pbl.js 60 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640
  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("/getFileR2r").all(async (req, res, next) => {
  168. //mode的处理
  169. if (req.body[0]) {
  170. let fileid = decodeURIComponent(req.body[0].fileids).split(",");
  171. try {
  172. const response = await client.documents.list({ ids: fileid });
  173. res.status(200).json({ result: response.results });
  174. } catch (error) {
  175. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  176. }
  177. }
  178. });
  179. //获取文档列表
  180. router.route("/getFileD").all(async (req, res, next) => {
  181. //mode的处理
  182. if (req.body[0]) {
  183. p = Object.values(req.body[0]);
  184. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  185. mysql.usselect(p, async function (ret) {
  186. let data = ret[0];
  187. // const array = data.map(item => item.documentid);
  188. const array = data.map(item => item.documentid);
  189. try {
  190. const response = await client.documents.list({ ids: array });
  191. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  192. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  193. // console.log(response);
  194. let data2 = response.results.map(doc => {
  195. doc.documentid = doc.id;
  196. delete doc.id;
  197. return doc;
  198. });
  199. // console.log(data2);
  200. // 合并data和data2,根据documentid
  201. const mergedData = data.map(item => {
  202. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  203. return { ...item, ...correspondingDoc };
  204. });
  205. res.status(200).json({ result: mergedData });
  206. } catch (error) {
  207. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  208. }
  209. });
  210. }
  211. });
  212. //获取文档列表
  213. router.route("/getFileArray").all(async (req, res, next) => {
  214. //mode的处理
  215. if (req.body[0]) {
  216. let data = req.body[0].array.split(",");
  217. // const array = data.map(item => item.documentid);
  218. const array = data;
  219. try {
  220. const response = await client.documents.list({ ids: array });
  221. res.status(200).json({ result: response.results });
  222. } catch (error) {
  223. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  224. }
  225. }
  226. });
  227. //获取文档列表
  228. router.route("/getFileD2").all(async (req, res, next) => {
  229. //mode的处理
  230. if (req.body) {
  231. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  232. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  233. mysql.usselect(p, async function (ret) {
  234. let data = ret[0];
  235. // const array = data.map(item => item.documentid);
  236. const array = data.map(item => item.documentid);
  237. try {
  238. const response = await client.documents.list({ ids: array });
  239. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  240. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  241. // console.log(response);
  242. let data2 = response.results.map(doc => {
  243. doc.documentid = doc.id;
  244. delete doc.id;
  245. return doc;
  246. });
  247. // console.log(data2);
  248. // 合并data和data2,根据documentid
  249. const mergedData = data.map(item => {
  250. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  251. return { ...item, ...correspondingDoc };
  252. });
  253. res.status(200).json({ result: mergedData });
  254. } catch (error) {
  255. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  256. }
  257. });
  258. }
  259. });
  260. //获取文档列表
  261. router.route("/getFileD2").all(async (req, res, next) => {
  262. //mode的处理
  263. if (req.body) {
  264. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  265. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  266. mysql.usselect(p, async function (ret) {
  267. let data = ret[0];
  268. // const array = data.map(item => item.documentid);
  269. const array = data.map(item => item.documentid);
  270. try {
  271. const response = await client.documents.list({ ids: array });
  272. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  273. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  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. // console.log(data2);
  281. // 合并data和data2,根据documentid
  282. const mergedData = data.map(item => {
  283. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  284. return { ...item, ...correspondingDoc };
  285. });
  286. res.status(200).json({ result: mergedData });
  287. } catch (error) {
  288. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  289. }
  290. });
  291. }
  292. });
  293. //获取文档列表
  294. router.route("/getFileOpen").all(async (req, res, next) => {
  295. //mode的处理
  296. if (req.body[0]) {
  297. p = Object.values(req.body[0]);
  298. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  299. mysql.usselect(p, async function (ret) {
  300. let data = ret[0];
  301. let user = ret[1];
  302. // const array = data.map(item => item.documentid);
  303. const array = data.map(item => item.documentid);
  304. try {
  305. const response = await client.documents.list({ ids: array });
  306. let data2 = response.results.map(doc => {
  307. doc.documentid = doc.id;
  308. delete doc.id;
  309. return doc;
  310. });
  311. // 合并data和data2,根据documentid
  312. const mergedData = data.map(item => {
  313. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  314. return { ...item, ...correspondingDoc };
  315. });
  316. res.status(200).json({ result: mergedData, user: user });
  317. } catch (error) {
  318. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  319. }
  320. });
  321. }
  322. });
  323. //获取文档列表2
  324. router.route("/getFileP").all(async (req, res, next) => {
  325. //mode的处理
  326. if (req.body) {
  327. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  328. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  329. mysql.usselect(p, async function (ret) {
  330. let data = ret[0];
  331. // const array = data.map(item => item.documentid);
  332. try {
  333. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  334. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  335. // console.log(response);
  336. let data2 = response.results.map(doc => {
  337. doc.documentid = doc.id;
  338. delete doc.id;
  339. return doc;
  340. });
  341. // 合并data和data2,根据documentid
  342. const mergedData = data.map(item => {
  343. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  344. return { ...item, ...correspondingDoc };
  345. });
  346. res.status(200).json({ result: mergedData });
  347. } catch (error) {
  348. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  349. }
  350. });
  351. }
  352. });
  353. //获取能关联的文件
  354. router.route("/getFile2").all(async (req, res, next) => {
  355. //mode的处理
  356. if (req.body[0]) {
  357. p = Object.values(req.body[0]);
  358. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  359. mysql.usselect(p, async function (ret) {
  360. let data = ret[0];
  361. const array = data.map(item => item.documentid);
  362. try {
  363. const response = await client.documents.list({ ids: array });
  364. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  365. // console.log(response);
  366. let data2 = response.results.map(doc => {
  367. doc.documentid = doc.id;
  368. delete doc.id;
  369. return doc;
  370. });
  371. // 合并data和data2,根据documentid
  372. const mergedData = data.map(item => {
  373. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  374. return { ...item, ...correspondingDoc };
  375. });
  376. res.status(200).json({ result: mergedData });
  377. } catch (error) {
  378. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  379. }
  380. });
  381. }
  382. });
  383. //添加文件夹
  384. router.route("/addFolder").all(async (req, res, next) => {
  385. //mode的处理
  386. if (req.body[0]) {
  387. try {
  388. const response = await client.collections.create({
  389. name: decodeURIComponent(req.body[0].n),
  390. description: decodeURIComponent(req.body[0].fd)
  391. });
  392. p = [...Object.values(req.body[0]), response.results.id];
  393. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  394. mysql.usselect(p, async function (ret) {
  395. res.end(JSON.stringify(ret));
  396. });
  397. } catch (error) {
  398. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  399. }
  400. }
  401. });
  402. //添加文件夹
  403. router.route("/addFolderP").all(async (req, res, next) => {
  404. //mode的处理
  405. if (req.body) {
  406. try {
  407. const response = await client.collections.create({
  408. name: decodeURIComponent(req.body.n),
  409. description: decodeURIComponent(req.body.fd)
  410. });
  411. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  412. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  413. mysql.usselect(p, async function (ret) {
  414. res.end(JSON.stringify(ret));
  415. });
  416. } catch (error) {
  417. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  418. }
  419. }
  420. });
  421. //删除文件夹
  422. router.route("/deleteFolder").all(async (req, res, next) => {
  423. //mode的处理
  424. if (req.body[0]) {
  425. try {
  426. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  427. const ids = req.body[0].ids;
  428. // 并行处理文件夹下的文件删除和文件夹删除
  429. await Promise.all(folderids.map(async folderid => {
  430. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  431. await Promise.all(files.results.map(file =>
  432. client.collections.removeDocument({ id: folderid, documentId: file.id })
  433. ));
  434. }));
  435. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  436. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  437. mysql.usselect(params, async function (ret) {
  438. res.end(JSON.stringify(ret));
  439. });
  440. } catch (error) {
  441. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  442. }
  443. }
  444. });
  445. //修改文件夹
  446. router.route("/updateFolder").all(async (req, res, next) => {
  447. //mode的处理
  448. if (req.body[0]) {
  449. try {
  450. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  451. const response = await client.collections.update({
  452. id: folderid,
  453. name: decodeURIComponent(req.body[0].n),
  454. description: decodeURIComponent(req.body[0].d)
  455. });
  456. p = Object.values(folderData); // 获取剩余属性的键
  457. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  458. mysql.usselect(p, async function (ret) {
  459. res.end(JSON.stringify(ret));
  460. });
  461. } catch (error) {
  462. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  463. }
  464. }
  465. });
  466. //修改文件夹
  467. router.route("/updateFolder2").all(async (req, res, next) => {
  468. //mode的处理
  469. if (req.body[0]) {
  470. try {
  471. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  472. const response = await client.collections.update({
  473. id: folderid,
  474. name: decodeURIComponent(req.body[0].n),
  475. description: decodeURIComponent(req.body[0].d)
  476. });
  477. p = Object.values(folderData); // 获取剩余属性的键
  478. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  479. mysql.usselect(p, async function (ret) {
  480. res.end(JSON.stringify(ret));
  481. });
  482. } catch (error) {
  483. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  484. }
  485. }
  486. });
  487. //移除文件
  488. router.route("/removeFile").all(async (req, res, next) => {
  489. if (req.body[0]) {
  490. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  491. let folderid = req.body[0].folderid;
  492. let ids = req.body[0].ids;
  493. // 使用 Promise.all 确保所有删除操作都被执行
  494. const removePromises = documentids.map(documentid =>
  495. client.collections.removeDocument({ id: folderid, documentId: documentid })
  496. );
  497. // 不管client.documents报错没有也要执行removeFile
  498. Promise.all(removePromises).catch(error => {
  499. console.error("移除文件时发生错误:", error);
  500. });
  501. p = [ids, folderid];
  502. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  503. mysql.usselect(p, async function (ret) {
  504. res.end(JSON.stringify(ret));
  505. });
  506. }
  507. });
  508. //移入文件
  509. router.route("/moveFile").all(async (req, res, next) => {
  510. if (req.body[0]) {
  511. try {
  512. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  513. let folderid = req.body[0].folderid
  514. let ids = decodeURIComponent(req.body[0].ids).split(",")
  515. // 使用 Promise.all 确保所有添加操作都被执行
  516. await Promise.all(documentids.map(documentid =>
  517. client.collections.addDocument({
  518. id: folderid,
  519. documentId: documentid
  520. }).catch(error => {
  521. console.error("移动文件时发生错误:", error);
  522. })
  523. ));
  524. for (const id of ids) {
  525. p = [id, folderid];
  526. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  527. await mysql.usselect(p, async function (ret) { });
  528. }
  529. res.status(200).json({ messages: "移动成功" });
  530. } catch (error) {
  531. // 这里不返回错误,确保deleteFile执行完
  532. console.error("处理过程中发生错误:", error);
  533. }
  534. }
  535. });
  536. //删除文件
  537. router.route("/deleteFile").all(async (req, res, next) => {
  538. if (req.body[0]) {
  539. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  540. let ids = req.body[0].ids;
  541. // 使用 Promise.all 确保所有删除操作都被执行
  542. Promise.all(documentids.map(documentid =>
  543. client.documents.delete({ id: documentid }).catch(error => {
  544. console.error("删除文件时发生错误:", error);
  545. })
  546. ));
  547. p = [ids];
  548. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  549. mysql.usselect(p, async function (ret) {
  550. res.end(JSON.stringify(ret));
  551. });
  552. }
  553. });
  554. //删除文件
  555. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  556. if (req.body[0]) {
  557. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  558. let ids = req.body[0].ids;
  559. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  560. console.log("文件删除成功");
  561. }).catch(error => {
  562. console.error("删除文件时发生错误:", error);
  563. })
  564. p = [ids];
  565. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  566. mysql.usselect(p, async function (ret) {
  567. res.end(JSON.stringify(ret));
  568. });
  569. }
  570. });
  571. router.route("/getFolder").all((req, res, next) => {
  572. getmysql2(req, res, "getFolder", req.body.uid);
  573. });
  574. router.route("/getFolder2").all((req, res, next) => {
  575. var json = queryString(req.url);
  576. getmysql2(req, res, "getFolder", json['uid']);
  577. });
  578. router.route("/getFolderPage").all((req, res, next) => {
  579. 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);
  580. });
  581. router.route("/getFolderPage3").all((req, res, next) => {
  582. 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);
  583. });
  584. router.route("/getFolderPage32").all((req, res, next) => {
  585. var json = queryString(req.url);
  586. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  587. });
  588. router.route("/getFolderPageOpen").all((req, res, next) => {
  589. 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);
  590. });
  591. router.route("/haveFile").all((req, res, next) => {
  592. postmysql(req, res, "haveFile");
  593. });
  594. router.route("/haveFolder").all((req, res, next) => {
  595. postmysql(req, res, "haveFolder");
  596. });
  597. //search文件
  598. router.route("/searchFile").all(async (req, res, next) => {
  599. if (req.body) {
  600. try {
  601. let filters = {}
  602. if (req.body.type == 1) {
  603. filters = { "document_id": { "$in": req.body.ids } }
  604. } else if (req.body.type == 2) {
  605. filters = { "collection_ids": { "$in": req.body.ids } }
  606. }
  607. const response = await client.retrieval.search({
  608. query: req.body.query,
  609. search_settings: {
  610. filters: filters,
  611. useSemanticSearch: true,
  612. chunkSettings: {
  613. limit: req.body.limit ? req.body.limit : 10,
  614. enabled: true
  615. },
  616. }
  617. });
  618. res.status(200).json({ result: response.results });
  619. } catch (error) {
  620. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  621. }
  622. }
  623. });
  624. //提取文件实体
  625. router.route("/extractFile").all(async (req, res, next) => {
  626. if (req.body) {
  627. try {
  628. const response = client.documents.extract({
  629. id: req.body.id
  630. })
  631. res.status(200).json({ result: response.results });
  632. } catch (error) {
  633. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  634. }
  635. }
  636. });
  637. //添加上传好的知识库文件
  638. router.route("/addFile").all((req, res, next) => {
  639. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  640. })
  641. //获取文件详情
  642. router.route("/getFileDetail").all(async (req, res, next) => {
  643. if (req.body[0]) {
  644. p = Object.values(req.body[0]);
  645. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  646. mysql.usselect(p, async function (ret) {
  647. let data = ret[0][0];
  648. // const array = data.map(item => item.documentid);
  649. try {
  650. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  651. // console.log(response);
  652. let data2 = response.results[0]
  653. // 合并data和data2,根据documentid
  654. const mergedData = { ...data, ...data2 };
  655. res.status(200).json({ result: mergedData });
  656. } catch (error) {
  657. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  658. }
  659. });
  660. }
  661. });
  662. //获取切片
  663. router.route("/getChunks").all(async (req, res, next) => {
  664. if (req.body[0]) {
  665. try {
  666. const { page = 1, limit = 10, documentId } = req.body[0];
  667. const offset = (page - 1) * limit;
  668. const response = await client.documents.listChunks({
  669. id: documentId,
  670. includeVectors: false,
  671. offset,
  672. limit,
  673. });
  674. res.status(200).json({ result: response });
  675. } catch (error) {
  676. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  677. }
  678. }
  679. });
  680. //获取实体
  681. router.route("/getEntities").all(async (req, res, next) => {
  682. if (req.body[0]) {
  683. try {
  684. const { page = 1, limit = 10, documentId } = req.body[0];
  685. const offset = (page - 1) * limit;
  686. const response = await client.documents.listEntities({
  687. id: documentId,
  688. includeVectors: false,
  689. offset,
  690. limit,
  691. });
  692. res.status(200).json({ result: response });
  693. } catch (error) {
  694. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  695. }
  696. }
  697. });
  698. //获取关系
  699. router.route("/getRelationships").all(async (req, res, next) => {
  700. if (req.body[0]) {
  701. try {
  702. const { page = 1, limit = 10, documentId } = req.body[0];
  703. const offset = (page - 1) * limit;
  704. const response = await client.documents.listRelationships({
  705. id: documentId,
  706. includeVectors: false,
  707. offset,
  708. limit,
  709. });
  710. res.status(200).json({ result: response });
  711. } catch (error) {
  712. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  713. }
  714. }
  715. });
  716. //提取整个文件夹的关系
  717. router.route("/resetGraphs").all(async (req, res, next) => {
  718. if (req.body[0]) {
  719. try {
  720. // const response = await client.graphs.pull({
  721. // collectionId: req.body[0].collectionId,
  722. // });
  723. const response = await client.graphs.reset({
  724. collectionId: req.body[0].collectionId,
  725. });
  726. res.status(200).json({ result: response });
  727. } catch (error) {
  728. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  729. }
  730. }
  731. });
  732. //获取整个文件夹的关系
  733. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  734. if (req.body[0]) {
  735. try {
  736. await client.graphs.pull({
  737. collectionId: req.body[0].collectionId,
  738. });
  739. const response = await client.graphs.listRelationships({
  740. collectionId: req.body[0].collectionId,
  741. offset: req.body[0].page ? req.body[0].page : 0,
  742. limit: 1000
  743. });
  744. res.status(200).json({ result: response });
  745. } catch (error) {
  746. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  747. }
  748. }
  749. });
  750. //提取知识图谱
  751. router.route("/getGraphs").all(async (req, res, next) => {
  752. console.log(req.body);
  753. //mode的处理
  754. if (req.body[0]) {
  755. try {
  756. //创建临时文件夹
  757. const folder = await client.collections.create({
  758. name: "临时文件夹",
  759. description: "临时文件夹"
  760. });
  761. let folderid = folder.results.id
  762. console.log(folder.results.id);
  763. //将要提取的文件添加到临时文件夹
  764. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  765. // 使用 Promise.all 确保所有添加操作都被执行
  766. await Promise.all(documentids.map(documentid =>
  767. client.collections.addDocument({
  768. id: folderid,
  769. documentId: documentid
  770. }).catch(error => {
  771. console.error("移动文件时发生错误:", error);
  772. })
  773. ));
  774. //将临时文件夹提取知识图谱
  775. await client.graphs.pull({
  776. collectionId: folderid
  777. });
  778. let offset = 0;
  779. let results = [];
  780. let totalEntries = 0;
  781. do {
  782. const response = await client.graphs.listRelationships({
  783. collectionId: folderid,
  784. offset: offset,
  785. limit: 1000
  786. });
  787. results = results.concat(response.results);
  788. totalEntries = response.totalEntries;
  789. offset += 1000;
  790. } while (offset < totalEntries);
  791. //移除临时文件夹的文件
  792. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  793. await Promise.all(files.results.map(file =>
  794. client.collections.removeDocument({ id: folderid, documentId: file.id })
  795. ));
  796. //删除临时文件夹
  797. await client.collections.delete({ id: folderid });
  798. res.status(200).json({ result: results });
  799. } catch (error) {
  800. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  801. }
  802. }
  803. });
  804. //获取实体和关系
  805. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  806. if (req.body[0]) {
  807. try {
  808. const { documentId } = req.body[0];
  809. const offset = (page - 1) * limit;
  810. const response = await client.documents.listEntities({
  811. id: documentId,
  812. includeVectors: false,
  813. });
  814. const response2 = await client.documents.listRelationships({
  815. id: documentId,
  816. includeVectors: false,
  817. });
  818. res.status(200).json({ Entities: response, Relationships: response2 });
  819. } catch (error) {
  820. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  821. }
  822. }
  823. });
  824. //提取功能
  825. router.route("/extractFile2").all(async (req, res, next) => {
  826. try {
  827. let setting = {}
  828. if(req.body[0].prompt){
  829. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  830. }
  831. console.log(setting);
  832. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  833. console.log(response);
  834. res.status(200).json({ result: response.results });
  835. } catch (error) {
  836. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  837. }
  838. });
  839. //复制文件夹
  840. router.route("/copyFolder").all(async (req, res, next) => {
  841. if (req.body[0]) {
  842. const response = await client.collections.create({
  843. name: decodeURIComponent(req.body[0].n),
  844. description: decodeURIComponent(req.body[0].fd)
  845. });
  846. let folderid = response.results.id;
  847. p2 = [req.body[0].folderid]
  848. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  849. mysql.usselect(p2, async function (ret2) {
  850. let file = ret2[0]
  851. if (file.length) {
  852. let documentids = file.map(item => item.documentid);
  853. // 使用 Promise.all 确保所有添加操作都被执行
  854. await Promise.all(documentids.map(documentid =>
  855. client.collections.addDocument({
  856. id: folderid,
  857. documentId: documentid
  858. }).catch(error => {
  859. console.error("移动文件时发生错误:", error);
  860. })
  861. ));
  862. }
  863. let ids = file.map(item => item.fileid);
  864. console.log(ids);
  865. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  866. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  867. mysql.usselect(p, async function (ret) {
  868. res.end(JSON.stringify(ret));
  869. });
  870. });
  871. }
  872. });
  873. // 下载文件功能
  874. router.route("/downloadFile").all(async (req, res, next) => {
  875. var json = queryString(req.url);
  876. let params = {
  877. documentid: (json && json['documentid'] || req.body[0].documentid),
  878. filename: (json && json['filename'] || req.body[0].filename),
  879. };
  880. try {
  881. const response = await client.documents.download({ id: params.documentid });
  882. console.log(response);
  883. // 根据response的类型设置Content-Type
  884. const contentType = response.type || 'application/octet-stream';
  885. res.setHeader('Content-Type', contentType);
  886. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  887. // 直接发送blob数据
  888. res.send(Buffer.from(await response.arrayBuffer()));
  889. } catch (error) {
  890. console.error("下载文件时发生错误:", error);
  891. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  892. }
  893. });
  894. // 下载文件功能
  895. router.route("/downloadFile2").all(async (req, res, next) => {
  896. var json = queryString(req.url);
  897. let params = {
  898. documentid: (json && json['documentid'] || req.body[0].documentid),
  899. filename: (json && json['filename'] || req.body[0].filename),
  900. };
  901. try {
  902. const response = await client.documents.download({ id: params.documentid });
  903. console.log(response);
  904. // 直接返回blob数据
  905. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  906. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  907. const buffer = Buffer.from(await response.arrayBuffer());
  908. res.send(buffer); // 发送blob数据
  909. } catch (error) {
  910. console.error("下载文件时发生错误:", error);
  911. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  912. }
  913. });
  914. //召回测试
  915. router.route("/ragFolder").all(async (req, res, next) => {
  916. let reqJson = req.body[0]
  917. if (reqJson) {
  918. reqJson.query = decodeURIComponent(reqJson.query)
  919. try {
  920. let filters = {}
  921. if (reqJson.type == 1) {
  922. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  923. } else if (reqJson.type == 2) {
  924. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  925. } else if (reqJson.type == 3) {
  926. filters = { "collection_ids": { "$eq": reqJson.ids } }
  927. }
  928. console.log(111);
  929. const response = await client.retrieval.search({
  930. query: reqJson.query,
  931. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  932. // searchMode: 'basic',
  933. searchSettings: {
  934. useHybridSearch: reqJson.useType == 3,
  935. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  936. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  937. hybrid_settings: {
  938. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  939. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  940. },
  941. filters: filters,
  942. limit: reqJson.limit ? reqJson.limit : 100,
  943. },
  944. ragGenerationConfig: {
  945. stream: false,
  946. temperature: reqJson.temperature ? reqJson.temperature : 0,
  947. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  948. }
  949. });
  950. // res.status(200).json({ result: response.results });
  951. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  952. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  953. mysql.usselect(p, async function (ret) {
  954. res.status(200).json({ result: response.results });
  955. });
  956. } catch (error) {
  957. console.log(error);
  958. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  959. }
  960. }
  961. });
  962. //召回测试
  963. router.route("/searchFolder").all(async (req, res, next) => {
  964. let reqJson = req.body[0]
  965. if (reqJson) {
  966. try {
  967. let filters = {}
  968. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  969. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  970. filters = {
  971. "$or": [
  972. { "collection_ids": { "$in": folderids } },
  973. { "document_id": { "$in": documentids } }
  974. ]
  975. };
  976. let params = {
  977. query: reqJson.query,
  978. searchSettings: {
  979. useHybridSearch: reqJson.useType == 3,
  980. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  981. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  982. hybrid_settings: {
  983. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  984. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  985. },
  986. filters: filters,
  987. limit: reqJson.limit ? reqJson.limit : 100,
  988. },
  989. ragGenerationConfig: {
  990. stream: false,
  991. temperature: reqJson.temperature ? reqJson.temperature : 0,
  992. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  993. }
  994. }
  995. // console.log(params);
  996. const response = await client.retrieval.search(params);
  997. res.status(200).json({ result: response.results });
  998. } catch (error) {
  999. console.log(error);
  1000. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1001. }
  1002. }
  1003. });
  1004. //召回测试
  1005. router.route("/searchFolder2").all(async (req, res, next) => {
  1006. let reqJson = req.body
  1007. if (Object.keys(reqJson).length) {
  1008. try {
  1009. let filters = {}
  1010. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  1011. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  1012. filters = {
  1013. "$or": [
  1014. { "collection_ids": { "$in": folderids } },
  1015. { "document_id": { "$in": documentids } }
  1016. ]
  1017. };
  1018. if(!folderids.length && !documentids.length){
  1019. res.status(200).json({ result: [] });
  1020. return
  1021. }
  1022. let params = {
  1023. query: reqJson.query,
  1024. searchSettings: {
  1025. useHybridSearch: reqJson.useType == 3,
  1026. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1027. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1028. hybrid_settings: {
  1029. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1030. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1031. },
  1032. filters: filters,
  1033. limit: reqJson.limit ? reqJson.limit : 100,
  1034. },
  1035. ragGenerationConfig: {
  1036. stream: false,
  1037. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1038. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1039. }
  1040. }
  1041. // console.log(params);
  1042. const response = await client.retrieval.search(params);
  1043. res.status(200).json({ result: response.results });
  1044. } catch (error) {
  1045. console.log(error);
  1046. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1047. }
  1048. }else{
  1049. res.status(200).json({ result: "请传参" });
  1050. }
  1051. });
  1052. //召回测试
  1053. router.route("/searchFolder3").all(async (req, res, next) => {
  1054. let reqJson = req.body[0]
  1055. if (reqJson) {
  1056. reqJson.query = decodeURIComponent(reqJson.query)
  1057. try {
  1058. let filters = {}
  1059. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1060. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1061. filters = {
  1062. "$or": [
  1063. { "collection_ids": { "$in": folderids } },
  1064. { "document_id": { "$in": documentids } }
  1065. ]
  1066. };
  1067. if(!folderids.length && !documentids.length){
  1068. res.status(200).json({ result: [] });
  1069. return
  1070. }
  1071. let params = {
  1072. query: reqJson.query,
  1073. searchSettings: {
  1074. useHybridSearch: reqJson.useType == 3,
  1075. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1076. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1077. hybrid_settings: {
  1078. fullTextWeight: reqJson.fullTextWeight,
  1079. semanticWeight: reqJson.semanticWeight,
  1080. },
  1081. filters: filters,
  1082. limit: reqJson.limit ? reqJson.limit : 100,
  1083. },
  1084. ragGenerationConfig: {
  1085. stream: false,
  1086. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1087. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1088. }
  1089. }
  1090. // console.log(params);
  1091. const response = await client.retrieval.search(params);
  1092. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1093. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1094. mysql.usselect(p, async function (ret) {
  1095. res.status(200).json({ result: response.results });
  1096. });
  1097. } catch (error) {
  1098. console.log(error);
  1099. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1100. }
  1101. }
  1102. });
  1103. //召回测试
  1104. router.route("/searchFolder32").all(async (req, res, next) => {
  1105. let reqJson = req.body[0]
  1106. if (reqJson) {
  1107. reqJson.query = decodeURIComponent(reqJson.query)
  1108. try {
  1109. let filters = {}
  1110. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1111. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1112. filters = {
  1113. "$or": [
  1114. { "collection_ids": { "$in": folderids } },
  1115. { "document_id": { "$in": documentids } }
  1116. ]
  1117. };
  1118. if(!folderids.length && !documentids.length){
  1119. res.status(200).json({ result: [] });
  1120. return
  1121. }
  1122. let params = {
  1123. query: reqJson.query,
  1124. searchSettings: {
  1125. useHybridSearch: reqJson.useType == 3,
  1126. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1127. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1128. hybrid_settings: {
  1129. fullTextWeight: reqJson.fullTextWeight,
  1130. semanticWeight: reqJson.semanticWeight,
  1131. },
  1132. filters: filters,
  1133. limit: reqJson.limit ? reqJson.limit : 100,
  1134. },
  1135. ragGenerationConfig: {
  1136. stream: false,
  1137. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1138. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1139. }
  1140. }
  1141. console.log(params);
  1142. const response = await client.retrieval.search(params);
  1143. res.status(200).json({ result: response.results });
  1144. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1145. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1146. // mysql.usselect(p, async function (ret) {
  1147. // });
  1148. } catch (error) {
  1149. console.log(error);
  1150. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1151. }
  1152. }
  1153. });
  1154. //获取召回测试数据
  1155. router.route("/getRagData").all(async (req, res, next) => {
  1156. //mode的处理
  1157. if (req.body[0]) {
  1158. p = Object.values(req.body[0]);
  1159. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1160. mysql.usselect(p, async function (ret) {
  1161. res.status(200).json(ret);
  1162. });
  1163. }
  1164. });
  1165. //删除召回测试数据
  1166. router.route("/delRagData").all(async (req, res, next) => {
  1167. //mode的处理
  1168. if (req.body[0]) {
  1169. p = Object.values(req.body[0]);
  1170. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1171. mysql.usselect(p, async function (ret) {
  1172. res.status(200).json(ret);
  1173. });
  1174. }
  1175. });
  1176. function generateAPIKey(username) {
  1177. let timeStamp = new Date().getTime();
  1178. let hashStr = timeStamp + username;
  1179. let digest = crypto
  1180. .createHash("sha256")
  1181. .update(hashStr, "utf8")
  1182. .digest("hex");
  1183. return digest;
  1184. }
  1185. queryString = function (url) {
  1186. var a = url.split("?");
  1187. var json = querystring.parse(a[1]);
  1188. return json;
  1189. };
  1190. //post存儲過程處理
  1191. postmysql = function (req, res, functionname) {
  1192. //mode的處理
  1193. if (req.body[0]) {
  1194. p = Object.values(req.body[0]);
  1195. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1196. //執行存儲過程
  1197. mysql.usselect(p, function (ret) {
  1198. res.end(JSON.stringify(ret));
  1199. });
  1200. }
  1201. };
  1202. //post存儲過程處理
  1203. postmysql2 = function (req, res, functionname) {
  1204. //mode的處理
  1205. if (req.body) {
  1206. p = Object.values(req.body)[0].split(",");
  1207. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1208. //執行存儲過程
  1209. mysql.usselect(p, function (ret) {
  1210. res.end(JSON.stringify(ret));
  1211. });
  1212. }
  1213. };
  1214. //post存儲過程處理
  1215. postmysql3 = function (req, res, functionname) {
  1216. //mode的處理
  1217. if (req.body) {
  1218. console.log(req);
  1219. p = Object.values(req.body);
  1220. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1221. //執行存儲過程
  1222. mysql.usselect(p, function (ret) {
  1223. res.end(JSON.stringify(ret));
  1224. });
  1225. }
  1226. };
  1227. //get獲取數據庫信息
  1228. getmysql = function (req, res, functionname, ...p) {
  1229. p = p || [];
  1230. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1231. //執行存儲過程
  1232. mysql.usselect(p, function (ret) {
  1233. res.end(JSON.stringify(ret));
  1234. }); //
  1235. };
  1236. //get獲取數據庫信息
  1237. getmysql2 = function (req, res, functionname, ...p) {
  1238. p = p || [];
  1239. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1240. //執行存儲過程
  1241. mysql.usselect(p, function (ret) {
  1242. res.end(JSON.stringify(ret));
  1243. }); //
  1244. };
  1245. //get獲取數據庫信息
  1246. getmysql3 = function (req, res, functionname, ...p) {
  1247. p = p || [];
  1248. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1249. //執行存儲過程
  1250. mysql.usselect(p, function (ret) {
  1251. res.end(JSON.stringify(ret));
  1252. }); //
  1253. };
  1254. //post存储过程处理
  1255. userpostmysql = function (req, res, functionname) {
  1256. //mode的处理
  1257. if (req.body.mode) {
  1258. p = req.body.mode.split(",");
  1259. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1260. //执行存储过程
  1261. mysql.usselect(p, function (ret) {
  1262. res.end(JSON.stringify(ret));
  1263. });
  1264. }
  1265. };
  1266. //get获取数据库信息
  1267. usergetmysql = function (req, res, functionname, ...p) {
  1268. p = p || [];
  1269. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1270. //执行存储过程
  1271. mysql.usselect(p, function (ret) {
  1272. res.end(JSON.stringify(ret));
  1273. }); //
  1274. };
  1275. //post存儲過程處理
  1276. postmysqlLocal = function (req, res) {
  1277. //mode的處理
  1278. if (req.body[0]) {
  1279. p = Object.values(req.body[0]);
  1280. p.unshift(_localhost[0], _localhost[1]);
  1281. //執行存儲過程
  1282. mysql.usselect(p, function (ret) {
  1283. res.end(JSON.stringify(ret));
  1284. });
  1285. }
  1286. };
  1287. postmysqlOnline = function (req, res) {
  1288. //mode的處理
  1289. if (req.body[0]) {
  1290. p = Object.values(req.body[0]);
  1291. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1292. //執行存儲過程
  1293. mysql.usselect(p, function (ret) {
  1294. res.end(JSON.stringify(ret));
  1295. });
  1296. }
  1297. };
  1298. // 新增测试接口,直接返回成功
  1299. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1300. // console.log('收到的 req.body:', req.body);
  1301. // console.log('收到的 req.files:', req.files);
  1302. // console.log('收到的 headers:', req.headers);
  1303. const requestUrl = req.body.url; // 前端传入 url
  1304. if (!requestUrl) {
  1305. return res.status(400).json({error: 'Missing url'});
  1306. }
  1307. const browser = await initBrowser();
  1308. const page = await browser.newPage();
  1309. try {
  1310. console.log("开始抓取页面:", requestUrl);
  1311. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1312. await page.goto(requestUrl);
  1313. const html = await page.content();
  1314. // 新增:HTML 转 Markdown
  1315. const turndownService = new TurndownService();
  1316. // 用 cheerio 只提取正文部分
  1317. const cheerioDoc = cheerio.load(html);
  1318. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1319. // 用 cheerio 过滤无关标签
  1320. const cheerioMain = cheerio.load(mainHtml || '');
  1321. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1322. mainHtml = cheerioMain.html();
  1323. const markdownContent = turndownService.turndown(mainHtml);
  1324. // 新增:解析 <title> 标签内容
  1325. let title = '';
  1326. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1327. if (titleMatch && titleMatch[1]) {
  1328. title = titleMatch[1].trim();
  1329. }
  1330. let finalTitle;
  1331. if (title && title.trim()) {
  1332. finalTitle = title + '.md';
  1333. } else {
  1334. finalTitle = 'untitled-' + uuidv4() + '.md';
  1335. }
  1336. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1337. // 解析用户传入的参数
  1338. const {
  1339. collection_ids,
  1340. id: customId,
  1341. metadata: customMetadata,
  1342. ingestion_mode,
  1343. run_with_orchestration
  1344. } = req.body;
  1345. // 处理 metadata 字段
  1346. let finalMetadata = customMetadata;
  1347. if (typeof finalMetadata === 'string') {
  1348. try {
  1349. finalMetadata = JSON.parse(finalMetadata);
  1350. } catch (e) {
  1351. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1352. }
  1353. }
  1354. // if (!finalMetadata) {
  1355. // finalMetadata = {
  1356. // title: title || uuidv4(),
  1357. // // source_url: url,
  1358. // content_type: "markdown",
  1359. // // scraped_at: new Date().toISOString(),
  1360. // // upload_id: Date.now() + '-' + Math.random()
  1361. // };
  1362. // }
  1363. // 处理 collection_ids 字段
  1364. let finalCollectionIds = collection_ids;
  1365. if (typeof finalCollectionIds === 'string') {
  1366. try {
  1367. finalCollectionIds = JSON.parse(finalCollectionIds);
  1368. } catch (e) {
  1369. // 不是合法 JSON 字符串就当作普通字符串
  1370. }
  1371. }
  1372. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1373. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1374. finalMetadata.collection_ids = finalCollectionIds;
  1375. }
  1376. // 补全 finalMetadata.title 和 content_type
  1377. if (!finalMetadata.title) {
  1378. finalMetadata.title = finalTitle;
  1379. }
  1380. finalMetadata.content_type = 'markdown';
  1381. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1382. const FormData = require('form-data');
  1383. const fs = require('fs');
  1384. const path = require('path');
  1385. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1386. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1387. const form = new FormData();
  1388. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1389. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1390. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1391. }
  1392. if (customId) {
  1393. form.append('id', customId);
  1394. }
  1395. form.append('metadata', JSON.stringify(finalMetadata));
  1396. form.append('ingestion_mode', ingestion_mode || 'fast');
  1397. form.append('run_with_orchestration', 'false');
  1398. // 打印上传参数,方便排查问题
  1399. console.log('Request Headers:', form.getHeaders());
  1400. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1401. const response = await axios.post(url, form, {
  1402. headers: {
  1403. ...form.getHeaders(),
  1404. // 'User-Agent': 'NodeClient/1.0'
  1405. }
  1406. });
  1407. fs.unlinkSync(tmpFilePath);
  1408. result = {
  1409. results: response.data.results,
  1410. title: finalTitle
  1411. }
  1412. res.json(result);
  1413. } catch (error) {
  1414. console.error('Message:', error.message);
  1415. console.error('Stack:', error.stack);
  1416. res.status(500).json({error: error.message});
  1417. } finally {
  1418. await page.close(); // 确保页面总是被关闭
  1419. }
  1420. });
  1421. module.exports = router;