230 lines
7.7 KiB
TypeScript
230 lines
7.7 KiB
TypeScript
|
import * as puppeteer from 'puppeteer';
|
|||
|
|
import { Logger } from '@nestjs/common';
|
|||
|
|
|
|||
|
|
// 模拟人类鼠标移动
|
|||
|
|
async function simulateHumanMouseMovement(page: puppeteer.Page) {
|
|||
|
|
const viewport = page.viewport();
|
|||
|
|
if (!viewport) return;
|
|||
|
|
|
|||
|
|
const movements = 5 + Math.floor(Math.random() * 5); // 5-10次随机移动
|
|||
|
|
|
|||
|
|
for (let i = 0; i < movements; i++) {
|
|||
|
|
const x = Math.floor(Math.random() * viewport.width);
|
|||
|
|
const y = Math.floor(Math.random() * viewport.height);
|
|||
|
|
|
|||
|
|
await page.mouse.move(x, y, {
|
|||
|
|
steps: 10 + Math.floor(Math.random() * 20) // 10-30步，使移动更平滑
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
// 随机停顿 100-500ms
|
|||
|
|
await new Promise(r => setTimeout(r, 100 + Math.random() * 400));
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 模拟人类滚动
|
|||
|
|
async function simulateHumanScrolling(page: puppeteer.Page) {
|
|||
|
|
const scrollCount = 3 + Math.floor(Math.random() * 5); // 3-7次滚动
|
|||
|
|
|
|||
|
|
for (let i = 0; i < scrollCount; i++) {
|
|||
|
|
const scrollDistance = 100 + Math.floor(Math.random() * 400); // 100-500px
|
|||
|
|
|
|||
|
|
await page.evaluate((distance) => {
|
|||
|
|
window.scrollBy({
|
|||
|
|
top: distance,
|
|||
|
|
behavior: 'smooth'
|
|||
|
|
});
|
|||
|
|
}, scrollDistance);
|
|||
|
|
|
|||
|
|
// 随机停顿 500-1500ms
|
|||
|
|
await new Promise(r => setTimeout(r, 500 + Math.random() * 1000));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
// 滚动回顶部
|
|||
|
|
await page.evaluate(() => {
|
|||
|
|
window.scrollTo({ top: 0, behavior: 'smooth' });
|
|||
|
|
});
|
|||
|
|
await new Promise(r => setTimeout(r, 1000));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
export interface CdtResult {
|
|||
|
|
title: string;
|
|||
|
|
publishDate: Date;
|
|||
|
|
url: string;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
export const CdtCrawler = {
|
|||
|
|
name: '中国大唐集团电子商务平台',
|
|||
|
|
url: 'https://tang.cdt-ec.com/home/index.html',
|
|||
|
|
baseUrl: 'https://tang.cdt-ec.com',
|
|||
|
|
|
|||
|
|
async crawl(browser: puppeteer.Browser): Promise<CdtResult[]> {
|
|||
|
|
const logger = new Logger('CdtCrawler');
|
|||
|
|
const page = await browser.newPage();
|
|||
|
|
|
|||
|
|
const username = process.env.PROXY_USERNAME;
|
|||
|
|
const password = process.env.PROXY_PASSWORD;
|
|||
|
|
if (username && password) {
|
|||
|
|
await page.authenticate({ username, password });
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
await page.setUserAgent('Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36');
|
|||
|
|
|
|||
|
|
const allResults: CdtResult[] = [];
|
|||
|
|
let currentPage = 1;
|
|||
|
|
const maxPages = 5;
|
|||
|
|
|
|||
|
|
try {
|
|||
|
|
logger.log(`Navigating to ${this.url}...`);
|
|||
|
|
await page.goto(this.url, { waitUntil: 'networkidle2', timeout: 60000 });
|
|||
|
|
|
|||
|
|
// 模拟人类行为
|
|||
|
|
logger.log('Simulating human mouse movements...');
|
|||
|
|
await simulateHumanMouseMovement(page);
|
|||
|
|
|
|||
|
|
logger.log('Simulating human scrolling...');
|
|||
|
|
await simulateHumanScrolling(page);
|
|||
|
|
|
|||
|
|
// 点击"招标公告"标签
|
|||
|
|
logger.log('Looking for "招标公告" tab...');
|
|||
|
|
await page.waitForFunction(() => {
|
|||
|
|
const tabs = Array.from(document.querySelectorAll('span.notice-tab'));
|
|||
|
|
return tabs.some(tab => tab.textContent && tab.textContent.includes('招标公告'));
|
|||
|
|
}, { timeout: 30000 });
|
|||
|
|
|
|||
|
|
await page.evaluate(() => {
|
|||
|
|
const tabs = Array.from(document.querySelectorAll('span.notice-tab'));
|
|||
|
|
const target = tabs.find(tab => tab.textContent && tab.textContent.includes('招标公告')) as HTMLElement;
|
|||
|
|
if (target) target.click();
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
logger.log('Clicked "招标公告" tab.');
|
|||
|
|
await new Promise(r => setTimeout(r, 2000));
|
|||
|
|
|
|||
|
|
// 模拟人类行为
|
|||
|
|
logger.log('Simulating human mouse movements...');
|
|||
|
|
await simulateHumanMouseMovement(page);
|
|||
|
|
|
|||
|
|
logger.log('Simulating human scrolling...');
|
|||
|
|
await simulateHumanScrolling(page);
|
|||
|
|
|
|||
|
|
// 点击"招标公告"下的"更多+"链接
|
|||
|
|
logger.log('Looking for "更多+" link under "招标公告"...');
|
|||
|
|
await page.waitForFunction(() => {
|
|||
|
|
const titles = Array.from(document.querySelectorAll('span.h-notice-title'));
|
|||
|
|
return titles.some(title => title.textContent && title.textContent.includes('招标公告'));
|
|||
|
|
}, { timeout: 30000 });
|
|||
|
|
|
|||
|
|
await page.evaluate(() => {
|
|||
|
|
const titles = Array.from(document.querySelectorAll('span.h-notice-title'));
|
|||
|
|
const targetTitle = titles.find(title => title.textContent && title.textContent.includes('招标公告'));
|
|||
|
|
if (targetTitle) {
|
|||
|
|
const parent = targetTitle.parentElement;
|
|||
|
|
if (parent) {
|
|||
|
|
const moreLink = parent.querySelector('a.h-notice-more') as HTMLElement;
|
|||
|
|
if (moreLink) moreLink.click();
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
logger.log('Clicked "更多+" link under "招标公告".');
|
|||
|
|
await page.waitForNavigation({ waitUntil: 'networkidle2', timeout: 60000 }).catch(() => {});
|
|||
|
|
await new Promise(r => setTimeout(r, 3000));
|
|||
|
|
|
|||
|
|
// 模拟人类行为
|
|||
|
|
logger.log('Simulating human mouse movements...');
|
|||
|
|
await simulateHumanMouseMovement(page);
|
|||
|
|
|
|||
|
|
logger.log('Simulating human scrolling...');
|
|||
|
|
await simulateHumanScrolling(page);
|
|||
|
|
|
|||
|
|
while (currentPage <= maxPages) {
|
|||
|
|
const content = await page.content();
|
|||
|
|
const pageResults = this.extract(content);
|
|||
|
|
if (pageResults.length === 0) {
|
|||
|
|
logger.warn(`No results found on page ${currentPage}, stopping.`);
|
|||
|
|
break;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
allResults.push(...pageResults);
|
|||
|
|
logger.log(`Extracted ${pageResults.length} items from page ${currentPage}`);
|
|||
|
|
|
|||
|
|
// 模拟人类行为 - 翻页前
|
|||
|
|
logger.log('Simulating human mouse movements before pagination...');
|
|||
|
|
await simulateHumanMouseMovement(page);
|
|||
|
|
|
|||
|
|
logger.log('Simulating human scrolling before pagination...');
|
|||
|
|
await simulateHumanScrolling(page);
|
|||
|
|
|
|||
|
|
// Find the "Next Page" button - layui pagination
|
|||
|
|
const nextButtonSelector = 'a.layui-laypage-next:not(.layui-disabled)';
|
|||
|
|
const nextButton = await page.$(nextButtonSelector);
|
|||
|
|
|
|||
|
|
if (!nextButton) {
|
|||
|
|
logger.log('Next page button not found. Reached end of list.');
|
|||
|
|
break;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
logger.log(`Navigating to page ${currentPage + 1}...`);
|
|||
|
|
|
|||
|
|
try {
|
|||
|
|
await Promise.all([
|
|||
|
|
page.waitForNavigation({ waitUntil: 'networkidle2', timeout: 60000 }),
|
|||
|
|
nextButton.click(),
|
|||
|
|
]);
|
|||
|
|
} catch (navError) {
|
|||
|
|
logger.error(`Navigation to page ${currentPage + 1} failed: ${navError.message}`);
|
|||
|
|
break;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
currentPage++;
|
|||
|
|
|
|||
|
|
// 模拟人类行为 - 翻页后
|
|||
|
|
logger.log('Simulating human mouse movements after pagination...');
|
|||
|
|
await simulateHumanMouseMovement(page);
|
|||
|
|
|
|||
|
|
logger.log('Simulating human scrolling after pagination...');
|
|||
|
|
await simulateHumanScrolling(page);
|
|||
|
|
|
|||
|
|
// Random delay between pages
|
|||
|
|
const delay = Math.floor(Math.random() * (3000 - 1000 + 1)) + 1000;
|
|||
|
|
await new Promise(resolve => setTimeout(resolve, delay));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return allResults;
|
|||
|
|
|
|||
|
|
} catch (error) {
|
|||
|
|
logger.error(`Failed to crawl ${this.name}: ${error.message}`);
|
|||
|
|
return allResults;
|
|||
|
|
} finally {
|
|||
|
|
await page.close();
|
|||
|
|
}
|
|||
|
|
},
|
|||
|
|
|
|||
|
|
extract(html: string): CdtResult[] {
|
|||
|
|
const results: CdtResult[] = [];
|
|||
|
|
/**
|
|||
|
|
* Regex groups for tang.cdt-ec.com:
|
|||
|
|
* 1: URL
|
|||
|
|
* 2: Title (项目名称)
|
|||
|
|
* 3: Date (发布时间)
|
|||
|
|
*/
|
|||
|
|
const regex = /<tr[^>]*data-index="[^"]*"[^>]*>[\s\S]*?<a[^>]*class="layui-table-link"[^>]*href="([^"]*)"[^>]*>([^<]*)<\/a>[\s\S]*?<td[^>]*data-field="publish_time"[^>]*>[\s\S]*?<div[^>]*class="layui-table-cell[^"]*"[^>]*>([^<]*)<\/div>[\s\S]*?<\/td>[\s\S]*?<\/tr>/gs;
|
|||
|
|
|
|||
|
|
let match;
|
|||
|
|
while ((match = regex.exec(html)) !== null) {
|
|||
|
|
const url = match[1]?.trim();
|
|||
|
|
const title = match[2]?.trim();
|
|||
|
|
const dateStr = match[3]?.trim();
|
|||
|
|
|
|||
|
|
if (title && url) {
|
|||
|
|
results.push({
|
|||
|
|
title,
|
|||
|
|
publishDate: dateStr ? new Date(dateStr) : new Date(),
|
|||
|
|
url: url.startsWith('http') ? url : this.baseUrl + url
|
|||
|
|
});
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
return results;
|
|||
|
|
}
|
|||
|
|
};
|