douyin-user-data-download

下载抖音用户主页数据!

当前为 2024-06-21 提交的版本,查看 最新版本

  1. // ==UserScript==
  2. // @name douyin-user-data-download
  3. // @namespace http://tampermonkey.net/
  4. // @version 0.5.3
  5. // @description 下载抖音用户主页数据!
  6. // @author xxmdmst
  7. // @match https://www.douyin.com/*
  8. // @icon https://xxmdmst.oss-cn-beijing.aliyuncs.com/imgs/favicon.ico
  9. // @grant GM_registerMenuCommand
  10. // @grant GM_setValue
  11. // @grant GM_getValue
  12. // @require https://cdnjs.cloudflare.com/ajax/libs/jszip/3.6.0/jszip.min.js
  13. // @license MIT
  14. // ==/UserScript==
  15.  
  16. (function () {
  17. let localDownload;
  18. let localDownloadUrl = GM_getValue("localDownloadUrl", 'http://localhost:8080/data');
  19. const startPipeline = (start) => {
  20. if (confirm(start ? "是否开启本地下载通道?\n开启后会向本地服务发送数据,服务地址:\n" + localDownloadUrl : "是否关闭本地下载通道?")) {
  21. GM_setValue("localDownload", start);
  22. window.location.reload();
  23. }
  24. }
  25. localDownload = GM_getValue("localDownload", false);
  26. if (localDownload) {
  27. GM_registerMenuCommand("✅关闭上报本地通道", () => {
  28. startPipeline(false);
  29. })
  30. } else {
  31. GM_registerMenuCommand("⛔️开启上报本地通道", () => {
  32. startPipeline(true);
  33. })
  34. }
  35.  
  36. GM_registerMenuCommand("♐设置本地上报地址", () => {
  37. localDownloadUrl = GM_getValue("localDownloadUrl", 'http://localhost:8080/data');
  38. let newlocalDownloadUrl = prompt("请输入新的上报地址:", localDownloadUrl);
  39. if (newlocalDownloadUrl === null) return;
  40. newlocalDownloadUrl = newlocalDownloadUrl.trim();
  41. if (!newlocalDownloadUrl) {
  42. newlocalDownloadUrl = "http://localhost:8080/data";
  43. toast("设置了空白地址,已经恢复默认地址为:" + newlocalDownloadUrl);
  44. localDownloadUrl = newlocalDownloadUrl;
  45. } else if (localDownloadUrl !== newlocalDownloadUrl) {
  46. GM_setValue("localDownloadUrl", newlocalDownloadUrl);
  47. toast("当前上报地址已经修改为:" + newlocalDownloadUrl);
  48. }
  49. GM_setValue("localDownloadUrl", newlocalDownloadUrl);
  50. localDownloadUrl = newlocalDownloadUrl;
  51. });
  52. GM_registerMenuCommand("🔄清空信息内容", () => msg_pre.textContent = "")
  53. let max_author_num = GM_getValue("max_author_num", 1000);
  54. GM_registerMenuCommand("👤设置最大缓存作者数", () => {
  55. let new_max_author_num = prompt("设置最大缓存作者数:", max_author_num);
  56. if (new_max_author_num === null) return;
  57. if (!/^\d+$/.test(new_max_author_num)) {
  58. toast("请输入正整数!");
  59. return;
  60. }
  61. max_author_num = parseInt(new_max_author_num);
  62. GM_setValue("max_author_num", max_author_num);
  63. toast("当前最大缓存作者数已经修改为:" + max_author_num);
  64. })
  65. let table;
  66.  
  67. function initGbkTable() {
  68. // https://en.wikipedia.org/wiki/GBK_(character_encoding)#Encoding
  69. const ranges = [
  70. [0xA1, 0xA9, 0xA1, 0xFE],
  71. [0xB0, 0xF7, 0xA1, 0xFE],
  72. [0x81, 0xA0, 0x40, 0xFE],
  73. [0xAA, 0xFE, 0x40, 0xA0],
  74. [0xA8, 0xA9, 0x40, 0xA0],
  75. [0xAA, 0xAF, 0xA1, 0xFE],
  76. [0xF8, 0xFE, 0xA1, 0xFE],
  77. [0xA1, 0xA7, 0x40, 0xA0],
  78. ];
  79. const codes = new Uint16Array(23940);
  80. let i = 0;
  81.  
  82. for (const [b1Begin, b1End, b2Begin, b2End] of ranges) {
  83. for (let b2 = b2Begin; b2 <= b2End; b2++) {
  84. if (b2 !== 0x7F) {
  85. for (let b1 = b1Begin; b1 <= b1End; b1++) {
  86. codes[i++] = b2 << 8 | b1
  87. }
  88. }
  89. }
  90. }
  91. table = new Uint16Array(65536);
  92. table.fill(0xFFFF);
  93. const str = new TextDecoder('gbk').decode(codes);
  94. for (let i = 0; i < str.length; i++) {
  95. table[str.charCodeAt(i)] = codes[i]
  96. }
  97. }
  98.  
  99. function str2gbk(str, opt = {}) {
  100. if (!table) {
  101. initGbkTable()
  102. }
  103. const NodeJsBufAlloc = typeof Buffer === 'function' && Buffer.allocUnsafe;
  104. const defaultOnAlloc = NodeJsBufAlloc
  105. ? (len) => NodeJsBufAlloc(len)
  106. : (len) => new Uint8Array(len);
  107. const defaultOnError = () => 63;
  108. const onAlloc = opt.onAlloc || defaultOnAlloc;
  109. const onError = opt.onError || defaultOnError;
  110.  
  111. const buf = onAlloc(str.length * 2);
  112. let n = 0;
  113.  
  114. for (let i = 0; i < str.length; i++) {
  115. const code = str.charCodeAt(i);
  116. if (code < 0x80) {
  117. buf[n++] = code;
  118. continue
  119. }
  120. const gbk = table[code];
  121.  
  122. if (gbk !== 0xFFFF) {
  123. buf[n++] = gbk;
  124. buf[n++] = gbk >> 8
  125. } else if (code === 8364) {
  126. buf[n++] = 0x80
  127. } else {
  128. const ret = onError(i, str);
  129. if (ret === -1) {
  130. break
  131. }
  132. if (ret > 0xFF) {
  133. buf[n++] = ret;
  134. buf[n++] = ret >> 8
  135. } else {
  136. buf[n++] = ret
  137. }
  138. }
  139. }
  140. return buf.subarray(0, n)
  141. }
  142.  
  143. const toast = (msg, duration) => {
  144. duration = isNaN(duration) ? 3000 : duration;
  145. let toastDom = document.createElement('pre');
  146. toastDom.textContent = msg;
  147. toastDom.style.cssText = 'padding:2px 15px;min-height: 36px;line-height: 36px;text-align: center;transform: translate(-50%);border-radius: 4px;color: rgb(255, 255, 255);position: fixed;top: 50%;left: 50%;z-index: 9999999;background: rgb(0, 0, 0);font-size: 16px;'
  148. document.body.appendChild(toastDom);
  149. setTimeout(function () {
  150. const d = 0.5;
  151. toastDom.style.transition = `transform ${d}s ease-in, opacity ${d}s ease-in`;
  152. toastDom.style.opacity = '0';
  153. setTimeout(function () {
  154. document.body.removeChild(toastDom)
  155. }, d * 1000);
  156. }, duration);
  157. }
  158.  
  159. function formatSeconds(seconds) {
  160. const timeUnits = ['小时', '分', '秒'];
  161. const timeValues = [
  162. Math.floor(seconds / 3600),
  163. Math.floor((seconds % 3600) / 60),
  164. seconds % 60
  165. ];
  166. return timeValues.map((value, index) => value > 0 ? value + timeUnits[index] : '').join('');
  167. }
  168.  
  169. const timeFormat = (timestamp = null, fmt = 'yyyy-mm-dd') => {
  170. // 其他更多是格式化有如下:
  171. // yyyy:mm:dd|yyyy:mm|yyyy年mm月dd日|yyyy年mm月dd日 hh时MM分等,可自定义组合
  172. timestamp = parseInt(timestamp);
  173. // 如果为null,则格式化当前时间
  174. if (!timestamp) timestamp = Number(new Date());
  175. // 判断用户输入的时间戳是秒还是毫秒,一般前端js获取的时间戳是毫秒(13位),后端传过来的为秒(10位)
  176. if (timestamp.toString().length === 10) timestamp *= 1000;
  177. let date = new Date(timestamp);
  178. let ret;
  179. let opt = {
  180. "y{4,}": date.getFullYear().toString(), // 年
  181. "y+": date.getFullYear().toString().slice(2,), // 年
  182. "m+": (date.getMonth() + 1).toString(), // 月
  183. "d+": date.getDate().toString(), // 日
  184. "h+": date.getHours().toString(), // 时
  185. "M+": date.getMinutes().toString(), // 分
  186. "s+": date.getSeconds().toString() // 秒
  187. // 有其他格式化字符需求可以继续添加,必须转化成字符串
  188. };
  189. for (let k in opt) {
  190. ret = new RegExp("(" + k + ")").exec(fmt);
  191. if (ret) {
  192. fmt = fmt.replace(ret[1], (ret[1].length === 1) ? (opt[k]) : (opt[k].padStart(ret[1].length, "0")))
  193. }
  194. }
  195. return fmt
  196. };
  197. window.all_aweme_map = new Map();
  198. window.user_map = new Map();
  199. const user_local_data = localStorage.getItem('user_local_data');
  200. if (user_local_data) {
  201. JSON.parse(user_local_data).forEach((userInfo) => {
  202. user_map.set(userInfo.uid, userInfo);
  203. });
  204. }
  205. let current_user_id = null;
  206. const user_key = {
  207. "nickname": "昵称",
  208. "following_count": "关注",
  209. "mplatform_followers_count": "粉丝",
  210. "total_favorited": "获赞",
  211. "unique_id": "抖音号",
  212. "ip_location": "IP属地",
  213. "gender": "性别",
  214. "city": "位置",
  215. "signature": "签名",
  216. "aweme_count": "作品数",
  217. }
  218.  
  219. function copyText(text, node) {
  220. let oldText = node.textContent;
  221. navigator.clipboard.writeText(text).then(r => {
  222. node.textContent = "复制成功";
  223. toast("复制成功\n" + text.slice(0, 20) + (text.length > 20 ? "..." : ""), 2000);
  224. }).catch((e) => {
  225. node.textContent = "复制失败";
  226. toast("复制失败", 2000);
  227. })
  228. setTimeout(() => node.textContent = oldText, 2000);
  229. }
  230.  
  231. function copyUserData(node) {
  232. if (!current_user_id) {
  233. toast("还没有捕获到用户数据!");
  234. return;
  235. }
  236. let text = [];
  237. let userInfo = user_map.get(current_user_id);
  238. for (let key in user_key) {
  239. let value = (userInfo[key] || "").toString().trim()
  240. if (value) text.push(user_key[key] + ":" + value);
  241. }
  242. copyText(text.join("\n"), node);
  243. }
  244.  
  245. function createVideoButton(text, top, func) {
  246. const button = document.createElement("button");
  247. button.textContent = text;
  248. button.style.position = "absolute";
  249. button.style.right = "0px";
  250. button.style.top = top;
  251. button.style.opacity = "0.5";
  252. if (func) {
  253. button.addEventListener("click", (event) => {
  254. event.preventDefault();
  255. event.stopPropagation();
  256. func();
  257. });
  258. }
  259. return button;
  260. }
  261.  
  262. function createDownloadLink(blob, filename, ext, prefix = "") {
  263. if (filename === null) {
  264. filename = current_user_id ? user_map.get(current_user_id).nickname : document.title;
  265. }
  266. const url = URL.createObjectURL(blob);
  267. const link = document.createElement('a');
  268. link.href = url;
  269. link.download = prefix + filename.replace(/[\/:*?"<>|\s]/g, "").slice(0, 40) + "." + ext;
  270. link.click();
  271. URL.revokeObjectURL(url);
  272. }
  273.  
  274. function txt2file(txt, filename, ext) {
  275. createDownloadLink(new Blob([txt], {type: 'text/plain'}), filename, ext);
  276. }
  277.  
  278. function getAwemeName(aweme) {
  279. let name = aweme.item_title ? aweme.item_title : aweme.caption;
  280. if (!name) name = aweme.desc ? aweme.desc : aweme.awemeId;
  281. return (aweme.date ? `【${aweme.date.slice(0, 10)}】` : "") + name.replace(/[\/:*?"<>|\s]+/g, "").slice(0, 27).replace(/\.\d+$/g, "");
  282. }
  283.  
  284. const downloadUrl = (url, node, filename = window.title, ext = "mp4") => {
  285. // toast("准备就绪,等待视频下载完毕后弹出下载界面!");
  286. let xhr = new XMLHttpRequest();
  287. xhr.open('GET', url.replace("http://", "https://"), true);
  288. xhr.responseType = 'blob';
  289. let textContent = node.textContent;
  290. xhr.onload = (e) => {
  291. createDownloadLink(xhr.response, filename, ext);
  292. setTimeout(() => node.textContent = textContent, 2000);
  293. };
  294. xhr.onprogress = (event) => {
  295. if (event.lengthComputable) {
  296. node.textContent = "下载" + (event.loaded * 100 / event.total).toFixed(1) + '%';
  297. }
  298. };
  299. xhr.send();
  300. };
  301. const downloadVideo = (aweme, node) => {
  302. toast("准备就绪,等待视频下载完毕后弹出下载界面!");
  303. let xhr = new XMLHttpRequest();
  304. let url = aweme.url.replace("http://", "https://");
  305. let filename = aweme ? getAwemeName(aweme) : window.title;
  306. let ext = aweme && aweme.images ? "mp3" : "mp4";
  307. downloadUrl(url, node, filename, ext);
  308. };
  309. const downloadImage = (aweme, downloadImageButton) => {
  310. const zip = new JSZip();
  311. let textContent = downloadImageButton.textContent;
  312. downloadImageButton.textContent = "图片下载并打包中...";
  313. const promises = aweme.images.map((link, index) => {
  314. return fetch(link)
  315. .then((response) => response.arrayBuffer())
  316. .then((buffer) => {
  317. downloadImageButton.textContent = `图片已下载【${index + 1}/${aweme.images.length}】`;
  318. zip.file(`image_${index + 1}.jpg`, buffer);
  319. });
  320. });
  321. Promise.all(promises)
  322. .then(() => {
  323. return zip.generateAsync({type: "blob"});
  324. })
  325. .then((content) => {
  326. createDownloadLink(content, getAwemeName(aweme), "zip", "【图文】");
  327. setTimeout(() => downloadImageButton.textContent = textContent, 2000);
  328. });
  329. };
  330.  
  331. function createButtonGroup(aNode) {
  332. if (aNode.dataset.vid) return;
  333. let match = aNode.href.match(/(?:video|note)\/(\d+)/);
  334. if (!match) return;
  335. let videoId = match[1];
  336. let aweme = all_aweme_map.get(videoId);
  337. let copyDescButton = createVideoButton("复制描述", "0px");
  338. copyDescButton.addEventListener("click", (event) => {
  339. event.preventDefault();
  340. event.stopPropagation();
  341. copyText(aweme.desc, copyDescButton);
  342. })
  343. aNode.appendChild(copyDescButton);
  344. aNode.appendChild(createVideoButton("打开视频源", "20px", () => window.open(aweme.url)));
  345.  
  346. let downloadVideoButton = createVideoButton("下载视频", "40px");
  347. downloadVideoButton.addEventListener("click", () => downloadVideo(aweme, downloadVideoButton));
  348. aNode.appendChild(downloadVideoButton);
  349.  
  350. if (aweme.images) {
  351. let downloadImageButton = createVideoButton("图片打包下载", "60px");
  352. downloadImageButton.addEventListener("click", () => downloadImage(aweme, downloadImageButton));
  353. aNode.appendChild(downloadImageButton);
  354. }
  355. aNode.dataset.vid = videoId;
  356. }
  357.  
  358. function flush() {
  359. data_button.p2.textContent = `${all_aweme_map.size}`;
  360. user_button.p2.textContent = `${user_map.size}`;
  361. let img_num = Array.from(all_aweme_map.values()).filter(a => a.images).length;
  362. img_button.p2.textContent = `${img_num}`;
  363. msg_pre.textContent = `已加载${all_aweme_map.size}个作品,${img_num}个图文\n本地已缓存(${user_map.size}/${max_author_num})个作者\n激活上方头像可展开下载按钮`;
  364. }
  365.  
  366. const formatDouyinAwemeData = item => Object.assign(
  367. {
  368. "awemeId": item.aweme_id,
  369. "item_title": item.item_title,
  370. "caption": item.caption,
  371. "desc": item.desc,
  372. "tag": item.text_extra ? item.text_extra.map(tag => tag.hashtag_name).filter(tag => tag).join("#") : "",
  373. "video_tag": item.video_tag ? item.video_tag.map(tag => tag.tag_name).filter(tag => tag).join("->") : "",
  374. "date": timeFormat(item.create_time, "yyyy-mm-dd hh:MM:ss"),
  375. "create_time": item.create_time,
  376. },
  377. item.statistics ? {
  378. "diggCount": item.statistics.digg_count,
  379. "commentCount": item.statistics.comment_count,
  380. "collectCount": item.statistics.collect_count,
  381. "shareCount": item.statistics.share_count
  382. } : {},
  383. item.video ? {
  384. "duration": formatSeconds(Math.round(item.video.duration / 1000)),
  385. "url": item.video.play_addr.url_list[0],
  386. "cover": item.video.cover.url_list[0],
  387. "images": item.images ? item.images.map(row => row.url_list.pop()) : null,
  388. } : {},
  389. item.author ? {
  390. "uid": item.author.uid,
  391. "nickname": item.author.nickname
  392. } : {}
  393. );
  394.  
  395.  
  396. function formatAwemeData(json_data) {
  397. return json_data.aweme_list.map(formatDouyinAwemeData);
  398. }
  399.  
  400. function formatUserData(userInfo) {
  401. for (let key in userInfo) {
  402. if (!userInfo[key]) userInfo[key] = "";
  403. }
  404. return {
  405. "uid": userInfo.uid,
  406. "nickname": userInfo.nickname,
  407. "following_count": userInfo.following_count,
  408. "mplatform_followers_count": userInfo.mplatform_followers_count,
  409. "total_favorited": userInfo.total_favorited,
  410. "unique_id": userInfo.unique_id ? userInfo.unique_id : userInfo.short_id,
  411. "ip_location": userInfo.ip_location.replace("IP属地:", ""),
  412. "gender": userInfo.gender ? " 男女".charAt(userInfo.gender).trim() : "",
  413. "city": [userInfo.province, userInfo.city, userInfo.district].filter(x => x).join("·"),
  414. "signature": userInfo.signature,
  415. "aweme_count": userInfo.aweme_count,
  416. "create_time": Date.now()
  417. }
  418. }
  419.  
  420. function sendLocalData(jsonData) {
  421. if (!localDownload) return;
  422. fetch(localDownloadUrl, {
  423. method: 'POST',
  424. headers: {
  425. 'Content-Type': 'application/json'
  426. },
  427. body: JSON.stringify(jsonData)
  428. })
  429. .then(response => response.json())
  430. .then(responseData => {
  431. console.log('成功:', responseData);
  432. })
  433. .catch(error => {
  434. console.log('上报失败,请检查本地程序是否已经启动!');
  435. });
  436. }
  437.  
  438. function interceptResponse() {
  439. const originalSend = XMLHttpRequest.prototype.send;
  440. XMLHttpRequest.prototype.send = function () {
  441. originalSend.apply(this, arguments);
  442. if (!this._url) return;
  443. this.url = this._url;
  444. if (this.url.startsWith("http"))
  445. this.url = new URL(this.url).pathname
  446. if (!this.url.startsWith("/aweme/v1/web/")) return;
  447. const self = this;
  448. let func = this.onreadystatechange;
  449. this.onreadystatechange = (e) => {
  450. if (self.readyState === 4) {
  451. let data = JSON.parse(self.response);
  452. let jsonData;
  453. if (self.url.startsWith("/aweme/v1/web/user/profile/other")) {
  454. let userInfo = formatUserData(data.user);
  455. user_map.set(userInfo.uid, userInfo);
  456. current_user_id = userInfo.uid;
  457. console.log("加载作者:", current_user_id);
  458. let user_local_data = Array.from(user_map.values()).sort((a, b) => b.create_time - a.create_time);
  459. localStorage.setItem('user_local_data', JSON.stringify(user_local_data.slice(0, max_author_num)));
  460. } else if ([
  461. "/aweme/v1/web/aweme/post/",
  462. "/aweme/v1/web/aweme/related/",
  463. "/aweme/v1/web/aweme/favorite/",
  464. "/aweme/v1/web/mix/aweme/",
  465. "/aweme/v1/web/tab/feed/",
  466. "/aweme/v1/web/aweme/listcollection/",
  467. "/aweme/v1/web/history/read/"
  468. ].some(prefix => self.url.startsWith(prefix))) {
  469. jsonData = formatAwemeData(data);
  470. } else if ([
  471. "/aweme/v1/web/follow/feed/",
  472. "/aweme/v1/web/familiar/feed/",
  473. ].some(prefix => self.url.startsWith(prefix))) {
  474. jsonData = data.data.filter(item => item.aweme).map(item => formatDouyinAwemeData(item.aweme));
  475. } else if (self.url.startsWith("/aweme/v1/web/general/search/single/")) {
  476. jsonData = [];
  477. for (let obj of data.data) {
  478. if (obj.aweme_info) jsonData.push(formatDouyinAwemeData(obj.aweme_info))
  479. if (obj.user_list) {
  480. for (let user of obj.user_list) {
  481. user.items.forEach(aweme => jsonData.push(formatDouyinAwemeData(aweme)))
  482. }
  483. }
  484. }
  485. } else if (self.url.startsWith("/aweme/v1/web/module/feed/")) {
  486. jsonData = data.cards.map(item => formatDouyinAwemeData(JSON.parse(item.aweme)));
  487. } else if (self.url.startsWith("/aweme/v1/web/aweme/detail/")) {
  488. jsonData = [formatDouyinAwemeData(data.aweme_detail)]
  489. }
  490. if (jsonData) jsonData = jsonData.filter(item => item.url && item.awemeId);
  491. if (jsonData) {
  492. sendLocalData(jsonData);
  493. jsonData.forEach(aweme => {
  494. all_aweme_map.set(aweme.awemeId, aweme);
  495. })
  496. if (domLoadedTimer === null) flush();
  497. }
  498. }
  499. if (func) func.apply(self, e);
  500. };
  501. };
  502. }
  503.  
  504. function downloadData(node, encoding) {
  505. if (all_aweme_map.size === 0) {
  506. alert("还没有发现任何作品数据!");
  507. return;
  508. }
  509. if (node.disabled) {
  510. toast("下载正在处理中,请不要重复点击按钮!");
  511. return;
  512. }
  513. node.disabled = true;
  514. try {
  515. let text = "作者昵称,作品描述,作品链接,点赞数,评论数,收藏数,分享数,发布时间,时长,标签,分类,封面,下载链接\n";
  516. let user_aweme_list = Array.from(all_aweme_map.values()).sort((a, b) => b.create_time - a.create_time);
  517. user_aweme_list.forEach(aweme => {
  518. text += [aweme.nickname,
  519. '"' + aweme.desc.replace(/,/g, ',').replace(/"/g, '""') + '"',
  520. "https://www.douyin.com/video/" + aweme.awemeId,
  521. aweme.diggCount, aweme.commentCount,
  522. aweme.collectCount, aweme.shareCount, aweme.date,
  523. aweme.duration, aweme.tag, aweme.video_tag,
  524. aweme.cover, '"' + aweme.url + '"'].join(",") + "\n"
  525. });
  526. if (encoding === "gbk") text = str2gbk(text);
  527. txt2file(text, "【" + timeFormat(Date.now(), "yyyy-mm-dd") + "】抖音当前已加载数据", "csv");
  528. } finally {
  529. node.disabled = false;
  530. }
  531. }
  532.  
  533. function downloadUserData(node, encoding) {
  534. if (user_map.size === 0) {
  535. toast("还没有发现任何作者数据!请访问用户主页后再试!\n以https://www.douyin.com/user/开头的链接。");
  536. return;
  537. }
  538. if (node.disabled) {
  539. toast("下载正在处理中,请不要重复点击按钮!");
  540. return;
  541. }
  542. node.disabled = true;
  543. try {
  544. let text = "昵称,关注,粉丝,获赞,抖音号,IP属地,性别,位置,签名,作品数,主页\n";
  545. Array.from(user_map.values()).forEach(user_info => {
  546. text += [user_info.nickname, user_info.following_count, user_info.mplatform_followers_count,
  547. user_info.total_favorited, user_info.unique_id, user_info.ip_location,
  548. user_info.gender, user_info.city,
  549. '"' + user_info.signature.replace(/,/g, ',').replace(/"/g, '""') + '"',
  550. user_info.aweme_count, "https://www.douyin.com/user/" + user_info.uid].join(",") + "\n"
  551. });
  552. if (encoding === "gbk") text = str2gbk(text);
  553. txt2file(text, null, "csv");
  554. } finally {
  555. node.disabled = false;
  556. }
  557. }
  558.  
  559. let img_button, data_button, user_button, msg_pre;
  560.  
  561. function createMsgBox() {
  562. msg_pre = document.createElement('pre');
  563. msg_pre.textContent = '等待上方头像加载完毕';
  564. msg_pre.style.color = 'white';
  565. msg_pre.style.position = 'fixed';
  566. msg_pre.style.right = '5px';
  567. msg_pre.style.top = '60px';
  568. msg_pre.style.color = 'white';
  569. msg_pre.style.zIndex = '503';
  570. msg_pre.style.opacity = "0.5";
  571. document.body.appendChild(msg_pre);
  572. }
  573.  
  574. function scrollPageToBottom(scroll_button) {
  575. let scrollInterval;
  576.  
  577. function scrollLoop() {
  578. let endText = document.querySelector("div[data-e2e='user-post-list'] > ul[data-e2e='scroll-list'] + div div").innerText;
  579. if (endText.includes("没有更多了")) {
  580. clearInterval(scrollInterval);
  581. scrollInterval = null;
  582. scroll_button.p1.textContent = "已加载全部!";
  583. } else {
  584. scrollTo(0, document.body.scrollHeight);
  585. }
  586. }
  587.  
  588. scroll_button.addEventListener('click', () => {
  589. if (!scrollInterval) {
  590. if (!location.href.startsWith("https://www.douyin.com/user/")) {
  591. toast("不支持非用户主页开启下拉!");
  592. } else if (!document.querySelector("div[data-e2e='user-post-list']")) {
  593. toast("没有找到用户作品列表!");
  594. } else {
  595. scrollInterval = setInterval(scrollLoop, 1200);
  596. scroll_button.p1.textContent = "停止自动下拉";
  597. }
  598. } else {
  599. clearInterval(scrollInterval);
  600. scrollInterval = null;
  601. scroll_button.p1.textContent = "开启自动下拉";
  602. }
  603. });
  604. }
  605.  
  606. function createCommonElement(tagName, attrs = {}, text = "") {
  607. const tag = document.createElement(tagName);
  608. for (const [k, v] of Object.entries(attrs)) {
  609. tag.setAttribute(k, v);
  610. }
  611. if (text) tag.textContent = text;
  612. tag.addEventListener('click', (event) => event.stopPropagation());
  613. return tag;
  614. }
  615.  
  616. function createAllButton() {
  617. let dom = document.querySelector("#douyin-header-menuCt pace-island > div > div:nth-last-child(1) ul a:nth-last-child(1)");
  618. let baseNode = dom.cloneNode(true);
  619. baseNode.removeAttribute("target");
  620. baseNode.removeAttribute("rel");
  621. baseNode.removeAttribute("href");
  622. let svgChild = baseNode.querySelector("svg");
  623. if (svgChild) baseNode.removeChild(svgChild);
  624.  
  625. function createNewButton(name, num = "0") {
  626. let button = baseNode.cloneNode(true);
  627. button.p1 = button.querySelector("p:nth-child(1)");
  628. button.p2 = button.querySelector("p:nth-child(2)");
  629. button.p1.textContent = name;
  630. button.p2.textContent = num;
  631. dom.after(button);
  632. return button;
  633. }
  634.  
  635. img_button = createNewButton("图文打包下载");
  636. img_button.addEventListener('click', () => downloadImg(img_button));
  637.  
  638. let downloadCoverButton = createNewButton("封面打包下载", "");
  639. downloadCoverButton.addEventListener('click', () => downloadCover(downloadCoverButton));
  640.  
  641. data_button = createNewButton("下载已加载的数据");
  642. data_button.p1.after(createCommonElement("label", {'for': 'gbk'}, 'gbk'));
  643. let checkbox = createCommonElement("input", {'type': 'checkbox', 'id': 'gbk'});
  644. checkbox.checked = localStorage.getItem("gbk") === "1";
  645. checkbox.onclick = (event) => {
  646. event.stopPropagation();
  647. localStorage.setItem("gbk", checkbox.checked ? "1" : "0");
  648. };
  649. data_button.p1.after(checkbox);
  650. data_button.addEventListener('click', () => downloadData(data_button, checkbox.checked ? "gbk" : "utf-8"));
  651.  
  652. user_button = createNewButton("下载已游览的作者数据");
  653. user_button.addEventListener('click', () => downloadUserData(user_button, checkbox.checked ? "gbk" : "utf-8"));
  654.  
  655. scrollPageToBottom(createNewButton("开启自动下拉到底", ""));
  656.  
  657. let share_button = document.querySelector("#frame-user-info-share-button");
  658. if (share_button) {
  659. let node = share_button.cloneNode(true);
  660. node.span = node.querySelector("span");
  661. node.span.innerHTML = "复制作者信息";
  662. node.onclick = () => copyUserData(node.span);
  663. share_button.after(node);
  664. }
  665. }
  666.  
  667. async function downloadCover(node) {
  668. if (all_aweme_map.size === 0) {
  669. toast("还没有发现任何作品数据!");
  670. return;
  671. }
  672. if (node.disabled) {
  673. toast("下载正在处理中,请不要重复点击按钮!");
  674. return;
  675. }
  676. node.disabled = true;
  677. try {
  678. const zip = new JSZip();
  679. msg_pre.textContent = `下载封面并打包中...`;
  680. let user_aweme_list = Array.from(all_aweme_map.values()).sort((a, b) => b.create_time - a.create_time);
  681. let promises = user_aweme_list.map((aweme, index) => {
  682. let awemeName = getAwemeName(aweme) + ".jpg";
  683. return fetch(aweme.cover)
  684. .then(response => response.arrayBuffer())
  685. .then(buffer => zip.file(awemeName, buffer))
  686. .then(() => msg_pre.textContent = `${index + 1}/${user_aweme_list.length} ` + awemeName)
  687. });
  688. Promise.all(promises).then(() => {
  689. return zip.generateAsync({type: "blob"})
  690. }).then((content) => {
  691. createDownloadLink(content, null, "zip", "【封面】");
  692. msg_pre.textContent = "封面打包完成";
  693. node.disabled = false;
  694. })
  695. } finally {
  696. node.disabled = false;
  697. }
  698. }
  699.  
  700. async function downloadImg(node) {
  701. if (node.disabled) {
  702. toast("下载正在处理中,请不要重复点击按钮!");
  703. return;
  704. }
  705. node.disabled = true;
  706. try {
  707. const zip = new JSZip();
  708. let flag = true;
  709. let aweme_img_list = Array.from(all_aweme_map.values()).sort((a, b) => b.create_time - a.create_time).filter(a => a.images);
  710. for (let [i, aweme] of aweme_img_list.entries()) {
  711. let awemeName = getAwemeName(aweme);
  712. msg_pre.textContent = `${i + 1}/${aweme_img_list.length} ` + awemeName;
  713. let folder = zip.folder(awemeName);
  714. await Promise.all(aweme.images.map((link, index) => {
  715. return fetch(link)
  716. .then((res) => res.arrayBuffer())
  717. .then((buffer) => {
  718. folder.file(`image_${index + 1}.jpg`, buffer);
  719. });
  720. }));
  721. flag = false;
  722. }
  723. if (flag) {
  724. alert("当前页面未发现图文链接");
  725. node.disabled = false;
  726. return;
  727. }
  728. msg_pre.textContent = "图文打包中...";
  729. zip.generateAsync({type: "blob"})
  730. .then((content) => {
  731. createDownloadLink(content, null, "zip", "【图文】");
  732. msg_pre.textContent = "图文打包完成";
  733. node.disabled = false;
  734. });
  735. } finally {
  736. node.disabled = false;
  737. }
  738. }
  739.  
  740. function douyinVideoDownloader() {
  741. const adjustMargin = (toolDom) => {
  742. let virtualDom = toolDom.querySelector('.virtual');
  743. if (location.href.includes('search') && !location.href.includes('modal_id')) {
  744. toolDom.style.marginTop = "0px";
  745. virtualDom.style.marginBottom = "37px";
  746. } else {
  747. toolDom.style.marginTop = "-68px";
  748. virtualDom.style.marginBottom = "0px";
  749. }
  750. }
  751. const clonePlayclarity2Download = (xgPlayer, videoId, videoContainer) => {
  752. let toolDom = xgPlayer.querySelector(`.xgplayer-playclarity-setting[data-vid]`);
  753. let attrs = {class: "item", style: "text-align:center;"};
  754.  
  755. let aweme = all_aweme_map.get(videoId);
  756. if (toolDom) {
  757. toolDom.dataset.vid = videoId;
  758. videoContainer.dataset.vid = videoId;
  759. adjustMargin(toolDom);
  760. let virtualDom = toolDom.querySelector('.virtual');
  761. if (!aweme) return;
  762. if (!aweme.images && virtualDom.dataset.image) {
  763. virtualDom.removeChild(virtualDom.lastElementChild);
  764. delete virtualDom.dataset.image;
  765. } else if (aweme.images && !virtualDom.dataset.image) {
  766. let downloadDom2 = createCommonElement("div", attrs, "图文下载");
  767. virtualDom.appendChild(downloadDom2);
  768. downloadDom2.onclick = () => {
  769. aweme = all_aweme_map.get(toolDom.dataset.vid);
  770. if (!aweme) {
  771. toast('未捕获到对应数据源!');
  772. } else if (!aweme.images) {
  773. toast('捕获的数据源,不含图片信息!');
  774. } else {
  775. downloadImage(aweme, downloadDom2);
  776. }
  777. };
  778. virtualDom.dataset.image = videoId;
  779. }
  780. return;
  781. }
  782. // console.log("打开视频", videoId);
  783. // if (!aweme) return;
  784. // toast('当前打开的视频未捕获到数据源,若需要下载请转入观看历史下载!');
  785. const parser = new DOMParser();
  786. const doc = parser.parseFromString('<xg-icon class="xgplayer-playclarity-setting" data-state="normal" data-index="7.6">' +
  787. '<div class="gear"><div class="virtual"></div><div class="btn">下载</div></div></xg-icon>', 'text/html');
  788. toolDom = doc.body.firstChild;
  789.  
  790. toolDom.dataset.vid = videoId;
  791. toolDom.dataset.index = "7.6";
  792. videoContainer.dataset.vid = videoId;
  793. toolDom.style.paddingTop = '100px';
  794. adjustMargin(toolDom);
  795.  
  796. let downloadText = toolDom.querySelector('.btn');
  797. if (!downloadText) return;
  798. downloadText.textContent = '下载';
  799. downloadText.style = 'font-size:14px;font-weight:600;';
  800.  
  801. let virtualDom = toolDom.querySelector('.virtual');
  802. if (!virtualDom) return;
  803. toolDom.onmouseover = () => virtualDom.style.display = 'block';
  804. toolDom.onmouseout = () => virtualDom.style.display = 'none';
  805. virtualDom.innerHTML = '';
  806.  
  807. let copyDescDom = createCommonElement("div", attrs, "复制描述");
  808. virtualDom.appendChild(copyDescDom);
  809.  
  810. function checkDatasetVid() {
  811. if (toolDom.dataset.vid === "null") toolDom.dataset.vid = player.root.closest('div[data-e2e="feed-active-video"]').getAttribute('data-e2e-vid');
  812. }
  813.  
  814. copyDescDom.onclick = () => {
  815. checkDatasetVid();
  816. console.log("复制对象:", toolDom.dataset.vid);
  817. aweme = window.all_aweme_map.get(toolDom.dataset.vid);
  818. let textContent = aweme && aweme.desc ? aweme.desc : "";
  819. let videoDescNode = player.root.querySelector('div[data-e2e="video-desc"]');
  820. if (!textContent && videoDescNode) {
  821. textContent = videoDescNode.textContent
  822. }
  823. if (!textContent) {
  824. toast('没有发现描述信息!');
  825. } else {
  826. copyText(textContent, copyDescDom);
  827. }
  828. }
  829. let toLinkDom = createCommonElement("div", attrs, "打开视频");
  830. virtualDom.appendChild(toLinkDom);
  831. toLinkDom.onclick = () => {
  832. checkDatasetVid();
  833. aweme = all_aweme_map.get(toolDom.dataset.vid);
  834. if (aweme && aweme.url) window.open(aweme.url);
  835. else {
  836. window.open(player.videoList[0].playAddr[0].src);
  837. }
  838. };
  839. let downloadDom = createCommonElement("div", attrs, "下载视频");
  840. virtualDom.appendChild(downloadDom);
  841. downloadDom.onclick = () => {
  842. checkDatasetVid();
  843. aweme = all_aweme_map.get(toolDom.dataset.vid);
  844. console.log('下载视频:', toolDom.dataset.vid, aweme);
  845. if (aweme && aweme.url) {
  846. downloadVideo(aweme, downloadDom);
  847. } else if (player) {
  848. downloadUrl(player.videoList[0].playAddr[0].src, downloadDom);
  849. } else {
  850. toast('未捕获到对应数据源!')
  851. }
  852. };
  853. if (aweme && aweme.images) {
  854. let downloadDom2 = createCommonElement("div", attrs, "图文下载");
  855. virtualDom.appendChild(downloadDom2);
  856. downloadDom2.onclick = () => {
  857. aweme = all_aweme_map.get(toolDom.dataset.vid);
  858. if (!aweme) {
  859. toast('未捕获到对应数据源!');
  860. } else if (!aweme.images) {
  861. toast('捕获的数据源,不含图片信息!');
  862. } else {
  863. downloadImage(aweme, downloadDom2);
  864. }
  865. };
  866. virtualDom.dataset.image = videoId;
  867. }
  868. xgPlayer.appendChild(toolDom);
  869. }
  870. const run = (node) => {
  871. if (!node) return;
  872. let activeVideoElement = node.closest('div[data-e2e="feed-active-video"]');
  873. let videoId, xgPlayer, videoContainer;
  874. if (activeVideoElement) {
  875. videoId = activeVideoElement.getAttribute('data-e2e-vid');
  876. xgPlayer = activeVideoElement.querySelector('.xg-right-grid');
  877. videoContainer = activeVideoElement.querySelector("video");
  878. } else {
  879. let playVideoElements = Array.from(document.querySelectorAll('video')).filter(v => v.autoplay);
  880. videoContainer = location.href.includes('modal_id')
  881. ? playVideoElements[0]
  882. : playVideoElements[playVideoElements.length - 1];
  883. xgPlayer = node.closest('.xg-right-grid');
  884. let detailVideoInfo = document.querySelector("[data-e2e='detail-video-info']");
  885. videoId = detailVideoInfo ? detailVideoInfo.getAttribute('data-e2e-aweme-id') : null;
  886. videoId = videoId ? videoId : new URLSearchParams(location.search).get('modal_id');
  887. }
  888. if (!xgPlayer || !videoContainer) return;
  889. clonePlayclarity2Download(xgPlayer, videoId, videoContainer);
  890. }
  891. const rootObserver = new MutationObserver((mutations) => {
  892. mutations.forEach((mutation) => {
  893. mutation.addedNodes.forEach((node) => {
  894. if (node.className === "gear" || (node.className === "xgplayer-icon" && node.dataset.e2e === "video-player-auto-play") ||
  895. (node.classList && node.classList.contains("xgplayer-inner-autoplay"))) {
  896. run(node);
  897. }
  898. // if (node.closest && node.closest('.xg-right-grid')) {
  899. // console.log(node.outerHTML, node);
  900. // }
  901. });
  902. });
  903. });
  904. rootObserver.observe(document.body, {childList: true, subtree: true});
  905. const checkVideoNode = () => {
  906. if (typeof player === "undefined" || !player.video) return;
  907. if (player.root.querySelector(`.xgplayer-playclarity-setting[data-vid]`)) return;
  908. let xgPlayer = player.root.querySelector('.xg-right-grid');
  909. if (!xgPlayer) return;
  910. let activeVideoElement = player.root.closest('div[data-e2e="feed-active-video"]');
  911. let videoId = activeVideoElement ? activeVideoElement.getAttribute('data-e2e-vid') : "";
  912. videoId = videoId ? videoId : new URLSearchParams(location.search).get('modal_id');
  913. clonePlayclarity2Download(xgPlayer, videoId, player.video);
  914. };
  915. setInterval(checkVideoNode, 1000);
  916. }
  917.  
  918. function userDetailObserver() {
  919. const observeList = (scrollList) => {
  920. if (!scrollList) return;
  921. console.log('开始监听新创建的视频列表!');
  922. listObserver.observe(scrollList, {childList: true});
  923. };
  924. const listObserver = new MutationObserver((mutationsList) => {
  925. for (const mutation of mutationsList) {
  926. if (mutation.type !== 'childList') continue;
  927. mutation.addedNodes.forEach(node => {
  928. createButtonGroup(node.querySelector("a"));
  929. });
  930. }
  931. });
  932. const rootObserver = new MutationObserver((mutationsList) => {
  933. for (let mutation of mutationsList) {
  934. if (mutation.type !== 'childList') continue;
  935. mutation.addedNodes.forEach(node => {
  936. if (!node.querySelector) return;
  937. observeList(node.querySelector("ul[data-e2e='scroll-list']"));
  938. });
  939. mutation.removedNodes.forEach(node => {
  940. if (node.querySelector && node.querySelector("ul[data-e2e='scroll-list']")) {
  941. console.log('关闭了一个视频列表');
  942. listObserver.disconnect();
  943. }
  944. });
  945. }
  946. });
  947. rootObserver.observe(document.body, {childList: true, subtree: true});
  948. observeList(document.querySelector("div[data-e2e='user-detail'] ul[data-e2e='scroll-list']"));
  949. }
  950.  
  951. if (document.title === "验证码中间页") return;
  952. createMsgBox();
  953. interceptResponse();
  954. douyinVideoDownloader();
  955. userDetailObserver();
  956. let domLoadedTimer;
  957. const checkElementLoaded = () => {
  958. const element = document.querySelector('#douyin-header-menuCt pace-island > div > div:nth-last-child(1) ul a');
  959. if (element) {
  960. console.log('顶部栏加载完毕');
  961. msg_pre.textContent = "头像加载完成\n若需要下载用户数据,需进入目标用户主页\n若未捕获到数据,可以刷新重试";
  962. clearInterval(domLoadedTimer);
  963. domLoadedTimer = null;
  964. createAllButton();
  965. flush();
  966. }
  967. };
  968. document.window = window;
  969. window.onload = () => {
  970. domLoadedTimer = setInterval(checkElementLoaded, 700);
  971. }
  972. })();