pbl.js 61 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673
  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. // const array = data.map(item => item.documentid);
  217. try {
  218. let data = decodeURIComponent(req.body[0].array).split(",");
  219. const response = await client.documents.list({ ids: data });
  220. res.status(200).json({ result: response.results });
  221. } catch (error) {
  222. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  223. }
  224. }
  225. });
  226. //获取文档列表
  227. router.route("/getFileD2").all(async (req, res, next) => {
  228. //mode的处理
  229. if (req.body) {
  230. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  231. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  232. mysql.usselect(p, async function (ret) {
  233. let data = ret[0];
  234. // const array = data.map(item => item.documentid);
  235. const array = data.map(item => item.documentid);
  236. try {
  237. const response = await client.documents.list({ ids: array });
  238. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  239. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  240. // console.log(response);
  241. let data2 = response.results.map(doc => {
  242. doc.documentid = doc.id;
  243. delete doc.id;
  244. return doc;
  245. });
  246. // console.log(data2);
  247. // 合并data和data2,根据documentid
  248. const mergedData = data.map(item => {
  249. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  250. return { ...item, ...correspondingDoc };
  251. });
  252. res.status(200).json({ result: mergedData });
  253. } catch (error) {
  254. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  255. }
  256. });
  257. }
  258. });
  259. //获取文档列表
  260. router.route("/getFileD2").all(async (req, res, next) => {
  261. //mode的处理
  262. if (req.body) {
  263. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  264. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  265. mysql.usselect(p, async function (ret) {
  266. let data = ret[0];
  267. // const array = data.map(item => item.documentid);
  268. const array = data.map(item => item.documentid);
  269. try {
  270. const response = await client.documents.list({ ids: array });
  271. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  272. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  273. // console.log(response);
  274. let data2 = response.results.map(doc => {
  275. doc.documentid = doc.id;
  276. delete doc.id;
  277. return doc;
  278. });
  279. // console.log(data2);
  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. //获取文档列表
  293. router.route("/getFileOpen").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], "getFileOpen");
  298. mysql.usselect(p, async function (ret) {
  299. let data = ret[0];
  300. let user = ret[1];
  301. // const array = data.map(item => item.documentid);
  302. const array = data.map(item => item.documentid);
  303. try {
  304. const response = await client.documents.list({ ids: array });
  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, user: user });
  316. } catch (error) {
  317. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  318. }
  319. });
  320. }
  321. });
  322. //获取文档列表2
  323. router.route("/getFileP").all(async (req, res, next) => {
  324. //mode的处理
  325. if (req.body) {
  326. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  327. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  328. mysql.usselect(p, async function (ret) {
  329. let data = ret[0];
  330. // const array = data.map(item => item.documentid);
  331. try {
  332. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  333. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  334. // console.log(response);
  335. let data2 = response.results.map(doc => {
  336. doc.documentid = doc.id;
  337. delete doc.id;
  338. return doc;
  339. });
  340. // 合并data和data2,根据documentid
  341. const mergedData = data.map(item => {
  342. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  343. return { ...item, ...correspondingDoc };
  344. });
  345. res.status(200).json({ result: mergedData });
  346. } catch (error) {
  347. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  348. }
  349. });
  350. }
  351. });
  352. //获取能关联的文件
  353. router.route("/getFile2").all(async (req, res, next) => {
  354. //mode的处理
  355. if (req.body[0]) {
  356. p = Object.values(req.body[0]);
  357. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  358. mysql.usselect(p, async function (ret) {
  359. let data = ret[0];
  360. const array = data.map(item => item.documentid);
  361. try {
  362. const response = await client.documents.list({ ids: array });
  363. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  364. // console.log(response);
  365. let data2 = response.results.map(doc => {
  366. doc.documentid = doc.id;
  367. delete doc.id;
  368. return doc;
  369. });
  370. // 合并data和data2,根据documentid
  371. const mergedData = data.map(item => {
  372. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  373. return { ...item, ...correspondingDoc };
  374. });
  375. res.status(200).json({ result: mergedData });
  376. } catch (error) {
  377. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  378. }
  379. });
  380. }
  381. });
  382. //添加文件夹
  383. router.route("/addFolder").all(async (req, res, next) => {
  384. //mode的处理
  385. if (req.body[0]) {
  386. try {
  387. const response = await client.collections.create({
  388. name: decodeURIComponent(req.body[0].n),
  389. description: decodeURIComponent(req.body[0].fd)
  390. });
  391. p = [...Object.values(req.body[0]), response.results.id];
  392. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  393. mysql.usselect(p, async function (ret) {
  394. res.end(JSON.stringify(ret));
  395. });
  396. } catch (error) {
  397. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  398. }
  399. }
  400. });
  401. //添加文件夹
  402. router.route("/addFolderP").all(async (req, res, next) => {
  403. //mode的处理
  404. if (req.body) {
  405. try {
  406. const response = await client.collections.create({
  407. name: decodeURIComponent(req.body.n),
  408. description: decodeURIComponent(req.body.fd)
  409. });
  410. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  411. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  412. mysql.usselect(p, async function (ret) {
  413. res.end(JSON.stringify(ret));
  414. });
  415. } catch (error) {
  416. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  417. }
  418. }
  419. });
  420. //删除文件夹
  421. router.route("/deleteFolder").all(async (req, res, next) => {
  422. //mode的处理
  423. if (req.body[0]) {
  424. try {
  425. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  426. const ids = req.body[0].ids;
  427. // 并行处理文件夹下的文件删除和文件夹删除
  428. await Promise.all(folderids.map(async folderid => {
  429. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  430. await Promise.all(files.results.map(file =>
  431. client.collections.removeDocument({ id: folderid, documentId: file.id })
  432. ));
  433. }));
  434. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  435. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  436. mysql.usselect(params, async function (ret) {
  437. res.end(JSON.stringify(ret));
  438. });
  439. } catch (error) {
  440. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  441. }
  442. }
  443. });
  444. //修改文件夹
  445. router.route("/updateFolder").all(async (req, res, next) => {
  446. //mode的处理
  447. if (req.body[0]) {
  448. try {
  449. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  450. const response = await client.collections.update({
  451. id: folderid,
  452. name: decodeURIComponent(req.body[0].n),
  453. description: decodeURIComponent(req.body[0].d)
  454. });
  455. p = Object.values(folderData); // 获取剩余属性的键
  456. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  457. mysql.usselect(p, async function (ret) {
  458. res.end(JSON.stringify(ret));
  459. });
  460. } catch (error) {
  461. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  462. }
  463. }
  464. });
  465. //修改文件夹
  466. router.route("/updateFolder2").all(async (req, res, next) => {
  467. //mode的处理
  468. if (req.body[0]) {
  469. try {
  470. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  471. const response = await client.collections.update({
  472. id: folderid,
  473. name: decodeURIComponent(req.body[0].n),
  474. description: decodeURIComponent(req.body[0].d)
  475. });
  476. p = Object.values(folderData); // 获取剩余属性的键
  477. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  478. mysql.usselect(p, async function (ret) {
  479. res.end(JSON.stringify(ret));
  480. });
  481. } catch (error) {
  482. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  483. }
  484. }
  485. });
  486. //移除文件
  487. router.route("/removeFile").all(async (req, res, next) => {
  488. if (req.body[0]) {
  489. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  490. let folderid = req.body[0].folderid;
  491. let ids = req.body[0].ids;
  492. // 使用 Promise.all 确保所有删除操作都被执行
  493. const removePromises = documentids.map(documentid =>
  494. client.collections.removeDocument({ id: folderid, documentId: documentid })
  495. );
  496. // 不管client.documents报错没有也要执行removeFile
  497. Promise.all(removePromises).catch(error => {
  498. console.error("移除文件时发生错误:", error);
  499. });
  500. p = [ids, folderid];
  501. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  502. mysql.usselect(p, async function (ret) {
  503. res.end(JSON.stringify(ret));
  504. });
  505. }
  506. });
  507. //移入文件
  508. router.route("/moveFile").all(async (req, res, next) => {
  509. if (req.body[0]) {
  510. try {
  511. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  512. let folderid = req.body[0].folderid
  513. let ids = decodeURIComponent(req.body[0].ids).split(",")
  514. // 使用 Promise.all 确保所有添加操作都被执行
  515. await Promise.all(documentids.map(documentid =>
  516. client.collections.addDocument({
  517. id: folderid,
  518. documentId: documentid
  519. }).catch(error => {
  520. console.error("移动文件时发生错误:", error);
  521. })
  522. ));
  523. for (const id of ids) {
  524. p = [id, folderid];
  525. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  526. await mysql.usselect(p, async function (ret) { });
  527. }
  528. res.status(200).json({ messages: "移动成功" });
  529. } catch (error) {
  530. // 这里不返回错误,确保deleteFile执行完
  531. console.error("处理过程中发生错误:", error);
  532. }
  533. }
  534. });
  535. //删除文件
  536. router.route("/deleteFile").all(async (req, res, next) => {
  537. if (req.body[0]) {
  538. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  539. let ids = req.body[0].ids;
  540. // 使用 Promise.all 确保所有删除操作都被执行
  541. Promise.all(documentids.map(documentid =>
  542. client.documents.delete({ id: documentid }).catch(error => {
  543. console.error("删除文件时发生错误:", error);
  544. })
  545. ));
  546. p = [ids];
  547. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  548. mysql.usselect(p, async function (ret) {
  549. res.end(JSON.stringify(ret));
  550. });
  551. }
  552. });
  553. //删除文件
  554. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  555. if (req.body[0]) {
  556. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  557. let ids = req.body[0].ids;
  558. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  559. console.log("文件删除成功");
  560. }).catch(error => {
  561. console.error("删除文件时发生错误:", error);
  562. })
  563. p = [ids];
  564. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  565. mysql.usselect(p, async function (ret) {
  566. res.end(JSON.stringify(ret));
  567. });
  568. }
  569. });
  570. router.route("/getFolder").all((req, res, next) => {
  571. getmysql2(req, res, "getFolder", req.body.uid);
  572. });
  573. router.route("/getFolder2").all((req, res, next) => {
  574. var json = queryString(req.url);
  575. getmysql2(req, res, "getFolder", json['uid']);
  576. });
  577. router.route("/getFolderPage").all((req, res, next) => {
  578. 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);
  579. });
  580. router.route("/getFolderPage3").all((req, res, next) => {
  581. 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);
  582. });
  583. router.route("/getFolderPage32").all((req, res, next) => {
  584. var json = queryString(req.url);
  585. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  586. });
  587. router.route("/getFolderPageOpen").all((req, res, next) => {
  588. 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);
  589. });
  590. router.route("/haveFile").all((req, res, next) => {
  591. postmysql(req, res, "haveFile");
  592. });
  593. router.route("/haveFolder").all((req, res, next) => {
  594. postmysql(req, res, "haveFolder");
  595. });
  596. //search文件
  597. router.route("/searchFile").all(async (req, res, next) => {
  598. if (req.body) {
  599. try {
  600. let filters = {}
  601. if (req.body.type == 1) {
  602. filters = { "document_id": { "$in": req.body.ids } }
  603. } else if (req.body.type == 2) {
  604. filters = { "collection_ids": { "$in": req.body.ids } }
  605. }
  606. const response = await client.retrieval.search({
  607. query: req.body.query,
  608. search_settings: {
  609. filters: filters,
  610. useSemanticSearch: true,
  611. chunkSettings: {
  612. limit: req.body.limit ? req.body.limit : 10,
  613. enabled: true
  614. },
  615. }
  616. });
  617. res.status(200).json({ result: response.results });
  618. } catch (error) {
  619. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  620. }
  621. }
  622. });
  623. //提取文件实体
  624. router.route("/extractFile").all(async (req, res, next) => {
  625. if (req.body) {
  626. try {
  627. const response = client.documents.extract({
  628. id: req.body.id
  629. })
  630. res.status(200).json({ result: response.results });
  631. } catch (error) {
  632. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  633. }
  634. }
  635. });
  636. //添加上传好的知识库文件
  637. router.route("/addFile").all((req, res, next) => {
  638. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  639. })
  640. //获取文件详情
  641. router.route("/getFileDetail").all(async (req, res, next) => {
  642. if (req.body[0]) {
  643. p = Object.values(req.body[0]);
  644. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  645. mysql.usselect(p, async function (ret) {
  646. let data = ret[0][0];
  647. // const array = data.map(item => item.documentid);
  648. try {
  649. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  650. // console.log(response);
  651. let data2 = response.results[0]
  652. // 合并data和data2,根据documentid
  653. const mergedData = { ...data, ...data2 };
  654. res.status(200).json({ result: mergedData });
  655. } catch (error) {
  656. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  657. }
  658. });
  659. }
  660. });
  661. //获取切片
  662. router.route("/getChunks").all(async (req, res, next) => {
  663. if (req.body[0]) {
  664. try {
  665. const { page = 1, limit = 10, documentId } = req.body[0];
  666. const offset = (page - 1) * limit;
  667. const response = await client.documents.listChunks({
  668. id: documentId,
  669. includeVectors: false,
  670. offset,
  671. limit,
  672. });
  673. res.status(200).json({ result: response });
  674. } catch (error) {
  675. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  676. }
  677. }
  678. });
  679. //获取实体
  680. router.route("/getEntities").all(async (req, res, next) => {
  681. if (req.body[0]) {
  682. try {
  683. const { page = 1, limit = 10, documentId } = req.body[0];
  684. const offset = (page - 1) * limit;
  685. const response = await client.documents.listEntities({
  686. id: documentId,
  687. includeVectors: false,
  688. offset,
  689. limit,
  690. });
  691. res.status(200).json({ result: response });
  692. } catch (error) {
  693. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  694. }
  695. }
  696. });
  697. //获取关系
  698. router.route("/getRelationships").all(async (req, res, next) => {
  699. if (req.body[0]) {
  700. try {
  701. const { page = 1, limit = 10, documentId } = req.body[0];
  702. const offset = (page - 1) * limit;
  703. const response = await client.documents.listRelationships({
  704. id: documentId,
  705. includeVectors: false,
  706. offset,
  707. limit,
  708. });
  709. res.status(200).json({ result: response });
  710. } catch (error) {
  711. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  712. }
  713. }
  714. });
  715. //提取整个文件夹的关系
  716. router.route("/resetGraphs").all(async (req, res, next) => {
  717. if (req.body[0]) {
  718. try {
  719. // const response = await client.graphs.pull({
  720. // collectionId: req.body[0].collectionId,
  721. // });
  722. const response = await client.graphs.reset({
  723. collectionId: req.body[0].collectionId,
  724. });
  725. res.status(200).json({ result: response });
  726. } catch (error) {
  727. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  728. }
  729. }
  730. });
  731. //获取整个文件夹的关系
  732. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  733. if (req.body[0]) {
  734. try {
  735. await client.graphs.pull({
  736. collectionId: req.body[0].collectionId,
  737. });
  738. const response = await client.graphs.listRelationships({
  739. collectionId: req.body[0].collectionId,
  740. offset: req.body[0].page ? req.body[0].page : 0,
  741. limit: 1000
  742. });
  743. res.status(200).json({ result: response });
  744. } catch (error) {
  745. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  746. }
  747. }
  748. });
  749. //提取知识图谱
  750. router.route("/getGraphs").all(async (req, res, next) => {
  751. console.log(req.body);
  752. //mode的处理
  753. if (req.body[0]) {
  754. try {
  755. //创建临时文件夹
  756. const folder = await client.collections.create({
  757. name: "临时文件夹",
  758. description: "临时文件夹"
  759. });
  760. let folderid = folder.results.id
  761. console.log(folder.results.id);
  762. //将要提取的文件添加到临时文件夹
  763. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  764. // 使用 Promise.all 确保所有添加操作都被执行
  765. await Promise.all(documentids.map(documentid =>
  766. client.collections.addDocument({
  767. id: folderid,
  768. documentId: documentid
  769. }).catch(error => {
  770. console.error("移动文件时发生错误:", error);
  771. })
  772. ));
  773. //将临时文件夹提取知识图谱
  774. await client.graphs.pull({
  775. collectionId: folderid
  776. });
  777. let offset = 0;
  778. let results = [];
  779. let totalEntries = 0;
  780. do {
  781. const response = await client.graphs.listRelationships({
  782. collectionId: folderid,
  783. offset: offset,
  784. limit: 1000
  785. });
  786. results = results.concat(response.results);
  787. totalEntries = response.totalEntries;
  788. offset += 1000;
  789. } while (offset < totalEntries);
  790. //移除临时文件夹的文件
  791. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  792. await Promise.all(files.results.map(file =>
  793. client.collections.removeDocument({ id: folderid, documentId: file.id })
  794. ));
  795. //删除临时文件夹
  796. await client.collections.delete({ id: folderid });
  797. res.status(200).json({ result: results });
  798. } catch (error) {
  799. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  800. }
  801. }
  802. });
  803. //提取知识图谱
  804. router.route("/getGraphs2").all(async (req, res, next) => {
  805. console.log(req.body);
  806. //mode的处理
  807. if (req.body[0]) {
  808. try {
  809. let documentIds = decodeURIComponent(req.body[0].fileids).split(",");
  810. // 使用 Promise.all 并通过 map 返回新的 results 数组
  811. let results = await Promise.all(documentIds.map(async documentId => {
  812. const response = await client.documents.listRelationships({
  813. id: documentId,
  814. includeVectors: false,
  815. limit: 1000, // 修正这里的参数
  816. });
  817. return response.results;
  818. }));
  819. // 使用 flat() 合并所有数组
  820. results = results.flat();
  821. res.status(200).json({ result: results });
  822. } catch (error) {
  823. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  824. }
  825. }
  826. });
  827. //获取实体和关系
  828. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  829. if (req.body[0]) {
  830. try {
  831. const { documentId } = req.body[0];
  832. const offset = (page - 1) * limit;
  833. const response = await client.documents.listEntities({
  834. id: documentId,
  835. includeVectors: false,
  836. });
  837. const response2 = await client.documents.listRelationships({
  838. id: documentId,
  839. includeVectors: false,
  840. });
  841. res.status(200).json({ Entities: response, Relationships: response2 });
  842. } catch (error) {
  843. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  844. }
  845. }
  846. });
  847. //提取功能
  848. router.route("/extractFile2").all(async (req, res, next) => {
  849. try {
  850. let setting = {}
  851. if(req.body[0].prompt){
  852. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  853. }
  854. console.log(setting);
  855. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  856. console.log(response);
  857. res.status(200).json({ result: response.results });
  858. } catch (error) {
  859. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  860. }
  861. });
  862. //复制文件夹
  863. router.route("/copyFolder").all(async (req, res, next) => {
  864. if (req.body[0]) {
  865. const response = await client.collections.create({
  866. name: decodeURIComponent(req.body[0].n),
  867. description: decodeURIComponent(req.body[0].fd)
  868. });
  869. let folderid = response.results.id;
  870. p2 = [req.body[0].folderid]
  871. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  872. mysql.usselect(p2, async function (ret2) {
  873. let file = ret2[0]
  874. if (file.length) {
  875. let documentids = file.map(item => item.documentid);
  876. // 使用 Promise.all 确保所有添加操作都被执行
  877. await Promise.all(documentids.map(documentid =>
  878. client.collections.addDocument({
  879. id: folderid,
  880. documentId: documentid
  881. }).catch(error => {
  882. console.error("移动文件时发生错误:", error);
  883. })
  884. ));
  885. }
  886. let ids = file.map(item => item.fileid);
  887. console.log(ids);
  888. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  889. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  890. mysql.usselect(p, async function (ret) {
  891. res.end(JSON.stringify(ret));
  892. });
  893. });
  894. }
  895. });
  896. // 下载文件功能
  897. router.route("/downloadFile").all(async (req, res, next) => {
  898. var json = queryString(req.url);
  899. let params = {
  900. documentid: (json && json['documentid'] || req.body[0].documentid),
  901. filename: (json && json['filename'] || req.body[0].filename),
  902. };
  903. try {
  904. const response = await client.documents.download({ id: params.documentid });
  905. console.log(response);
  906. // 根据response的类型设置Content-Type
  907. const contentType = response.type || 'application/octet-stream';
  908. res.setHeader('Content-Type', contentType);
  909. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  910. // 直接发送blob数据
  911. res.send(Buffer.from(await response.arrayBuffer()));
  912. } catch (error) {
  913. console.error("下载文件时发生错误:", error);
  914. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  915. }
  916. });
  917. // 下载文件功能
  918. router.route("/downloadFile2").all(async (req, res, next) => {
  919. var json = queryString(req.url);
  920. let params = {
  921. documentid: (json && json['documentid'] || req.body[0].documentid),
  922. filename: (json && json['filename'] || req.body[0].filename),
  923. };
  924. try {
  925. const response = await client.documents.download({ id: params.documentid });
  926. console.log(response);
  927. // 直接返回blob数据
  928. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  929. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  930. const buffer = Buffer.from(await response.arrayBuffer());
  931. res.send(buffer); // 发送blob数据
  932. } catch (error) {
  933. console.error("下载文件时发生错误:", error);
  934. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  935. }
  936. });
  937. //召回测试
  938. router.route("/ragFolder").all(async (req, res, next) => {
  939. let reqJson = req.body[0]
  940. if (reqJson) {
  941. reqJson.query = decodeURIComponent(reqJson.query)
  942. try {
  943. let filters = {}
  944. if (reqJson.type == 1) {
  945. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  946. } else if (reqJson.type == 2) {
  947. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  948. } else if (reqJson.type == 3) {
  949. filters = { "collection_ids": { "$eq": reqJson.ids } }
  950. }
  951. console.log(111);
  952. const response = await client.retrieval.search({
  953. query: reqJson.query,
  954. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  955. // searchMode: 'basic',
  956. searchSettings: {
  957. useHybridSearch: reqJson.useType == 3,
  958. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  959. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  960. hybrid_settings: {
  961. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  962. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  963. },
  964. filters: filters,
  965. limit: reqJson.limit ? reqJson.limit : 100,
  966. },
  967. ragGenerationConfig: {
  968. stream: false,
  969. temperature: reqJson.temperature ? reqJson.temperature : 0,
  970. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  971. }
  972. });
  973. // res.status(200).json({ result: response.results });
  974. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  975. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  976. mysql.usselect(p, async function (ret) {
  977. res.status(200).json({ result: response.results });
  978. });
  979. } catch (error) {
  980. console.log(error);
  981. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  982. }
  983. }
  984. });
  985. //召回测试
  986. router.route("/searchFolder").all(async (req, res, next) => {
  987. let reqJson = req.body[0]
  988. if (reqJson) {
  989. try {
  990. let filters = {}
  991. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  992. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  993. filters = {
  994. "$or": [
  995. { "collection_ids": { "$in": folderids } },
  996. { "document_id": { "$in": documentids } }
  997. ]
  998. };
  999. let params = {
  1000. query: reqJson.query,
  1001. searchSettings: {
  1002. useHybridSearch: reqJson.useType == 3,
  1003. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1004. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1005. hybrid_settings: {
  1006. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1007. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1008. },
  1009. filters: filters,
  1010. limit: reqJson.limit ? reqJson.limit : 100,
  1011. },
  1012. ragGenerationConfig: {
  1013. stream: false,
  1014. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1015. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1016. }
  1017. }
  1018. // console.log(params);
  1019. const response = await client.retrieval.search(params);
  1020. res.status(200).json({ result: response.results });
  1021. } catch (error) {
  1022. console.log(error);
  1023. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1024. }
  1025. }
  1026. });
  1027. //召回测试
  1028. router.route("/searchFolder2").all(async (req, res, next) => {
  1029. let reqJson = req.body
  1030. if (Object.keys(reqJson).length) {
  1031. try {
  1032. let filters = {}
  1033. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  1034. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  1035. filters = {
  1036. "$or": [
  1037. { "collection_ids": { "$in": folderids } },
  1038. { "document_id": { "$in": documentids } }
  1039. ]
  1040. };
  1041. if(!folderids.length && !documentids.length){
  1042. res.status(200).json({ result: [] });
  1043. return
  1044. }
  1045. let params = {
  1046. query: reqJson.query,
  1047. searchSettings: {
  1048. useHybridSearch: reqJson.useType == 3,
  1049. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1050. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1051. hybrid_settings: {
  1052. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1053. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1054. },
  1055. filters: filters,
  1056. limit: reqJson.limit ? reqJson.limit : 100,
  1057. },
  1058. ragGenerationConfig: {
  1059. stream: false,
  1060. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1061. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1062. }
  1063. }
  1064. // console.log(params);
  1065. const response = await client.retrieval.search(params);
  1066. res.status(200).json({ result: response.results });
  1067. } catch (error) {
  1068. console.log(error);
  1069. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1070. }
  1071. }else{
  1072. res.status(200).json({ result: "请传参" });
  1073. }
  1074. });
  1075. //召回测试
  1076. router.route("/searchFolder3").all(async (req, res, next) => {
  1077. let reqJson = req.body[0]
  1078. if (reqJson) {
  1079. reqJson.query = decodeURIComponent(reqJson.query)
  1080. try {
  1081. let filters = {}
  1082. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1083. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1084. filters = {
  1085. "$or": [
  1086. { "collection_ids": { "$in": folderids } },
  1087. { "document_id": { "$in": documentids } }
  1088. ]
  1089. };
  1090. if(!folderids.length && !documentids.length){
  1091. res.status(200).json({ result: [] });
  1092. return
  1093. }
  1094. let params = {
  1095. query: reqJson.query,
  1096. searchSettings: {
  1097. useHybridSearch: reqJson.useType == 3,
  1098. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1099. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1100. hybrid_settings: {
  1101. fullTextWeight: reqJson.fullTextWeight,
  1102. semanticWeight: reqJson.semanticWeight,
  1103. },
  1104. filters: filters,
  1105. limit: reqJson.limit ? reqJson.limit : 100,
  1106. },
  1107. ragGenerationConfig: {
  1108. stream: false,
  1109. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1110. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1111. }
  1112. }
  1113. // console.log(params);
  1114. const response = await client.retrieval.search(params);
  1115. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1116. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1117. mysql.usselect(p, async function (ret) {
  1118. res.status(200).json({ result: response.results });
  1119. });
  1120. } catch (error) {
  1121. console.log(error);
  1122. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1123. }
  1124. }
  1125. });
  1126. //召回测试
  1127. router.route("/searchFolder32").all(async (req, res, next) => {
  1128. let reqJson = req.body[0]
  1129. if (reqJson) {
  1130. reqJson.query = decodeURIComponent(reqJson.query)
  1131. try {
  1132. let filters = {}
  1133. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1134. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1135. filters = {
  1136. "$or": [
  1137. { "collection_ids": { "$in": folderids } },
  1138. { "document_id": { "$in": documentids } }
  1139. ]
  1140. };
  1141. if(!folderids.length && !documentids.length){
  1142. res.status(200).json({ result: [] });
  1143. return
  1144. }
  1145. let params = {
  1146. query: reqJson.query,
  1147. searchSettings: {
  1148. useHybridSearch: reqJson.useType == 3,
  1149. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1150. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1151. hybrid_settings: {
  1152. fullTextWeight: reqJson.fullTextWeight,
  1153. semanticWeight: reqJson.semanticWeight,
  1154. },
  1155. filters: filters,
  1156. limit: reqJson.limit ? reqJson.limit : 100,
  1157. },
  1158. ragGenerationConfig: {
  1159. stream: false,
  1160. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1161. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1162. }
  1163. }
  1164. console.log(params);
  1165. const response = await client.retrieval.search(params);
  1166. res.status(200).json({ result: response.results });
  1167. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1168. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1169. // mysql.usselect(p, async function (ret) {
  1170. // });
  1171. } catch (error) {
  1172. console.log(error);
  1173. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1174. }
  1175. }
  1176. });
  1177. //获取召回测试数据
  1178. router.route("/getRagData").all(async (req, res, next) => {
  1179. //mode的处理
  1180. if (req.body[0]) {
  1181. p = Object.values(req.body[0]);
  1182. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1183. mysql.usselect(p, async function (ret) {
  1184. res.status(200).json(ret);
  1185. });
  1186. }
  1187. });
  1188. //删除召回测试数据
  1189. router.route("/delRagData").all(async (req, res, next) => {
  1190. //mode的处理
  1191. if (req.body[0]) {
  1192. p = Object.values(req.body[0]);
  1193. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1194. mysql.usselect(p, async function (ret) {
  1195. res.status(200).json(ret);
  1196. });
  1197. }
  1198. });
  1199. function generateAPIKey(username) {
  1200. let timeStamp = new Date().getTime();
  1201. let hashStr = timeStamp + username;
  1202. let digest = crypto
  1203. .createHash("sha256")
  1204. .update(hashStr, "utf8")
  1205. .digest("hex");
  1206. return digest;
  1207. }
  1208. queryString = function (url) {
  1209. var a = url.split("?");
  1210. var json = querystring.parse(a[1]);
  1211. return json;
  1212. };
  1213. //post存儲過程處理
  1214. postmysql = function (req, res, functionname) {
  1215. //mode的處理
  1216. if (req.body[0]) {
  1217. p = Object.values(req.body[0]);
  1218. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1219. //執行存儲過程
  1220. mysql.usselect(p, function (ret) {
  1221. res.end(JSON.stringify(ret));
  1222. });
  1223. }
  1224. };
  1225. //post存儲過程處理
  1226. postmysql2 = function (req, res, functionname) {
  1227. //mode的處理
  1228. if (req.body) {
  1229. p = Object.values(req.body)[0].split(",");
  1230. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1231. //執行存儲過程
  1232. mysql.usselect(p, function (ret) {
  1233. res.end(JSON.stringify(ret));
  1234. });
  1235. }
  1236. };
  1237. //post存儲過程處理
  1238. postmysql3 = function (req, res, functionname) {
  1239. //mode的處理
  1240. if (req.body) {
  1241. console.log(req);
  1242. p = Object.values(req.body);
  1243. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1244. //執行存儲過程
  1245. mysql.usselect(p, function (ret) {
  1246. res.end(JSON.stringify(ret));
  1247. });
  1248. }
  1249. };
  1250. //get獲取數據庫信息
  1251. getmysql = function (req, res, functionname, ...p) {
  1252. p = p || [];
  1253. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1254. //執行存儲過程
  1255. mysql.usselect(p, function (ret) {
  1256. res.end(JSON.stringify(ret));
  1257. }); //
  1258. };
  1259. //get獲取數據庫信息
  1260. getmysql2 = function (req, res, functionname, ...p) {
  1261. p = p || [];
  1262. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1263. //執行存儲過程
  1264. mysql.usselect(p, function (ret) {
  1265. res.end(JSON.stringify(ret));
  1266. }); //
  1267. };
  1268. //get獲取數據庫信息
  1269. getmysql3 = function (req, res, functionname, ...p) {
  1270. p = p || [];
  1271. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1272. //執行存儲過程
  1273. mysql.usselect(p, function (ret) {
  1274. res.end(JSON.stringify(ret));
  1275. }); //
  1276. };
  1277. //post存储过程处理
  1278. userpostmysql = function (req, res, functionname) {
  1279. //mode的处理
  1280. if (req.body.mode) {
  1281. p = req.body.mode.split(",");
  1282. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1283. //执行存储过程
  1284. mysql.usselect(p, function (ret) {
  1285. res.end(JSON.stringify(ret));
  1286. });
  1287. }
  1288. };
  1289. //get获取数据库信息
  1290. usergetmysql = function (req, res, functionname, ...p) {
  1291. p = p || [];
  1292. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1293. //执行存储过程
  1294. mysql.usselect(p, function (ret) {
  1295. res.end(JSON.stringify(ret));
  1296. }); //
  1297. };
  1298. //post存儲過程處理
  1299. postmysqlLocal = function (req, res) {
  1300. //mode的處理
  1301. if (req.body[0]) {
  1302. p = Object.values(req.body[0]);
  1303. p.unshift(_localhost[0], _localhost[1]);
  1304. //執行存儲過程
  1305. mysql.usselect(p, function (ret) {
  1306. res.end(JSON.stringify(ret));
  1307. });
  1308. }
  1309. };
  1310. postmysqlOnline = function (req, res) {
  1311. //mode的處理
  1312. if (req.body[0]) {
  1313. p = Object.values(req.body[0]);
  1314. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1315. //執行存儲過程
  1316. mysql.usselect(p, function (ret) {
  1317. res.end(JSON.stringify(ret));
  1318. });
  1319. }
  1320. };
  1321. // 新增测试接口,直接返回成功
  1322. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1323. // console.log('收到的 req.body:', req.body);
  1324. // console.log('收到的 req.files:', req.files);
  1325. // console.log('收到的 headers:', req.headers);
  1326. const requestUrl = req.body.url; // 前端传入 url
  1327. if (!requestUrl) {
  1328. return res.status(400).json({error: 'Missing url'});
  1329. }
  1330. const browser = await initBrowser();
  1331. const page = await browser.newPage();
  1332. try {
  1333. console.log("开始抓取页面:", requestUrl);
  1334. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1335. await page.goto(requestUrl);
  1336. const html = await page.content();
  1337. // 新增:HTML 转 Markdown
  1338. const turndownService = new TurndownService();
  1339. // 用 cheerio 只提取正文部分
  1340. const cheerioDoc = cheerio.load(html);
  1341. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1342. // 用 cheerio 过滤无关标签
  1343. const cheerioMain = cheerio.load(mainHtml || '');
  1344. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1345. mainHtml = cheerioMain.html();
  1346. const markdownContent = turndownService.turndown(mainHtml);
  1347. // 新增:解析 <title> 标签内容
  1348. let title = '';
  1349. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1350. if (titleMatch && titleMatch[1]) {
  1351. title = titleMatch[1].trim();
  1352. }
  1353. let finalTitle;
  1354. if (title && title.trim()) {
  1355. finalTitle = title + '.md';
  1356. } else {
  1357. finalTitle = 'untitled-' + uuidv4() + '.md';
  1358. }
  1359. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1360. // 解析用户传入的参数
  1361. const {
  1362. collection_ids,
  1363. id: customId,
  1364. metadata: customMetadata,
  1365. ingestion_mode,
  1366. run_with_orchestration
  1367. } = req.body;
  1368. // 处理 metadata 字段
  1369. let finalMetadata = customMetadata;
  1370. if (typeof finalMetadata === 'string') {
  1371. try {
  1372. finalMetadata = JSON.parse(finalMetadata);
  1373. } catch (e) {
  1374. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1375. }
  1376. }
  1377. // if (!finalMetadata) {
  1378. // finalMetadata = {
  1379. // title: title || uuidv4(),
  1380. // // source_url: url,
  1381. // content_type: "markdown",
  1382. // // scraped_at: new Date().toISOString(),
  1383. // // upload_id: Date.now() + '-' + Math.random()
  1384. // };
  1385. // }
  1386. // 处理 collection_ids 字段
  1387. let finalCollectionIds = collection_ids;
  1388. if (typeof finalCollectionIds === 'string') {
  1389. try {
  1390. finalCollectionIds = JSON.parse(finalCollectionIds);
  1391. } catch (e) {
  1392. // 不是合法 JSON 字符串就当作普通字符串
  1393. }
  1394. }
  1395. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1396. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1397. finalMetadata.collection_ids = finalCollectionIds;
  1398. }
  1399. // 补全 finalMetadata.title 和 content_type
  1400. if (!finalMetadata.title) {
  1401. finalMetadata.title = finalTitle;
  1402. }
  1403. finalMetadata.content_type = 'markdown';
  1404. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1405. const FormData = require('form-data');
  1406. const fs = require('fs');
  1407. const path = require('path');
  1408. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1409. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1410. const form = new FormData();
  1411. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1412. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1413. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1414. }
  1415. if (customId) {
  1416. form.append('id', customId);
  1417. }
  1418. form.append('metadata', JSON.stringify(finalMetadata));
  1419. form.append('ingestion_mode', ingestion_mode || 'fast');
  1420. form.append('run_with_orchestration', 'false');
  1421. // 打印上传参数,方便排查问题
  1422. console.log('Request Headers:', form.getHeaders());
  1423. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1424. const response = await axios.post(url, form, {
  1425. headers: {
  1426. ...form.getHeaders(),
  1427. // 'User-Agent': 'NodeClient/1.0'
  1428. }
  1429. });
  1430. fs.unlinkSync(tmpFilePath);
  1431. result = {
  1432. results: response.data.results,
  1433. title: finalTitle
  1434. }
  1435. res.json(result);
  1436. } catch (error) {
  1437. console.error('Message:', error.message);
  1438. console.error('Stack:', error.stack);
  1439. res.status(500).json({error: error.message});
  1440. } finally {
  1441. await page.close(); // 确保页面总是被关闭
  1442. }
  1443. });
  1444. module.exports = router;