You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
374 lines
13 KiB
374 lines
13 KiB
// ==UserScript==
|
|
// @name eh
|
|
// @namespace http://tampermonkey.net/
|
|
// @version 1.5
|
|
// @description 采集页面数据并发送到后端
|
|
// @author Jack
|
|
// @match https://e-hentai.org/*
|
|
// @grant GM_xmlhttpRequest
|
|
// @grant GM_notification
|
|
// ==/UserScript==
|
|
|
|
(function() {
|
|
'use strict';
|
|
|
|
// 全局配置
|
|
const BACKEND_IP = '127.0.0.1';
|
|
const BACKEND_PORT = '55830';
|
|
const BUTTON_LOCATION_SELECTOR = '#gd5 > p:nth-child(5)';
|
|
const DATA_LIST_SELECTOR = '#gdt a';
|
|
const ALL_IMG_DATA = {};
|
|
const source = 'eh';
|
|
|
|
// 位置配置 - 只需调整这个数字即可同时调整按钮和进度条位置
|
|
const TOP_POSITION = 32; // 百分比位置，按钮在32%，进度条在36%
|
|
|
|
// 并发配置
|
|
const CONCURRENT_LIMIT = 5; // 并发请求数量限制
|
|
const REQUEST_DELAY = 50; // 请求间延迟(毫秒)
|
|
|
|
// 创建按钮
|
|
const button = document.createElement('button');
|
|
button.id = 'data-sender-button';
|
|
button.textContent = "send data";
|
|
button.style.position = "fixed";
|
|
button.style.top = `${TOP_POSITION}%`;
|
|
button.style.right = "1%";
|
|
button.style.transform = "translateY(-50%)";
|
|
button.style.padding = "3px 8px";
|
|
button.style.fontSize = "10px";
|
|
button.style.backgroundColor = "#007baf";
|
|
button.style.color = "#fff";
|
|
button.style.border = "none";
|
|
button.style.borderRadius = "5px";
|
|
button.style.cursor = "pointer";
|
|
button.style.zIndex = "10000";
|
|
|
|
// 创建进度显示 - 位置在按钮下方4%
|
|
const progressDiv = document.createElement('div');
|
|
progressDiv.id = 'progress-display';
|
|
progressDiv.style.position = "fixed";
|
|
progressDiv.style.top = `${TOP_POSITION + 4}%`;
|
|
progressDiv.style.right = "1%";
|
|
progressDiv.style.padding = "5px 10px";
|
|
progressDiv.style.fontSize = "12px";
|
|
progressDiv.style.backgroundColor = "rgba(0,0,0,0.7)";
|
|
progressDiv.style.color = "#fff";
|
|
progressDiv.style.borderRadius = "5px";
|
|
progressDiv.style.zIndex = "9999";
|
|
progressDiv.style.display = "none";
|
|
document.body.appendChild(progressDiv);
|
|
|
|
const targetElement = document.querySelector(BUTTON_LOCATION_SELECTOR);
|
|
if (targetElement) {
|
|
targetElement.appendChild(button);
|
|
} else {
|
|
document.body.appendChild(button);
|
|
}
|
|
|
|
function formatNumber(num) {
|
|
return num.toString().padStart(4, '0');
|
|
}
|
|
|
|
function getBaseUrl(url) {
|
|
return url.replace(/([?&])p=\d+(&|$)/, (match, p1, p2) => {
|
|
return p2 === '&' ? p1 : '';
|
|
}).replace(/[?&]$/, '');
|
|
}
|
|
|
|
function buildPageUrl(baseUrl, page) {
|
|
if (page === 0) {
|
|
return baseUrl.includes('?') ? baseUrl : baseUrl;
|
|
} else {
|
|
const separator = baseUrl.includes('?') ? '&' : '?';
|
|
return baseUrl + separator + `p=${page}`;
|
|
}
|
|
}
|
|
|
|
function extractThumbnailLinks(htmlContent) {
|
|
const links = [];
|
|
const tempDiv = document.createElement('div');
|
|
tempDiv.innerHTML = htmlContent;
|
|
|
|
if (DATA_LIST_SELECTOR) {
|
|
const linkElements = tempDiv.querySelectorAll(DATA_LIST_SELECTOR);
|
|
linkElements.forEach(link => {
|
|
const hrefAttr = link.getAttribute('href');
|
|
if (hrefAttr) {
|
|
links.push(hrefAttr);
|
|
}
|
|
});
|
|
}
|
|
return links;
|
|
}
|
|
|
|
function extractActualImageUrl(htmlContent) {
|
|
const regex = /<img[^>]*id="img"[^>]*src="([^"]*)"[^>]*>/i;
|
|
const match = htmlContent.match(regex);
|
|
return match ? match[1] : null;
|
|
}
|
|
|
|
async function fetchImagePage(url) {
|
|
return new Promise((resolve, reject) => {
|
|
GM_xmlhttpRequest({
|
|
method: "GET",
|
|
url: url,
|
|
headers: {
|
|
"Referer": window.location.href,
|
|
"Cookie": document.cookie
|
|
},
|
|
onload: function(response) {
|
|
resolve({url, html: response.responseText});
|
|
},
|
|
onerror: function(error) {
|
|
reject({url, error});
|
|
}
|
|
});
|
|
});
|
|
}
|
|
|
|
// 并发处理函数
|
|
async function processThumbnailLinksConcurrently(thumbnailLinks, pageNum) {
|
|
const results = [];
|
|
const totalLinks = thumbnailLinks.length;
|
|
|
|
// 更新进度显示 - 页数显示为 pageNum + 1（从1开始）
|
|
const displayPageNum = pageNum + 1;
|
|
updateProgress(`第${displayPageNum}页: 0/${totalLinks}`, 0);
|
|
|
|
// 将链接分组，实现并发控制
|
|
for (let i = 0; i < thumbnailLinks.length; i += CONCURRENT_LIMIT) {
|
|
const chunk = thumbnailLinks.slice(i, i + CONCURRENT_LIMIT);
|
|
|
|
// 并发请求当前组
|
|
const promises = chunk.map(link => fetchImagePage(link));
|
|
|
|
try {
|
|
const chunkResults = await Promise.all(promises);
|
|
results.push(...chunkResults);
|
|
|
|
// 更新进度
|
|
const processed = Math.min(i + CONCURRENT_LIMIT, totalLinks);
|
|
updateProgress(`第${displayPageNum}页: ${processed}/${totalLinks}`, (processed / totalLinks) * 100);
|
|
|
|
// 组间延迟，避免请求过快
|
|
if (i + CONCURRENT_LIMIT < thumbnailLinks.length) {
|
|
await new Promise(resolve => setTimeout(resolve, REQUEST_DELAY));
|
|
}
|
|
} catch (error) {
|
|
console.error('并发请求组失败:', error);
|
|
}
|
|
}
|
|
|
|
// 处理结果，提取真实图片URL
|
|
const imageUrls = [];
|
|
for (const result of results) {
|
|
if (result.html) {
|
|
const actualImageUrl = extractActualImageUrl(result.html);
|
|
if (actualImageUrl) {
|
|
imageUrls.push(actualImageUrl);
|
|
}
|
|
}
|
|
}
|
|
|
|
return imageUrls;
|
|
}
|
|
|
|
function updateProgress(text, percentage) {
|
|
progressDiv.textContent = text;
|
|
progressDiv.style.display = "block";
|
|
|
|
// 可以添加进度条样式
|
|
progressDiv.style.background = `linear-gradient(90deg, #007baf ${percentage}%, rgba(0,0,0,0.7) ${percentage}%)`;
|
|
}
|
|
|
|
async function processPage(page) {
|
|
const baseUrl = getBaseUrl(window.location.href);
|
|
const pageUrl = buildPageUrl(baseUrl, page);
|
|
|
|
try {
|
|
let htmlContent;
|
|
|
|
if (page === 0) {
|
|
htmlContent = document.documentElement.innerHTML;
|
|
} else {
|
|
htmlContent = await new Promise((resolve, reject) => {
|
|
GM_xmlhttpRequest({
|
|
method: "GET",
|
|
url: pageUrl,
|
|
headers: {
|
|
"Referer": window.location.href,
|
|
"Cookie": document.cookie
|
|
},
|
|
onload: function(response) {
|
|
resolve(response.responseText);
|
|
},
|
|
onerror: function(error) {
|
|
reject(error);
|
|
}
|
|
});
|
|
});
|
|
}
|
|
|
|
const thumbnailLinks = extractThumbnailLinks(htmlContent);
|
|
|
|
if (thumbnailLinks.length === 0) {
|
|
console.log(`第${page + 1}页没有缩略图链接，可能是最后一页`);
|
|
return false;
|
|
}
|
|
|
|
console.log(`第${page + 1}页找到${thumbnailLinks.length}个缩略图链接，开始并发获取真实图片URL...`);
|
|
|
|
// 使用并发处理
|
|
const actualImageUrls = await processThumbnailLinksConcurrently(thumbnailLinks, page);
|
|
|
|
let hasNewImage = false;
|
|
actualImageUrls.forEach(url => {
|
|
const isDuplicate = Object.values(ALL_IMG_DATA).includes(url);
|
|
if (!isDuplicate) {
|
|
ALL_IMG_DATA[formatNumber(Object.keys(ALL_IMG_DATA).length + 1)] = url;
|
|
hasNewImage = true;
|
|
}
|
|
});
|
|
|
|
console.log(`第${page + 1}页采集完成，获取到${actualImageUrls.length}个真实图片链接`);
|
|
|
|
return hasNewImage;
|
|
|
|
} catch (error) {
|
|
console.error(`第${page + 1}页采集失败:`, error);
|
|
return false;
|
|
}
|
|
}
|
|
|
|
function sendDataToBackend(data) {
|
|
return new Promise((resolve, reject) => {
|
|
GM_xmlhttpRequest({
|
|
method: "POST",
|
|
url: `http://${BACKEND_IP}:${BACKEND_PORT}/api/save_json`,
|
|
headers: {
|
|
"Content-Type": "application/json",
|
|
},
|
|
data: JSON.stringify(data),
|
|
onload: function(response) {
|
|
if (response.status === 200) {
|
|
try {
|
|
const result = JSON.parse(response.responseText);
|
|
if (result.success) {
|
|
resolve(result);
|
|
} else {
|
|
reject(new Error(result.message || '后端保存失败'));
|
|
}
|
|
} catch (e) {
|
|
reject(new Error('解析响应失败: ' + e.message));
|
|
}
|
|
} else {
|
|
reject(new Error(`后端返回错误: ${response.status} - ${response.responseText}`));
|
|
}
|
|
},
|
|
onerror: function(error) {
|
|
reject(error);
|
|
}
|
|
});
|
|
});
|
|
}
|
|
|
|
button.addEventListener('click', async function() {
|
|
const currentUrl = window.location.href;
|
|
const pageTitle = document.title;
|
|
|
|
Object.keys(ALL_IMG_DATA).forEach(key => delete ALL_IMG_DATA[key]);
|
|
|
|
button.textContent = "采集中...";
|
|
button.disabled = true;
|
|
progressDiv.style.display = "block";
|
|
|
|
try {
|
|
let shouldContinue = true;
|
|
let page = 0;
|
|
let totalPagesProcessed = 0;
|
|
|
|
while (shouldContinue && page <= 100) {
|
|
// 显示页数为 page + 1（从1开始）
|
|
updateProgress(`正在处理第${page + 1}页...`, (page / 100) * 50);
|
|
|
|
const hasNewImages = await processPage(page);
|
|
totalPagesProcessed++;
|
|
|
|
if (!hasNewImages && page > 0) {
|
|
console.log(`第${page + 1}页没有新图片，停止采集`);
|
|
shouldContinue = false;
|
|
}
|
|
|
|
if (Object.keys(ALL_IMG_DATA).length >= 2200) {
|
|
console.log('图片数量达到上限2200，停止采集');
|
|
shouldContinue = false;
|
|
}
|
|
|
|
page++;
|
|
|
|
// 页面间延迟，避免请求过快
|
|
if (shouldContinue) {
|
|
await new Promise(resolve => setTimeout(resolve, 200));
|
|
}
|
|
}
|
|
|
|
updateProgress(`处理完成，准备发送数据...`, 80);
|
|
|
|
const data = {
|
|
url: currentUrl,
|
|
title: pageTitle,
|
|
source: source,
|
|
imgs: ALL_IMG_DATA,
|
|
totalImages: Object.keys(ALL_IMG_DATA).length
|
|
};
|
|
|
|
console.log('采集完成的所有数据:', data);
|
|
|
|
// 发送数据到后端
|
|
const response = await sendDataToBackend(data);
|
|
|
|
updateProgress(`数据发送成功！页面将在1秒后关闭...`, 100);
|
|
|
|
// 显示成功通知
|
|
if (typeof GM_notification !== 'undefined') {
|
|
GM_notification({
|
|
title: '数据采集完成',
|
|
text: `已采集 ${Object.keys(ALL_IMG_DATA).length} 张图片，页面即将关闭`,
|
|
timeout: 1000
|
|
});
|
|
}
|
|
|
|
// 1秒后自动关闭页面，无需用户确认
|
|
setTimeout(() => {
|
|
console.log(`数据保存成功，关闭页面。采集统计：
|
|
- 标题: ${pageTitle}
|
|
- 总图片数: ${Object.keys(ALL_IMG_DATA).length}
|
|
- 处理页数: ${totalPagesProcessed}
|
|
- 保存路径: ${response.folder || '未知'}`);
|
|
window.close();
|
|
}, 1000);
|
|
|
|
} catch (error) {
|
|
console.error('采集失败:', error);
|
|
updateProgress(`采集失败: ${error.message}`, 0);
|
|
|
|
// 失败时显示错误信息，但也不弹框，只在控制台显示
|
|
console.error('数据采集失败！错误:', error.message);
|
|
|
|
// 在进度条上显示错误信息
|
|
progressDiv.textContent = `采集失败: ${error.message}`;
|
|
progressDiv.style.backgroundColor = "rgba(255,0,0,0.7)";
|
|
|
|
} finally {
|
|
button.textContent = "send data";
|
|
button.disabled = false;
|
|
|
|
// 如果发生错误，10秒后隐藏进度显示
|
|
setTimeout(() => {
|
|
progressDiv.style.display = "none";
|
|
progressDiv.style.backgroundColor = "rgba(0,0,0,0.7)";
|
|
}, 10000);
|
|
}
|
|
});
|
|
})(); |