const phantom = require('phantom');
const fs = require('fs');
const cheerio = require('cheerio');
const request = require('request');

class ImageScraper {
    constructor() {
        this.searchUrl = 'https://image.baidu.com/search/index?ct=201326592&z=&tn=baiduimage&word=%E6%BC%AB%E5%A8%81%E5%9B%BE%E7%89%87&pn=0&ie=utf-8&oe=utf-8&cl=2&lm=-1&fr=ala&se=&sme=&width=1920&height=1080';
        this.currentPage = 1;
        this.imageLinks = [];
        this.thumbnailElements = [];
    }

    async initialize() {
        const phantomInstance = await phantom.create();
        try {
            await this.loadWebpage();
            await this.extractThumbnails();
            await this.downloadHighResImages();
        } catch (error) {
            console.error('Scraping error:', error);
        }
    }

    wait(seconds) {
        return new Promise((resolve) => {
            setTimeout(resolve, seconds * 1000);
        });
    }

    pause(milliseconds) {
        return new Promise((resolve) => {
            console.log(`Pausing for ${milliseconds / 1000} seconds...`);
            setTimeout(resolve, milliseconds);
        });
    }

    async loadWebpage() {
        await this.scrollPage(0);
    }

    async scrollPage(scrollCount) {
        const phantomInstance = await phantom.create();
        const page = await phantomInstance.createPage();
        
        await page.open(this.searchUrl);
        await page.property('viewportSize', {
            width: 1920,
            height: 1080
        });
        
        await this.wait(5);
        
        await page.property('scrollPosition', {
            left: 0,
            top: 1000 * scrollCount
        });

        const pageContent = await page.property('content');
        const $ = cheerio.load(pageContent);
        
        console.log('Thumbnail count:', $('.imgbox').length);
        
        if ($('.imgbox').length < 20) {
            await this.scrollPage(++scrollCount);
        }
        
        this.thumbnailElements = $('.imgitem');
    }

    async extractThumbnails() {
        const phantomInstance = await phantom.create();
        const page = await phantomInstance.createPage();
        const pageContent = await page.property('content');
        const $ = cheerio.load(pageContent);

        this.thumbnailElements.each((index, element) => {
            this.imageLinks.push({
                detailUrl: 'https://image.baidu.com' + $(element).find('a').attr('href')
            });
        });
    }

    async downloadHighResImages() {
        const phantomInstance = await phantom.create();
        const page = await phantomInstance.createPage();
        
        for (let i = 0; i < this.imageLinks.length; i++) {
            try {
                await page.open(this.imageLinks[i].detailUrl);
                await this.wait(2);
                
                const pageContent = await page.property('content');
                const $ = cheerio.load(pageContent);
                
                const imageSrc = $('#currentImg').attr('src');
                const fileExtension = imageSrc.split('.').pop();
                
                console.log('Downloading image', i + 1);
                request(imageSrc).pipe(fs.createWriteStream(`./downloads/${Date.now()}.${fileExtension}`));
                
                await this.pause(3000);
                console.log('Download completed');
                
                if (i === this.imageLinks.length - 1) {
                    phantomInstance.exit();
                }
            } catch (error) {
                console.error('Download error:', error);
            }
        }
    }
}

const scraper = new ImageScraper();
scraper.initialize();