pbl.js 53 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454
  1. // ROUTES FOR OUR APIselectClassStudent
  2. // =============================================================================
  3. const express = require("express");
  4. const request = require("request");
  5. const bcrypt = require("bcryptjs");
  6. const router = express.Router(); // get an instance of the express Router
  7. const querystring = require("querystring");
  8. var mysql = require("./mysql");
  9. // const _mysqlLabor = ["183.36.25.93", "pbl"]; //edu數據庫信息
  10. // const _mysqluser = ["183.36.25.93", "cocorobouser"]; //用户数据库信息
  11. // const _getmysqlLabor = ["123.58.32.151", "pbl"]; //edu數據庫信息f
  12. // const _getmysqluser = ["123.58.32.151", "cocorobouser"]; //edu數據庫信息
  13. const _mysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  14. const _mysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  15. // const _getmysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  16. // const _getmysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  17. const _getmysqlLabor2 = ["172.16.12.7", "pbl"]; //edu數據庫信息
  18. const _getmysqlLabor = ["172.16.12.7", "pbl"]; //edu數據庫信息
  19. const _getmysqluser = ["172.16.12.7", "cocorobouser"]; //edu數據庫信息
  20. const _localhost = ["127.0.0.1", "pbl"]; //edu數據庫信息
  21. const crypto = require("crypto");
  22. const https = require("https");
  23. const { Pool } = require('pg');
  24. const axios = require('axios');
  25. const multer = require("multer");
  26. const {chromium} = require('playwright');
  27. const TurndownService = require('turndown');
  28. const cheerio = require('cheerio');
  29. const { v4: uuidv4 } = require('uuid');
  30. const loginHandler = require('./login'); // 确保路径正确
  31. const { r2rClient } = require("r2r-js");
  32. const client = new r2rClient("https://r2rserver.cocorobo.cn/");
  33. // 配置 multer 存储选项
  34. const storage = multer.diskStorage({
  35. destination: function (req, file, cb) {
  36. req.setEncoding('utf-8');
  37. cb(null, 'uploads/'); // 文件上传目录
  38. },
  39. filename: function (req, file, cb) {
  40. req.setEncoding('utf-8');
  41. // 使用 Buffer 处理文件名,确保编码正确
  42. const fileExtension = file.originalname.split('.').pop(); // 获取文件后缀
  43. const fileNameWithoutExt = Buffer.from(file.originalname.slice(0, -(fileExtension.length + 1)), 'latin1').toString('utf8'); // 转换编码
  44. cb(null, fileNameWithoutExt + '-' + Date.now() + '.' + fileExtension); // 使用时间戳和原始文件名
  45. }
  46. });
  47. const upload = multer({
  48. storage: storage, limits: {
  49. fileSize: 100 * 1024 * 1024 // 100MB
  50. }
  51. });
  52. // 全局浏览器实例
  53. let globalBrowser = null;
  54. async function initBrowser() {
  55. if (!globalBrowser) {
  56. globalBrowser = await chromium.launch({headless: true});
  57. console.log('Browser initialized');
  58. }
  59. return globalBrowser;
  60. }
  61. router.route("/login").all(async (req, res, next) => {
  62. loginHandler(req, res)
  63. });
  64. router.route("/logout").all(async (req, res, next) => {
  65. try {
  66. const response = await client.users.logout();
  67. res.status(200).json({ result: response.results });
  68. } catch (error) {
  69. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  70. }
  71. });
  72. //获取文档列表
  73. router.route("/getDocument").all(async (req, res, next) => {
  74. try {
  75. const response = await client.documents.list({
  76. limit: 10,
  77. offset: 0,
  78. });
  79. console.log(response);
  80. res.status(200).json({ result: response.results });
  81. } catch (error) {
  82. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  83. }
  84. });
  85. //列出文档块
  86. router.route("/listChunks").all(async (req, res, next) => {
  87. try {
  88. const response = await client.documents.listChunks({
  89. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  90. });
  91. console.log(response);
  92. res.status(200).json({ result: response.results });
  93. } catch (error) {
  94. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  95. }
  96. });
  97. //提取实体和关系
  98. router.route("/extract").all(async (req, res, next) => {
  99. try {
  100. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  101. console.log(response);
  102. res.status(200).json({ result: response.results });
  103. } catch (error) {
  104. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  105. }
  106. });
  107. router.route("/selectUser").all((req, res, next) => {
  108. var json = queryString(req.url);
  109. getmysql(req, res, "selectUser", json["userid"]);
  110. });
  111. // 文件上传接口
  112. router.post('/upload', upload.single('file'), async (req, res) => {
  113. req.setEncoding('utf-8')
  114. try {
  115. console.log(req.file); // 打印上传的文件信息
  116. console.log(`File path: uploads/${req.file.filename}`);
  117. const path = require('path');
  118. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  119. console.log(filePath);
  120. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  121. const fs = require('fs');
  122. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  123. const response = await client.documents.create({
  124. file: file,
  125. metadata: { title: filename }
  126. });
  127. console.log(response);
  128. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  129. } catch (error) {
  130. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  131. }
  132. });
  133. //获取文档列表
  134. router.route("/getFile").all(async (req, res, next) => {
  135. //mode的处理
  136. if (req.body[0]) {
  137. p = Object.values(req.body[0]);
  138. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  139. mysql.usselect(p, async function (ret) {
  140. let data = ret[0];
  141. // const array = data.map(item => item.documentid);
  142. const array = data.map(item => item.documentid);
  143. try {
  144. const response = await client.documents.list({ ids: array });
  145. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  146. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  147. // console.log(response);
  148. let data2 = response.results.map(doc => {
  149. doc.documentid = doc.id;
  150. delete doc.id;
  151. return doc;
  152. });
  153. // console.log(data2);
  154. // 合并data和data2,根据documentid
  155. const mergedData = data.map(item => {
  156. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  157. return { ...item, ...correspondingDoc };
  158. });
  159. res.status(200).json({ result: mergedData });
  160. } catch (error) {
  161. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  162. }
  163. });
  164. }
  165. });
  166. //获取文档列表
  167. router.route("/getFileD").all(async (req, res, next) => {
  168. //mode的处理
  169. if (req.body[0]) {
  170. p = Object.values(req.body[0]);
  171. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  172. mysql.usselect(p, async function (ret) {
  173. let data = ret[0];
  174. // const array = data.map(item => item.documentid);
  175. const array = data.map(item => item.documentid);
  176. try {
  177. const response = await client.documents.list({ ids: array });
  178. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  179. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  180. // console.log(response);
  181. let data2 = response.results.map(doc => {
  182. doc.documentid = doc.id;
  183. delete doc.id;
  184. return doc;
  185. });
  186. // console.log(data2);
  187. // 合并data和data2,根据documentid
  188. const mergedData = data.map(item => {
  189. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  190. return { ...item, ...correspondingDoc };
  191. });
  192. res.status(200).json({ result: mergedData });
  193. } catch (error) {
  194. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  195. }
  196. });
  197. }
  198. });
  199. //获取文档列表
  200. router.route("/getFileD2").all(async (req, res, next) => {
  201. //mode的处理
  202. if (req.body) {
  203. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  204. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  205. mysql.usselect(p, async function (ret) {
  206. let data = ret[0];
  207. // const array = data.map(item => item.documentid);
  208. const array = data.map(item => item.documentid);
  209. try {
  210. const response = await client.documents.list({ ids: array });
  211. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  212. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  213. // console.log(response);
  214. let data2 = response.results.map(doc => {
  215. doc.documentid = doc.id;
  216. delete doc.id;
  217. return doc;
  218. });
  219. // console.log(data2);
  220. // 合并data和data2,根据documentid
  221. const mergedData = data.map(item => {
  222. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  223. return { ...item, ...correspondingDoc };
  224. });
  225. res.status(200).json({ result: mergedData });
  226. } catch (error) {
  227. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  228. }
  229. });
  230. }
  231. });
  232. //获取文档列表
  233. router.route("/getFileOpen").all(async (req, res, next) => {
  234. //mode的处理
  235. if (req.body[0]) {
  236. p = Object.values(req.body[0]);
  237. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  238. mysql.usselect(p, async function (ret) {
  239. let data = ret[0];
  240. let user = ret[1];
  241. // const array = data.map(item => item.documentid);
  242. const array = data.map(item => item.documentid);
  243. try {
  244. const response = await client.documents.list({ ids: array });
  245. let data2 = response.results.map(doc => {
  246. doc.documentid = doc.id;
  247. delete doc.id;
  248. return doc;
  249. });
  250. // 合并data和data2,根据documentid
  251. const mergedData = data.map(item => {
  252. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  253. return { ...item, ...correspondingDoc };
  254. });
  255. res.status(200).json({ result: mergedData, user: user });
  256. } catch (error) {
  257. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  258. }
  259. });
  260. }
  261. });
  262. //获取文档列表2
  263. router.route("/getFileP").all(async (req, res, next) => {
  264. //mode的处理
  265. if (req.body) {
  266. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  267. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  268. mysql.usselect(p, async function (ret) {
  269. let data = ret[0];
  270. // const array = data.map(item => item.documentid);
  271. try {
  272. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  273. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  274. // console.log(response);
  275. let data2 = response.results.map(doc => {
  276. doc.documentid = doc.id;
  277. delete doc.id;
  278. return doc;
  279. });
  280. // 合并data和data2,根据documentid
  281. const mergedData = data.map(item => {
  282. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  283. return { ...item, ...correspondingDoc };
  284. });
  285. res.status(200).json({ result: mergedData });
  286. } catch (error) {
  287. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  288. }
  289. });
  290. }
  291. });
  292. //获取能关联的文件
  293. router.route("/getFile2").all(async (req, res, next) => {
  294. //mode的处理
  295. if (req.body[0]) {
  296. p = Object.values(req.body[0]);
  297. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  298. mysql.usselect(p, async function (ret) {
  299. let data = ret[0];
  300. const array = data.map(item => item.documentid);
  301. try {
  302. const response = await client.documents.list({ ids: array });
  303. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  304. // console.log(response);
  305. let data2 = response.results.map(doc => {
  306. doc.documentid = doc.id;
  307. delete doc.id;
  308. return doc;
  309. });
  310. // 合并data和data2,根据documentid
  311. const mergedData = data.map(item => {
  312. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  313. return { ...item, ...correspondingDoc };
  314. });
  315. res.status(200).json({ result: mergedData });
  316. } catch (error) {
  317. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  318. }
  319. });
  320. }
  321. });
  322. //添加文件夹
  323. router.route("/addFolder").all(async (req, res, next) => {
  324. //mode的处理
  325. if (req.body[0]) {
  326. try {
  327. const response = await client.collections.create({
  328. name: decodeURIComponent(req.body[0].n),
  329. description: decodeURIComponent(req.body[0].fd)
  330. });
  331. p = [...Object.values(req.body[0]), response.results.id];
  332. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  333. mysql.usselect(p, async function (ret) {
  334. res.end(JSON.stringify(ret));
  335. });
  336. } catch (error) {
  337. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  338. }
  339. }
  340. });
  341. //添加文件夹
  342. router.route("/addFolderP").all(async (req, res, next) => {
  343. //mode的处理
  344. if (req.body) {
  345. try {
  346. const response = await client.collections.create({
  347. name: decodeURIComponent(req.body.n),
  348. description: decodeURIComponent(req.body.fd)
  349. });
  350. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  351. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  352. mysql.usselect(p, async function (ret) {
  353. res.end(JSON.stringify(ret));
  354. });
  355. } catch (error) {
  356. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  357. }
  358. }
  359. });
  360. //删除文件夹
  361. router.route("/deleteFolder").all(async (req, res, next) => {
  362. //mode的处理
  363. if (req.body[0]) {
  364. try {
  365. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  366. const ids = req.body[0].ids;
  367. // 并行处理文件夹下的文件删除和文件夹删除
  368. await Promise.all(folderids.map(async folderid => {
  369. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  370. await Promise.all(files.results.map(file =>
  371. client.collections.removeDocument({ id: folderid, documentId: file.id })
  372. ));
  373. }));
  374. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  375. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  376. mysql.usselect(params, async function (ret) {
  377. res.end(JSON.stringify(ret));
  378. });
  379. } catch (error) {
  380. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  381. }
  382. }
  383. });
  384. //修改文件夹
  385. router.route("/updateFolder").all(async (req, res, next) => {
  386. //mode的处理
  387. if (req.body[0]) {
  388. try {
  389. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  390. const response = await client.collections.update({
  391. id: folderid,
  392. name: decodeURIComponent(req.body[0].n),
  393. description: decodeURIComponent(req.body[0].d)
  394. });
  395. p = Object.values(folderData); // 获取剩余属性的键
  396. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  397. mysql.usselect(p, async function (ret) {
  398. res.end(JSON.stringify(ret));
  399. });
  400. } catch (error) {
  401. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  402. }
  403. }
  404. });
  405. //修改文件夹
  406. router.route("/updateFolder2").all(async (req, res, next) => {
  407. //mode的处理
  408. if (req.body[0]) {
  409. try {
  410. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  411. const response = await client.collections.update({
  412. id: folderid,
  413. name: decodeURIComponent(req.body[0].n),
  414. description: decodeURIComponent(req.body[0].d)
  415. });
  416. p = Object.values(folderData); // 获取剩余属性的键
  417. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  418. mysql.usselect(p, async function (ret) {
  419. res.end(JSON.stringify(ret));
  420. });
  421. } catch (error) {
  422. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  423. }
  424. }
  425. });
  426. //移除文件
  427. router.route("/removeFile").all(async (req, res, next) => {
  428. if (req.body[0]) {
  429. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  430. let folderid = req.body[0].folderid;
  431. let ids = req.body[0].ids;
  432. // 使用 Promise.all 确保所有删除操作都被执行
  433. const removePromises = documentids.map(documentid =>
  434. client.collections.removeDocument({ id: folderid, documentId: documentid })
  435. );
  436. // 不管client.documents报错没有也要执行removeFile
  437. Promise.all(removePromises).catch(error => {
  438. console.error("移除文件时发生错误:", error);
  439. });
  440. p = [ids, folderid];
  441. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  442. mysql.usselect(p, async function (ret) {
  443. res.end(JSON.stringify(ret));
  444. });
  445. }
  446. });
  447. //移入文件
  448. router.route("/moveFile").all(async (req, res, next) => {
  449. if (req.body[0]) {
  450. try {
  451. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  452. let folderid = req.body[0].folderid
  453. let ids = decodeURIComponent(req.body[0].ids).split(",")
  454. // 使用 Promise.all 确保所有添加操作都被执行
  455. await Promise.all(documentids.map(documentid =>
  456. client.collections.addDocument({
  457. id: folderid,
  458. documentId: documentid
  459. }).catch(error => {
  460. console.error("移动文件时发生错误:", error);
  461. })
  462. ));
  463. for (const id of ids) {
  464. p = [id, folderid];
  465. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  466. await mysql.usselect(p, async function (ret) { });
  467. }
  468. res.status(200).json({ messages: "移动成功" });
  469. } catch (error) {
  470. // 这里不返回错误,确保deleteFile执行完
  471. console.error("处理过程中发生错误:", error);
  472. }
  473. }
  474. });
  475. //删除文件
  476. router.route("/deleteFile").all(async (req, res, next) => {
  477. if (req.body[0]) {
  478. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  479. let ids = req.body[0].ids;
  480. // 使用 Promise.all 确保所有删除操作都被执行
  481. Promise.all(documentids.map(documentid =>
  482. client.documents.delete({ id: documentid }).catch(error => {
  483. console.error("删除文件时发生错误:", error);
  484. })
  485. ));
  486. p = [ids];
  487. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  488. mysql.usselect(p, async function (ret) {
  489. res.end(JSON.stringify(ret));
  490. });
  491. }
  492. });
  493. //删除文件
  494. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  495. if (req.body[0]) {
  496. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  497. let ids = req.body[0].ids;
  498. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  499. console.log("文件删除成功");
  500. }).catch(error => {
  501. console.error("删除文件时发生错误:", error);
  502. })
  503. p = [ids];
  504. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  505. mysql.usselect(p, async function (ret) {
  506. res.end(JSON.stringify(ret));
  507. });
  508. }
  509. });
  510. router.route("/getFolder").all((req, res, next) => {
  511. getmysql2(req, res, "getFolder", req.body.uid);
  512. });
  513. router.route("/getFolder2").all((req, res, next) => {
  514. var json = queryString(req.url);
  515. getmysql2(req, res, "getFolder", json['uid']);
  516. });
  517. router.route("/getFolderPage").all((req, res, next) => {
  518. getmysql2(req, res, "getFolderPage2", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].page, req.body[0].num);
  519. });
  520. router.route("/getFolderPage3").all((req, res, next) => {
  521. getmysql2(req, res, "getFolderPage3", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].tag, req.body[0].page, req.body[0].num);
  522. });
  523. router.route("/getFolderPage32").all((req, res, next) => {
  524. var json = queryString(req.url);
  525. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  526. });
  527. router.route("/getFolderPageOpen").all((req, res, next) => {
  528. getmysql2(req, res, "getFolderPageOpen", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].tag, req.body[0].page, req.body[0].num);
  529. });
  530. router.route("/haveFile").all((req, res, next) => {
  531. postmysql(req, res, "haveFile");
  532. });
  533. router.route("/haveFolder").all((req, res, next) => {
  534. postmysql(req, res, "haveFolder");
  535. });
  536. //search文件
  537. router.route("/searchFile").all(async (req, res, next) => {
  538. if (req.body) {
  539. try {
  540. let filters = {}
  541. if (req.body.type == 1) {
  542. filters = { "document_id": { "$in": req.body.ids } }
  543. } else if (req.body.type == 2) {
  544. filters = { "collection_ids": { "$in": req.body.ids } }
  545. }
  546. const response = await client.retrieval.search({
  547. query: req.body.query,
  548. search_settings: {
  549. filters: filters,
  550. useSemanticSearch: true,
  551. chunkSettings: {
  552. limit: req.body.limit ? req.body.limit : 10,
  553. enabled: true
  554. },
  555. }
  556. });
  557. res.status(200).json({ result: response.results });
  558. } catch (error) {
  559. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  560. }
  561. }
  562. });
  563. //提取文件实体
  564. router.route("/extractFile").all(async (req, res, next) => {
  565. if (req.body) {
  566. try {
  567. const response = client.documents.extract({
  568. id: req.body.id
  569. })
  570. res.status(200).json({ result: response.results });
  571. } catch (error) {
  572. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  573. }
  574. }
  575. });
  576. //添加上传好的知识库文件
  577. router.route("/addFile").all((req, res, next) => {
  578. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  579. })
  580. //获取文件详情
  581. router.route("/getFileDetail").all(async (req, res, next) => {
  582. if (req.body[0]) {
  583. p = Object.values(req.body[0]);
  584. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  585. mysql.usselect(p, async function (ret) {
  586. let data = ret[0][0];
  587. // const array = data.map(item => item.documentid);
  588. try {
  589. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  590. // console.log(response);
  591. let data2 = response.results[0]
  592. // 合并data和data2,根据documentid
  593. const mergedData = { ...data, ...data2 };
  594. res.status(200).json({ result: mergedData });
  595. } catch (error) {
  596. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  597. }
  598. });
  599. }
  600. });
  601. //获取切片
  602. router.route("/getChunks").all(async (req, res, next) => {
  603. if (req.body[0]) {
  604. try {
  605. const { page = 1, limit = 10, documentId } = req.body[0];
  606. const offset = (page - 1) * limit;
  607. const response = await client.documents.listChunks({
  608. id: documentId,
  609. includeVectors: false,
  610. offset,
  611. limit,
  612. });
  613. res.status(200).json({ result: response });
  614. } catch (error) {
  615. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  616. }
  617. }
  618. });
  619. //获取实体
  620. router.route("/getEntities").all(async (req, res, next) => {
  621. if (req.body[0]) {
  622. try {
  623. const { page = 1, limit = 10, documentId } = req.body[0];
  624. const offset = (page - 1) * limit;
  625. const response = await client.documents.listEntities({
  626. id: documentId,
  627. includeVectors: false,
  628. offset,
  629. limit,
  630. });
  631. res.status(200).json({ result: response });
  632. } catch (error) {
  633. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  634. }
  635. }
  636. });
  637. //获取关系
  638. router.route("/getRelationships").all(async (req, res, next) => {
  639. if (req.body[0]) {
  640. try {
  641. const { page = 1, limit = 10, documentId } = req.body[0];
  642. const offset = (page - 1) * limit;
  643. const response = await client.documents.listRelationships({
  644. id: documentId,
  645. includeVectors: false,
  646. offset,
  647. limit,
  648. });
  649. res.status(200).json({ result: response });
  650. } catch (error) {
  651. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  652. }
  653. }
  654. });
  655. //获取实体和关系
  656. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  657. if (req.body[0]) {
  658. try {
  659. const { documentId } = req.body[0];
  660. const offset = (page - 1) * limit;
  661. const response = await client.documents.listEntities({
  662. id: documentId,
  663. includeVectors: false,
  664. });
  665. const response2 = await client.documents.listRelationships({
  666. id: documentId,
  667. includeVectors: false,
  668. });
  669. res.status(200).json({ Entities: response, Relationships: response2 });
  670. } catch (error) {
  671. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  672. }
  673. }
  674. });
  675. //提取功能
  676. router.route("/extractFile2").all(async (req, res, next) => {
  677. try {
  678. const response = await client.documents.extract({ id: req.body[0].documentid });
  679. console.log(response);
  680. res.status(200).json({ result: response.results });
  681. } catch (error) {
  682. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  683. }
  684. });
  685. //复制文件夹
  686. router.route("/copyFolder").all(async (req, res, next) => {
  687. if (req.body[0]) {
  688. const response = await client.collections.create({
  689. name: decodeURIComponent(req.body[0].n),
  690. description: decodeURIComponent(req.body[0].fd)
  691. });
  692. let folderid = response.results.id;
  693. p2 = [req.body[0].folderid]
  694. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  695. mysql.usselect(p2, async function (ret2) {
  696. let file = ret2[0]
  697. if (file.length) {
  698. let documentids = file.map(item => item.documentid);
  699. // 使用 Promise.all 确保所有添加操作都被执行
  700. await Promise.all(documentids.map(documentid =>
  701. client.collections.addDocument({
  702. id: folderid,
  703. documentId: documentid
  704. }).catch(error => {
  705. console.error("移动文件时发生错误:", error);
  706. })
  707. ));
  708. }
  709. let ids = file.map(item => item.fileid);
  710. console.log(ids);
  711. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  712. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  713. mysql.usselect(p, async function (ret) {
  714. res.end(JSON.stringify(ret));
  715. });
  716. });
  717. }
  718. });
  719. // 下载文件功能
  720. router.route("/downloadFile").all(async (req, res, next) => {
  721. var json = queryString(req.url);
  722. let params = {
  723. documentid: (json && json['documentid'] || req.body[0].documentid),
  724. filename: (json && json['filename'] || req.body[0].filename),
  725. };
  726. try {
  727. const response = await client.documents.download({ id: params.documentid });
  728. console.log(response);
  729. // 根据response的类型设置Content-Type
  730. const contentType = response.type || 'application/octet-stream';
  731. res.setHeader('Content-Type', contentType);
  732. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  733. // 直接发送blob数据
  734. res.send(Buffer.from(await response.arrayBuffer()));
  735. } catch (error) {
  736. console.error("下载文件时发生错误:", error);
  737. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  738. }
  739. });
  740. // 下载文件功能
  741. router.route("/downloadFile2").all(async (req, res, next) => {
  742. var json = queryString(req.url);
  743. let params = {
  744. documentid: (json && json['documentid'] || req.body[0].documentid),
  745. filename: (json && json['filename'] || req.body[0].filename),
  746. };
  747. try {
  748. const response = await client.documents.download({ id: params.documentid });
  749. console.log(response);
  750. // 直接返回blob数据
  751. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  752. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  753. const buffer = Buffer.from(await response.arrayBuffer());
  754. res.send(buffer); // 发送blob数据
  755. } catch (error) {
  756. console.error("下载文件时发生错误:", error);
  757. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  758. }
  759. });
  760. //召回测试
  761. router.route("/ragFolder").all(async (req, res, next) => {
  762. let reqJson = req.body[0]
  763. if (reqJson) {
  764. reqJson.query = decodeURIComponent(reqJson.query)
  765. try {
  766. let filters = {}
  767. if (reqJson.type == 1) {
  768. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  769. } else if (reqJson.type == 2) {
  770. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  771. } else if (reqJson.type == 3) {
  772. filters = { "collection_ids": { "$eq": reqJson.ids } }
  773. }
  774. console.log(111);
  775. const response = await client.retrieval.search({
  776. query: reqJson.query,
  777. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  778. // searchMode: 'basic',
  779. searchSettings: {
  780. useHybridSearch: reqJson.useType == 3,
  781. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  782. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  783. hybrid_settings: {
  784. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  785. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  786. },
  787. filters: filters,
  788. limit: reqJson.limit ? reqJson.limit : 100,
  789. },
  790. ragGenerationConfig: {
  791. stream: false,
  792. temperature: reqJson.temperature ? reqJson.temperature : 0,
  793. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  794. }
  795. });
  796. // res.status(200).json({ result: response.results });
  797. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  798. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  799. mysql.usselect(p, async function (ret) {
  800. res.status(200).json({ result: response.results });
  801. });
  802. } catch (error) {
  803. console.log(error);
  804. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  805. }
  806. }
  807. });
  808. //召回测试
  809. router.route("/searchFolder").all(async (req, res, next) => {
  810. let reqJson = req.body[0]
  811. if (reqJson) {
  812. try {
  813. let filters = {}
  814. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  815. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  816. filters = {
  817. "$or": [
  818. { "collection_ids": { "$in": folderids } },
  819. { "document_id": { "$in": documentids } }
  820. ]
  821. };
  822. let params = {
  823. query: reqJson.query,
  824. searchSettings: {
  825. useHybridSearch: reqJson.useType == 3,
  826. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  827. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  828. hybrid_settings: {
  829. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  830. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  831. },
  832. filters: filters,
  833. limit: reqJson.limit ? reqJson.limit : 100,
  834. },
  835. ragGenerationConfig: {
  836. stream: false,
  837. temperature: reqJson.temperature ? reqJson.temperature : 0,
  838. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  839. }
  840. }
  841. // console.log(params);
  842. const response = await client.retrieval.search(params);
  843. res.status(200).json({ result: response.results });
  844. } catch (error) {
  845. console.log(error);
  846. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  847. }
  848. }
  849. });
  850. //召回测试
  851. router.route("/searchFolder2").all(async (req, res, next) => {
  852. let reqJson = req.body
  853. if (Object.keys(reqJson).length) {
  854. try {
  855. let filters = {}
  856. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  857. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  858. filters = {
  859. "$or": [
  860. { "collection_ids": { "$in": folderids } },
  861. { "document_id": { "$in": documentids } }
  862. ]
  863. };
  864. if(!folderids.length && !documentids.length){
  865. res.status(200).json({ result: [] });
  866. return
  867. }
  868. let params = {
  869. query: reqJson.query,
  870. searchSettings: {
  871. useHybridSearch: reqJson.useType == 3,
  872. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  873. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  874. hybrid_settings: {
  875. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  876. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  877. },
  878. filters: filters,
  879. limit: reqJson.limit ? reqJson.limit : 100,
  880. },
  881. ragGenerationConfig: {
  882. stream: false,
  883. temperature: reqJson.temperature ? reqJson.temperature : 0,
  884. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  885. }
  886. }
  887. // console.log(params);
  888. const response = await client.retrieval.search(params);
  889. res.status(200).json({ result: response.results });
  890. } catch (error) {
  891. console.log(error);
  892. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  893. }
  894. }else{
  895. res.status(200).json({ result: "请传参" });
  896. }
  897. });
  898. //召回测试
  899. router.route("/searchFolder3").all(async (req, res, next) => {
  900. let reqJson = req.body[0]
  901. if (reqJson) {
  902. reqJson.query = decodeURIComponent(reqJson.query)
  903. try {
  904. let filters = {}
  905. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  906. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  907. filters = {
  908. "$or": [
  909. { "collection_ids": { "$in": folderids } },
  910. { "document_id": { "$in": documentids } }
  911. ]
  912. };
  913. if(!folderids.length && !documentids.length){
  914. res.status(200).json({ result: [] });
  915. return
  916. }
  917. let params = {
  918. query: reqJson.query,
  919. searchSettings: {
  920. useHybridSearch: reqJson.useType == 3,
  921. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  922. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  923. hybrid_settings: {
  924. fullTextWeight: reqJson.fullTextWeight,
  925. semanticWeight: reqJson.semanticWeight,
  926. },
  927. filters: filters,
  928. limit: reqJson.limit ? reqJson.limit : 100,
  929. },
  930. ragGenerationConfig: {
  931. stream: false,
  932. temperature: reqJson.temperature ? reqJson.temperature : 0,
  933. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  934. }
  935. }
  936. // console.log(params);
  937. const response = await client.retrieval.search(params);
  938. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  939. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  940. mysql.usselect(p, async function (ret) {
  941. res.status(200).json({ result: response.results });
  942. });
  943. } catch (error) {
  944. console.log(error);
  945. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  946. }
  947. }
  948. });
  949. //召回测试
  950. router.route("/searchFolder32").all(async (req, res, next) => {
  951. let reqJson = req.body[0]
  952. if (reqJson) {
  953. reqJson.query = decodeURIComponent(reqJson.query)
  954. try {
  955. let filters = {}
  956. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  957. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  958. filters = {
  959. "$or": [
  960. { "collection_ids": { "$in": folderids } },
  961. { "document_id": { "$in": documentids } }
  962. ]
  963. };
  964. if(!folderids.length && !documentids.length){
  965. res.status(200).json({ result: [] });
  966. return
  967. }
  968. let params = {
  969. query: reqJson.query,
  970. searchSettings: {
  971. useHybridSearch: reqJson.useType == 3,
  972. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  973. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  974. hybrid_settings: {
  975. fullTextWeight: reqJson.fullTextWeight,
  976. semanticWeight: reqJson.semanticWeight,
  977. },
  978. filters: filters,
  979. limit: reqJson.limit ? reqJson.limit : 100,
  980. },
  981. ragGenerationConfig: {
  982. stream: false,
  983. temperature: reqJson.temperature ? reqJson.temperature : 0,
  984. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  985. }
  986. }
  987. console.log(params);
  988. const response = await client.retrieval.search(params);
  989. res.status(200).json({ result: response.results });
  990. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  991. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  992. // mysql.usselect(p, async function (ret) {
  993. // });
  994. } catch (error) {
  995. console.log(error);
  996. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  997. }
  998. }
  999. });
  1000. //获取召回测试数据
  1001. router.route("/getRagData").all(async (req, res, next) => {
  1002. //mode的处理
  1003. if (req.body[0]) {
  1004. p = Object.values(req.body[0]);
  1005. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1006. mysql.usselect(p, async function (ret) {
  1007. res.status(200).json(ret);
  1008. });
  1009. }
  1010. });
  1011. //删除召回测试数据
  1012. router.route("/delRagData").all(async (req, res, next) => {
  1013. //mode的处理
  1014. if (req.body[0]) {
  1015. p = Object.values(req.body[0]);
  1016. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1017. mysql.usselect(p, async function (ret) {
  1018. res.status(200).json(ret);
  1019. });
  1020. }
  1021. });
  1022. function generateAPIKey(username) {
  1023. let timeStamp = new Date().getTime();
  1024. let hashStr = timeStamp + username;
  1025. let digest = crypto
  1026. .createHash("sha256")
  1027. .update(hashStr, "utf8")
  1028. .digest("hex");
  1029. return digest;
  1030. }
  1031. queryString = function (url) {
  1032. var a = url.split("?");
  1033. var json = querystring.parse(a[1]);
  1034. return json;
  1035. };
  1036. //post存儲過程處理
  1037. postmysql = function (req, res, functionname) {
  1038. //mode的處理
  1039. if (req.body[0]) {
  1040. p = Object.values(req.body[0]);
  1041. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1042. //執行存儲過程
  1043. mysql.usselect(p, function (ret) {
  1044. res.end(JSON.stringify(ret));
  1045. });
  1046. }
  1047. };
  1048. //post存儲過程處理
  1049. postmysql2 = function (req, res, functionname) {
  1050. //mode的處理
  1051. if (req.body) {
  1052. p = Object.values(req.body)[0].split(",");
  1053. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1054. //執行存儲過程
  1055. mysql.usselect(p, function (ret) {
  1056. res.end(JSON.stringify(ret));
  1057. });
  1058. }
  1059. };
  1060. //post存儲過程處理
  1061. postmysql3 = function (req, res, functionname) {
  1062. //mode的處理
  1063. if (req.body) {
  1064. console.log(req);
  1065. p = Object.values(req.body);
  1066. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1067. //執行存儲過程
  1068. mysql.usselect(p, function (ret) {
  1069. res.end(JSON.stringify(ret));
  1070. });
  1071. }
  1072. };
  1073. //get獲取數據庫信息
  1074. getmysql = function (req, res, functionname, ...p) {
  1075. p = p || [];
  1076. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1077. //執行存儲過程
  1078. mysql.usselect(p, function (ret) {
  1079. res.end(JSON.stringify(ret));
  1080. }); //
  1081. };
  1082. //get獲取數據庫信息
  1083. getmysql2 = function (req, res, functionname, ...p) {
  1084. p = p || [];
  1085. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1086. //執行存儲過程
  1087. mysql.usselect(p, function (ret) {
  1088. res.end(JSON.stringify(ret));
  1089. }); //
  1090. };
  1091. //get獲取數據庫信息
  1092. getmysql3 = function (req, res, functionname, ...p) {
  1093. p = p || [];
  1094. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1095. //執行存儲過程
  1096. mysql.usselect(p, function (ret) {
  1097. res.end(JSON.stringify(ret));
  1098. }); //
  1099. };
  1100. //post存储过程处理
  1101. userpostmysql = function (req, res, functionname) {
  1102. //mode的处理
  1103. if (req.body.mode) {
  1104. p = req.body.mode.split(",");
  1105. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1106. //执行存储过程
  1107. mysql.usselect(p, function (ret) {
  1108. res.end(JSON.stringify(ret));
  1109. });
  1110. }
  1111. };
  1112. //get获取数据库信息
  1113. usergetmysql = function (req, res, functionname, ...p) {
  1114. p = p || [];
  1115. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1116. //执行存储过程
  1117. mysql.usselect(p, function (ret) {
  1118. res.end(JSON.stringify(ret));
  1119. }); //
  1120. };
  1121. //post存儲過程處理
  1122. postmysqlLocal = function (req, res) {
  1123. //mode的處理
  1124. if (req.body[0]) {
  1125. p = Object.values(req.body[0]);
  1126. p.unshift(_localhost[0], _localhost[1]);
  1127. //執行存儲過程
  1128. mysql.usselect(p, function (ret) {
  1129. res.end(JSON.stringify(ret));
  1130. });
  1131. }
  1132. };
  1133. postmysqlOnline = function (req, res) {
  1134. //mode的處理
  1135. if (req.body[0]) {
  1136. p = Object.values(req.body[0]);
  1137. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1138. //執行存儲過程
  1139. mysql.usselect(p, function (ret) {
  1140. res.end(JSON.stringify(ret));
  1141. });
  1142. }
  1143. };
  1144. // 新增测试接口,直接返回成功
  1145. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1146. // console.log('收到的 req.body:', req.body);
  1147. // console.log('收到的 req.files:', req.files);
  1148. // console.log('收到的 headers:', req.headers);
  1149. const requestUrl = req.body.url; // 前端传入 url
  1150. if (!requestUrl) {
  1151. return res.status(400).json({error: 'Missing url'});
  1152. }
  1153. const browser = await initBrowser();
  1154. const page = await browser.newPage();
  1155. try {
  1156. console.log("开始抓取页面:", requestUrl);
  1157. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1158. await page.goto(requestUrl);
  1159. const html = await page.content();
  1160. // 新增:HTML 转 Markdown
  1161. const turndownService = new TurndownService();
  1162. // 用 cheerio 只提取正文部分
  1163. const cheerioDoc = cheerio.load(html);
  1164. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1165. // 用 cheerio 过滤无关标签
  1166. const cheerioMain = cheerio.load(mainHtml || '');
  1167. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1168. mainHtml = cheerioMain.html();
  1169. const markdownContent = turndownService.turndown(mainHtml);
  1170. // 新增:解析 <title> 标签内容
  1171. let title = '';
  1172. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1173. if (titleMatch && titleMatch[1]) {
  1174. title = titleMatch[1].trim();
  1175. }
  1176. let finalTitle;
  1177. if (title && title.trim()) {
  1178. finalTitle = title + '.md';
  1179. } else {
  1180. finalTitle = 'untitled-' + uuidv4() + '.md';
  1181. }
  1182. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1183. // 解析用户传入的参数
  1184. const {
  1185. collection_ids,
  1186. id: customId,
  1187. metadata: customMetadata,
  1188. ingestion_mode,
  1189. run_with_orchestration
  1190. } = req.body;
  1191. // 处理 metadata 字段
  1192. let finalMetadata = customMetadata;
  1193. if (typeof finalMetadata === 'string') {
  1194. try {
  1195. finalMetadata = JSON.parse(finalMetadata);
  1196. } catch (e) {
  1197. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1198. }
  1199. }
  1200. // if (!finalMetadata) {
  1201. // finalMetadata = {
  1202. // title: title || uuidv4(),
  1203. // // source_url: url,
  1204. // content_type: "markdown",
  1205. // // scraped_at: new Date().toISOString(),
  1206. // // upload_id: Date.now() + '-' + Math.random()
  1207. // };
  1208. // }
  1209. // 处理 collection_ids 字段
  1210. let finalCollectionIds = collection_ids;
  1211. if (typeof finalCollectionIds === 'string') {
  1212. try {
  1213. finalCollectionIds = JSON.parse(finalCollectionIds);
  1214. } catch (e) {
  1215. // 不是合法 JSON 字符串就当作普通字符串
  1216. }
  1217. }
  1218. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1219. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1220. finalMetadata.collection_ids = finalCollectionIds;
  1221. }
  1222. // 补全 finalMetadata.title 和 content_type
  1223. if (!finalMetadata.title) {
  1224. finalMetadata.title = finalTitle;
  1225. }
  1226. finalMetadata.content_type = 'markdown';
  1227. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1228. const FormData = require('form-data');
  1229. const fs = require('fs');
  1230. const path = require('path');
  1231. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1232. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1233. const form = new FormData();
  1234. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1235. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1236. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1237. }
  1238. if (customId) {
  1239. form.append('id', customId);
  1240. }
  1241. form.append('metadata', JSON.stringify(finalMetadata));
  1242. form.append('ingestion_mode', ingestion_mode || 'fast');
  1243. form.append('run_with_orchestration', 'false');
  1244. // 打印上传参数,方便排查问题
  1245. console.log('Request Headers:', form.getHeaders());
  1246. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1247. const response = await axios.post(url, form, {
  1248. headers: {
  1249. ...form.getHeaders(),
  1250. // 'User-Agent': 'NodeClient/1.0'
  1251. }
  1252. });
  1253. fs.unlinkSync(tmpFilePath);
  1254. result = {
  1255. results: response.data.results,
  1256. title: finalTitle
  1257. }
  1258. res.json(result);
  1259. } catch (error) {
  1260. console.error('Message:', error.message);
  1261. console.error('Stack:', error.stack);
  1262. res.status(500).json({error: error.message});
  1263. } finally {
  1264. await page.close(); // 确保页面总是被关闭
  1265. }
  1266. });
  1267. module.exports = router;