
|
||||
type_custom: "custom",
|
||||
config_id: k.id,
|
||||
from_site: matchedMarket.market_code,
|
||||
timezone: matchedMarket.timezone,
|
||||
listingType: k.url?.includes("_sop=10") ? "FixedPrice" : "Auction",
|
||||
});
|
||||
}
|
||||
});
|
||||
}
|
||||
// console.log(`Total configs to process: ${configs.length}`);
|
||||
let runCount = 0;
|
||||
const MAX_RUNS = 60; // Chạy 60 chu kỳ (~1 tiếng)
|
||||
const ONE_HOUR_MS = 60 * 60 * 1000;
|
||||
const SESSION_START = Date.now(); // Thời điểm bắt đầu chạy script
|
||||
const MAX_TIME_EXTENSIONS = SESSION_START + ONE_HOUR_MS; // Thời điểm phải kết thúc
|
||||
|
||||
while (runCount < MAX_RUNS && Date.now() < MAX_TIME_EXTENSIONS) {
|
||||
runCount++;
|
||||
const startTime = Date.now();
|
||||
console.log(`--- Chu kỳ ${runCount}/${MAX_RUNS} --- ${MAX_TIME_EXTENSIONS - Date.now()} ---`);
|
||||
|
||||
try {
|
||||
const configs = await getConfigs(pool);
|
||||
for (const store of configs) {
|
||||
// console.log(`Processing ${store.data}`);
|
||||
// let items = await scrapeWithCheerio(store);
|
||||
let items = await scrapeWithPuppeteer(store);
|
||||
if (Date.now() > MAX_TIME_EXTENSIONS) break;
|
||||
const items = await scrapeWithPuppeteer(browser, store);
|
||||
for (const item of items) {
|
||||
// 2️⃣ Check if record exists
|
||||
const [rows] = await db.execute("SELECT id FROM items WHERE id = ?", [item.id]);
|
||||
if (rows.length > 0 || item.id === "123456") continue;
|
||||
|
||||
// 3️⃣ Insert new record
|
||||
const [rows] = await pool.execute("SELECT id FROM items WHERE id = ?", [item.id]);
|
||||
if (rows.length > 0) continue;
|
||||
const title = (item.title || "").replace("Opens in a new window or tab", "").trim();
|
||||
console.log(`Processing ${store.data}`);
|
||||
console.log({ ...item, title });
|
||||
await axios
|
||||
try {
|
||||
const res = await axios
|
||||
.post(
|
||||
process.env.API_DISTI_HOST + "/api/items/insert",
|
||||
`${process.env.API_DISTI_HOST}/api/items/insert`,
|
||||
{ ...item, title },
|
||||
{
|
||||
headers: {
|
||||
"x-key": "CanTho#1",
|
||||
},
|
||||
headers: { "x-key": "CanTho#1" },
|
||||
}
|
||||
)
|
||||
.then((res) => {
|
||||
|
|
@ -419,16 +399,31 @@ async function main() {
|
|||
.catch((err) => {
|
||||
console.log(err);
|
||||
});
|
||||
} catch (e) {
|
||||
console.error(`❌ Lỗi API ID ${item.id}:`, e.message);
|
||||
}
|
||||
}
|
||||
// console.log("✅ Done scraping.");
|
||||
}
|
||||
} catch (err) {
|
||||
console.error("Lỗi chu kỳ:", err.message);
|
||||
}
|
||||
|
||||
await db.end();
|
||||
// console.log("🔌 MySQL connection closed");
|
||||
const duration = Date.now() - startTime;
|
||||
const delay = Math.max(0, 60000 - duration);
|
||||
console.log(`Hoàn thành chu kỳ trong ${duration / 1000}s. Nghỉ ${delay / 1000}s.`);
|
||||
|
||||
if (runCount < MAX_RUNS && Date.now() + delay < MAX_TIME_EXTENSIONS) await wait(delay);
|
||||
}
|
||||
|
||||
console.log("🏁 Đã chạy đủ 60 lần. Đang làm mới tiến trình...");
|
||||
await browser.close();
|
||||
await pool.end();
|
||||
// Thực hiện dọn log một lần cuối trước khi thoát process hoàn toàn
|
||||
trimLogFileIfNeeded();
|
||||
process.exit(0);
|
||||
}
|
||||
|
||||
main().catch((err) => {
|
||||
console.error(err);
|
||||
console.error("FATAL ERROR:", err);
|
||||
process.exit(1);
|
||||
});
|
||||
|
|
|
|||
Loading…
Reference in New Issue