ImagesDownloader/Tampermonkey/hdk4.js

// ==UserScript==
// @name         hd4k_downloader
// @namespace    http://tampermonkey.net/
// @version      1.4
// @description  简单直接的自动翻页图片爬取
// @author       Jack
// @match        https://www.4khd.com/*
// @grant        GM_xmlhttpRequest
// ==/UserScript==

(function() {
    'use strict';

    const CONFIG = {
        maxPages: 50,
        pageDelay: 1500,
        backendUrl: 'http://127.0.0.1:55830/api/save_json'
    };

    let isCrawling = false;
    let allImages = {};
    let currentPage = 1;
    let imgIndex = 1;
    let crawledUrls = [];
    const source = 'hd4k';

    const createButton = () => {
        const button = document.createElement('button');
        button.textContent = '开始爬取';
        button.id = 'hd4k-btn';

        button.style.position = 'fixed';
        button.style.top = '14%';
        button.style.right = '1%';
        button.style.transform = 'translateY(-50%)';
        button.style.padding = '8px 16px';
        button.style.fontSize = '12px';
        button.style.fontWeight = 'bold';
        button.style.backgroundColor = '#2c80ff';
        button.style.color = '#fff';
        button.style.border = 'none';
        button.style.borderRadius = '8px';
        button.style.cursor = 'pointer';
        button.style.zIndex = '10000';
        button.style.boxShadow = '0 2px 5px rgba(0,0,0,0.2)';
        button.style.transition = 'all 0.3s ease';

        button.addEventListener('mouseenter', () => {
            if (!isCrawling) {
                button.style.backgroundColor = '#1a6ee0';
                button.style.transform = 'translateY(-50%) scale(1.05)';
            }
        });

        button.addEventListener('mouseleave', () => {
            if (!isCrawling) {
                button.style.backgroundColor = '#2c80ff';
                button.style.transform = 'translateY(-50%) scale(1)';
            }
        });

        button.addEventListener('click', startCrawling);

        return button;
    };

    const createStatusDisplay = () => {
        const statusDiv = document.createElement('div');
        statusDiv.id = 'hd4k-status';
        statusDiv.style.position = 'fixed';
        statusDiv.style.top = '18%';
        statusDiv.style.right = '1%';
        statusDiv.style.padding = '10px';
        statusDiv.style.backgroundColor = 'rgba(0,0,0,0.85)';
        statusDiv.style.color = '#fff';
        statusDiv.style.borderRadius = '5px';
        statusDiv.style.fontSize = '12px';
        statusDiv.style.zIndex = '9999';
        statusDiv.style.minWidth = '180px';
        statusDiv.style.display = 'none';

        return statusDiv;
    };

    const updateStatus = (message) => {
        const statusDiv = document.getElementById('hd4k-status');
        if (statusDiv) {
            statusDiv.innerHTML = message;
            statusDiv.style.display = 'block';
        }
        console.log(`[状态] ${message}`);
    };

    const getCurrentPageImages = () => {
        // 每次都在当前页面重新查找容器
        const container = document.querySelector('#basicExample');
        if (!container) {
            console.log('当前页面未找到图片容器 #basicExample');
            return [];
        }

        // 在容器内查找图片
        const images = container.querySelectorAll('img');
        const imageUrls = [];
        const seenUrls = new Set();

        images.forEach(img => {
            let src = img.src || img.dataset.src || img.currentSrc;

            if (src && src.trim() && !src.startsWith('data:') && !src.startsWith('blob:')) {
                // 处理URL
                let fullUrl = src;
                if (src.startsWith('//')) {
                    fullUrl = window.location.protocol + src;
                } else if (src.startsWith('/')) {
                    fullUrl = window.location.origin + src;
                } else if (!src.startsWith('http')) {
                    fullUrl = new URL(src, window.location.href).href;
                }

                const isImage = /\.(jpg|jpeg|png|gif|webp|bmp|tiff)(\?.*)?$/i.test(fullUrl);
                if (isImage && !seenUrls.has(fullUrl)) {
                    seenUrls.add(fullUrl);
                    imageUrls.push(fullUrl);
                }
            }
        });

        return imageUrls;
    };

    const buildPageUrl = (pageNum) => {
        const currentUrl = window.location.href;
        const htmlIndex = currentUrl.indexOf('html');

        if (htmlIndex === -1) {
            console.error('URL中没有找到html');
            return currentUrl;
        }

        const basePart = currentUrl.substring(0, htmlIndex + 4);

        if (pageNum === 1) {
            return basePart;
        } else {
            return basePart + '/' + pageNum;
        }
    };

    const getCurrentPageNumber = () => {
        const currentUrl = window.location.href;
        const htmlIndex = currentUrl.indexOf('html');

        if (htmlIndex === -1) return 1;

        const afterHtml = currentUrl.substring(htmlIndex + 4);
        const match = afterHtml.match(/^\/(\d+)/);

        if (match) {
            const pageNum = parseInt(match[1], 10);
            if (!isNaN(pageNum) && pageNum > 0) {
                return pageNum;
            }
        }

        return 1;
    };

    const sendToBackend = (data) => {
        return new Promise((resolve, reject) => {
            GM_xmlhttpRequest({
                method: 'POST',
                url: CONFIG.backendUrl,
                headers: {
                    'Content-Type': 'application/json'
                },
                data: JSON.stringify(data),
                onload: function(response) {
                    if (response.status >= 200 && response.status < 300) {
                        resolve(response);
                    } else {
                        reject(new Error(`HTTP ${response.status}: ${response.statusText}`));
                    }
                },
                onerror: function(error) {
                    reject(error);
                },
                timeout: 10000
            });
        });
    };

    const sendAllData = async () => {
        updateStatus('整理数据并发送到后端...');

        const finalImages = {};
        let totalCount = 0;

        const sortedPages = Object.keys(allImages).map(Number).sort((a, b) => a - b);

        for (const page of sortedPages) {
            if (allImages[page]) {
                for (const imgUrl of allImages[page]) {
                    const key = String(imgIndex).padStart(4, '0');
                    finalImages[key] = imgUrl;
                    imgIndex++;
                    totalCount++;
                }
            }
        }

        const data = {
            title: document.title || '无标题',
            source: source,
            url: buildPageUrl(1),
            totalPages: sortedPages.length,
            totalImages: totalCount,
            imgs: finalImages
        };

        console.log('准备发送的数据:', data);

        try {
            await sendToBackend(data);
            updateStatus(`✅ 发送成功！<br>共 ${sortedPages.length} 页<br>${totalCount} 张图片`);
            return true;
        } catch (error) {
            updateStatus(`❌ 发送失败: ${error.message}`);
            return false;
        }
    };

    const beginPageProcessing = async () => {
        const isCrawlSession = sessionStorage.getItem('hd4k_crawling') === 'true';

        if (!isCrawlSession) {
            console.log('不在爬取会话中，停止处理');
            return;
        }

        const currentUrl = window.location.href;

        if (crawledUrls.includes(currentUrl)) {
            updateStatus('检测到重复URL，停止爬取');
            await finishCrawling();
            return;
        }

        crawledUrls.push(currentUrl);
        sessionStorage.setItem('hd4k_crawled_urls', JSON.stringify(crawledUrls));

        const urlPageNum = getCurrentPageNumber();

        if (currentPage !== urlPageNum) {
            currentPage = urlPageNum;
        }

        updateStatus(`处理第 ${currentPage} 页...`);
        const imageUrls = getCurrentPageImages();

        if (imageUrls.length === 0) {
            updateStatus(`第 ${currentPage} 页: 未找到图片`);
            setTimeout(async () => {
                await finishCrawling();
            }, CONFIG.pageDelay);
            return;
        }

        console.log(`第 ${currentPage} 页找到 ${imageUrls.length} 张图片`);

        allImages[currentPage] = imageUrls;
        sessionStorage.setItem('hd4k_all_images', JSON.stringify(allImages));
        updateStatus(`第 ${currentPage} 页: 找到 ${imageUrls.length} 张图片`);

        setTimeout(async () => {
            const nextPage = currentPage + 1;

            if (nextPage > CONFIG.maxPages) {
                updateStatus(`已达到最大页数 ${CONFIG.maxPages}`);
                await finishCrawling();
                return;
            }

            const nextUrl = buildPageUrl(nextPage);

            if (crawledUrls.includes(nextUrl)) {
                updateStatus('下一页URL已爬取过，停止爬取');
                await finishCrawling();
                return;
            }

            updateStatus(`准备跳转到第 ${nextPage} 页`);
            sessionStorage.setItem('hd4k_current_page', nextPage.toString());

            setTimeout(() => {
                window.location.href = nextUrl;
            }, CONFIG.pageDelay);

        }, CONFIG.pageDelay);
    };

    const startCrawling = async () => {
        if (isCrawling) {
            alert('正在爬取中，请稍候...');
            return;
        }

        // 检查容器是否存在
        const container = document.querySelector('#basicExample');
        if (!container) {
            alert('未找到图片容器 #basicExample，请确认页面结构！\n\n可能原因：\n1. 页面未完全加载\n2. 图片在滚动后才加载\n3. 网站结构已变化');
            return;
        }

        const button = document.getElementById('hd4k-btn');
        button.textContent = '爬取中...';
        button.style.backgroundColor = '#ff9800';
        button.disabled = true;

        isCrawling = true;
        allImages = {};
        crawledUrls = [];
        currentPage = 1;
        imgIndex = 1;

        sessionStorage.removeItem('hd4k_all_images');
        sessionStorage.removeItem('hd4k_crawled_urls');

        updateStatus('开始自动翻页爬取...');

        const firstPageUrl = buildPageUrl(1);
        const currentUrl = window.location.href;

        sessionStorage.setItem('hd4k_crawling', 'true');
        sessionStorage.setItem('hd4k_current_page', '1');

        if (currentUrl !== firstPageUrl) {
            updateStatus(`跳转到第一页`);
            window.location.href = firstPageUrl;
            return;
        }

        beginPageProcessing();
    };

    const finishCrawling = async () => {
        sessionStorage.removeItem('hd4k_crawling');
        sessionStorage.removeItem('hd4k_current_page');

        if (Object.keys(allImages).length > 0) {
            await sendAllData();
        } else {
            updateStatus('未找到任何图片数据');
        }

        const button = document.getElementById('hd4k-btn');
        button.textContent = '开始爬取';
        button.style.backgroundColor = '#2c80ff';
        button.disabled = false;
        isCrawling = false;

        setTimeout(() => {
            const statusDiv = document.getElementById('hd4k-status');
            if (statusDiv) {
                statusDiv.style.display = 'none';
            }
        }, 5000);
    };

    const onPageLoad = () => {
        const isCrawlSession = sessionStorage.getItem('hd4k_crawling') === 'true';

        if (isCrawlSession) {
            isCrawling = true;
            currentPage = getCurrentPageNumber();

            const savedImages = sessionStorage.getItem('hd4k_all_images');
            const savedUrls = sessionStorage.getItem('hd4k_crawled_urls');

            if (savedImages) {
                allImages = JSON.parse(savedImages);
            }

            if (savedUrls) {
                crawledUrls = JSON.parse(savedUrls);
            }

            setTimeout(() => {
                beginPageProcessing();
            }, 1500);
        }
    };

    const init = () => {
        if (!document.getElementById('hd4k-btn')) {
            const button = createButton();
            const statusDiv = createStatusDisplay();

            document.body.appendChild(button);
            document.body.appendChild(statusDiv);

            updateStatus('HD4K下载器已加载<br>点击按钮开始自动翻页爬取');
            setTimeout(() => {
                const statusDiv = document.getElementById('hd4k-status');
                if (statusDiv) {
                    statusDiv.style.display = 'none';
                }
            }, 3000);

            onPageLoad();
        }
    };

    if (document.readyState === 'loading') {
        document.addEventListener('DOMContentLoaded', init);
    } else {
        const isCrawlSession = sessionStorage.getItem('hd4k_crawling') === 'true';
        if (isCrawlSession) {
            isCrawling = true;
            const button = createButton();
            button.textContent = '爬取中...';
            button.style.backgroundColor = '#ff9800';
            button.disabled = true;
            document.body.appendChild(button);

            const statusDiv = createStatusDisplay();
            document.body.appendChild(statusDiv);
            updateStatus('检测到未完成的爬取任务，继续执行...');

            setTimeout(onPageLoad, 1500);
        } else {
            init();
        }
    }

})();