pbl.js 63 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739
  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. //修改文件夹
  487. router.route("/updateFolder3").all(async (req, res, next) => {
  488. //mode的处理
  489. if (req.body[0]) {
  490. try {
  491. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  492. const response = await client.collections.update({
  493. id: folderid,
  494. name: decodeURIComponent(req.body[0].n),
  495. description: decodeURIComponent(req.body[0].d)
  496. });
  497. p = Object.values(folderData); // 获取剩余属性的键
  498. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder4");
  499. mysql.usselect(p, async function (ret) {
  500. res.end(JSON.stringify(ret));
  501. });
  502. } catch (error) {
  503. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  504. }
  505. }
  506. });
  507. //移除文件
  508. router.route("/removeFile").all(async (req, res, next) => {
  509. if (req.body[0]) {
  510. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  511. let folderid = req.body[0].folderid;
  512. let ids = req.body[0].ids;
  513. // 使用 Promise.all 确保所有删除操作都被执行
  514. const removePromises = documentids.map(documentid =>
  515. client.collections.removeDocument({ id: folderid, documentId: documentid })
  516. );
  517. // 不管client.documents报错没有也要执行removeFile
  518. Promise.all(removePromises).catch(error => {
  519. console.error("移除文件时发生错误:", error);
  520. });
  521. p = [ids, folderid];
  522. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  523. mysql.usselect(p, async function (ret) {
  524. res.end(JSON.stringify(ret));
  525. });
  526. }
  527. });
  528. //移入文件
  529. router.route("/moveFile").all(async (req, res, next) => {
  530. if (req.body[0]) {
  531. try {
  532. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  533. let folderid = req.body[0].folderid
  534. let ids = decodeURIComponent(req.body[0].ids).split(",")
  535. // 使用 Promise.all 确保所有添加操作都被执行
  536. await Promise.all(documentids.map(documentid =>
  537. client.collections.addDocument({
  538. id: folderid,
  539. documentId: documentid
  540. }).catch(error => {
  541. console.error("移动文件时发生错误:", error);
  542. })
  543. ));
  544. for (const id of ids) {
  545. p = [id, folderid];
  546. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  547. await mysql.usselect(p, async function (ret) { });
  548. }
  549. res.status(200).json({ messages: "移动成功" });
  550. } catch (error) {
  551. // 这里不返回错误,确保deleteFile执行完
  552. console.error("处理过程中发生错误:", error);
  553. }
  554. }
  555. });
  556. //删除文件
  557. router.route("/deleteFile").all(async (req, res, next) => {
  558. if (req.body[0]) {
  559. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  560. let ids = req.body[0].ids;
  561. // 使用 Promise.all 确保所有删除操作都被执行
  562. Promise.all(documentids.map(documentid =>
  563. client.documents.delete({ id: documentid }).catch(error => {
  564. console.error("删除文件时发生错误:", error);
  565. })
  566. ));
  567. p = [ids];
  568. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  569. mysql.usselect(p, async function (ret) {
  570. res.end(JSON.stringify(ret));
  571. });
  572. }
  573. });
  574. //删除文件
  575. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  576. if (req.body[0]) {
  577. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  578. let ids = req.body[0].ids;
  579. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  580. console.log("文件删除成功");
  581. }).catch(error => {
  582. console.error("删除文件时发生错误:", error);
  583. })
  584. p = [ids];
  585. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  586. mysql.usselect(p, async function (ret) {
  587. res.end(JSON.stringify(ret));
  588. });
  589. }
  590. });
  591. router.route("/getFolder").all((req, res, next) => {
  592. getmysql2(req, res, "getFolder", req.body.uid);
  593. });
  594. router.route("/getFolder2").all((req, res, next) => {
  595. var json = queryString(req.url);
  596. getmysql2(req, res, "getFolder", json['uid']);
  597. });
  598. router.route("/getFolderPage").all((req, res, next) => {
  599. 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);
  600. });
  601. router.route("/getFolderPage3").all((req, res, next) => {
  602. 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);
  603. });
  604. router.route("/getFolderPage32").all((req, res, next) => {
  605. var json = queryString(req.url);
  606. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  607. });
  608. router.route("/getFolderPageOpen").all((req, res, next) => {
  609. 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);
  610. });
  611. router.route("/getFolderPageOpen2").all((req, res, next) => {
  612. getmysql2(req, res, "getFolderPageOpen2", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].tag, req.body[0].juri2, req.body[0].page, req.body[0].num);
  613. });
  614. router.route("/haveFile").all((req, res, next) => {
  615. postmysql(req, res, "haveFile");
  616. });
  617. router.route("/haveFolder").all((req, res, next) => {
  618. postmysql(req, res, "haveFolder");
  619. });
  620. //search文件
  621. router.route("/searchFile").all(async (req, res, next) => {
  622. if (req.body) {
  623. try {
  624. let filters = {}
  625. if (req.body.type == 1) {
  626. filters = { "document_id": { "$in": req.body.ids } }
  627. } else if (req.body.type == 2) {
  628. filters = { "collection_ids": { "$in": req.body.ids } }
  629. }
  630. const response = await client.retrieval.search({
  631. query: req.body.query,
  632. search_settings: {
  633. filters: filters,
  634. useSemanticSearch: true,
  635. chunkSettings: {
  636. limit: req.body.limit ? req.body.limit : 10,
  637. enabled: true
  638. },
  639. }
  640. });
  641. res.status(200).json({ result: response.results });
  642. } catch (error) {
  643. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  644. }
  645. }
  646. });
  647. //提取文件实体
  648. router.route("/extractFile").all(async (req, res, next) => {
  649. if (req.body) {
  650. try {
  651. const response = client.documents.extract({
  652. id: req.body.id
  653. })
  654. res.status(200).json({ result: response.results });
  655. } catch (error) {
  656. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  657. }
  658. }
  659. });
  660. //添加上传好的知识库文件
  661. router.route("/addFile").all((req, res, next) => {
  662. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  663. })
  664. //获取文件详情
  665. router.route("/getFileDetail").all(async (req, res, next) => {
  666. if (req.body[0]) {
  667. p = Object.values(req.body[0]);
  668. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  669. mysql.usselect(p, async function (ret) {
  670. let data = ret[0][0];
  671. // const array = data.map(item => item.documentid);
  672. try {
  673. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  674. // console.log(response);
  675. let data2 = response.results[0]
  676. // 合并data和data2,根据documentid
  677. const mergedData = { ...data, ...data2 };
  678. res.status(200).json({ result: mergedData });
  679. } catch (error) {
  680. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  681. }
  682. });
  683. }
  684. });
  685. //获取切片
  686. router.route("/getChunks").all(async (req, res, next) => {
  687. if (req.body[0]) {
  688. try {
  689. const { page = 1, limit = 10, documentId } = req.body[0];
  690. const offset = (page - 1) * limit;
  691. const response = await client.documents.listChunks({
  692. id: documentId,
  693. includeVectors: false,
  694. offset,
  695. limit,
  696. });
  697. res.status(200).json({ result: response });
  698. } catch (error) {
  699. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  700. }
  701. }
  702. });
  703. //获取实体
  704. router.route("/getEntities").all(async (req, res, next) => {
  705. if (req.body[0]) {
  706. try {
  707. const { page = 1, limit = 10, documentId } = req.body[0];
  708. const offset = (page - 1) * limit;
  709. const response = await client.documents.listEntities({
  710. id: documentId,
  711. includeVectors: false,
  712. offset,
  713. limit,
  714. });
  715. res.status(200).json({ result: response });
  716. } catch (error) {
  717. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  718. }
  719. }
  720. });
  721. //获取关系
  722. router.route("/getRelationships").all(async (req, res, next) => {
  723. if (req.body[0]) {
  724. try {
  725. const { page = 1, limit = 10, documentId } = req.body[0];
  726. const offset = (page - 1) * limit;
  727. const response = await client.documents.listRelationships({
  728. id: documentId,
  729. includeVectors: false,
  730. offset,
  731. limit,
  732. });
  733. res.status(200).json({ result: response });
  734. } catch (error) {
  735. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  736. }
  737. }
  738. });
  739. //提取整个文件夹的关系
  740. router.route("/resetGraphs").all(async (req, res, next) => {
  741. if (req.body[0]) {
  742. try {
  743. // const response = await client.graphs.pull({
  744. // collectionId: req.body[0].collectionId,
  745. // });
  746. const response = await client.graphs.reset({
  747. collectionId: req.body[0].collectionId,
  748. });
  749. res.status(200).json({ result: response });
  750. } catch (error) {
  751. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  752. }
  753. }
  754. });
  755. //获取整个文件夹的关系
  756. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  757. if (req.body[0]) {
  758. try {
  759. await client.graphs.pull({
  760. collectionId: req.body[0].collectionId,
  761. });
  762. const response = await client.graphs.listRelationships({
  763. collectionId: req.body[0].collectionId,
  764. offset: req.body[0].page ? req.body[0].page : 0,
  765. limit: 1000
  766. });
  767. res.status(200).json({ result: response });
  768. } catch (error) {
  769. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  770. }
  771. }
  772. });
  773. //提取知识图谱
  774. router.route("/getGraphs").all(async (req, res, next) => {
  775. console.log(req.body);
  776. //mode的处理
  777. if (req.body[0]) {
  778. try {
  779. //创建临时文件夹
  780. const folder = await client.collections.create({
  781. name: "临时文件夹",
  782. description: "临时文件夹"
  783. });
  784. let folderid = folder.results.id
  785. console.log(folder.results.id);
  786. //将要提取的文件添加到临时文件夹
  787. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  788. // await Promise.all(documentids.map(documentid =>
  789. // client.documents.extract({
  790. // id: documentid,
  791. // runWithOrchestration: false
  792. // }).catch(error => {
  793. // console.error("提取文件发生错误:", error);
  794. // })
  795. // ));
  796. // 使用 Promise.all 确保所有添加操作都被执行
  797. await Promise.all(documentids.map(documentid =>
  798. client.collections.addDocument({
  799. id: folderid,
  800. documentId: documentid
  801. }).catch(error => {
  802. console.error("移动文件时发生错误:", error);
  803. })
  804. ));
  805. //将临时文件夹提取知识图谱
  806. await client.graphs.pull({
  807. collectionId: folderid
  808. });
  809. let offset = 0;
  810. let results = [];
  811. let totalEntries = 0;
  812. do {
  813. const response = await client.graphs.listRelationships({
  814. collectionId: folderid,
  815. offset: offset,
  816. limit: 1000
  817. });
  818. results = results.concat(response.results);
  819. totalEntries = response.totalEntries;
  820. offset += 1000;
  821. } while (offset < totalEntries);
  822. let offset2 = 0;
  823. let results2 = [];
  824. let totalEntries2 = 0;
  825. do {
  826. const response = await client.graphs.listEntities({
  827. collectionId: folderid,
  828. offset: offset2,
  829. limit: 1000
  830. });
  831. results2 = results2.concat(response.results);
  832. totalEntries2 = response.totalEntries;
  833. offset2 += 1000;
  834. } while (offset2 < totalEntries2);
  835. //移除临时文件夹的文件
  836. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  837. await Promise.all(files.results.map(file =>
  838. client.collections.removeDocument({ id: folderid, documentId: file.id })
  839. ));
  840. //删除临时文件夹
  841. await client.collections.delete({ id: folderid });
  842. res.status(200).json({ result: results, result2: results2 });
  843. } catch (error) {
  844. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  845. }
  846. }
  847. });
  848. //提取知识图谱
  849. router.route("/getGraphs2").all(async (req, res, next) => {
  850. console.log(req.body);
  851. //mode的处理
  852. if (req.body[0]) {
  853. try {
  854. let documentIds = decodeURIComponent(req.body[0].fileids).split(",");
  855. // 使用 Promise.all 并通过 map 返回新的 results 数组
  856. let results = await Promise.all(documentIds.map(async documentId => {
  857. try {
  858. const response = await client.documents.listRelationships({
  859. id: documentId,
  860. includeVectors: false,
  861. offset: 0,
  862. limit: 1000, // 修正这里的参数
  863. });
  864. return response.results;
  865. } catch (error) {
  866. console.error(`Error fetching relationships for documentId ${documentId}:`, error);
  867. return []; // 返回空数组
  868. }
  869. }));
  870. // 使用 flat() 合并所有数组
  871. results = results.flat();
  872. res.status(200).json({ result: results });
  873. } catch (error) {
  874. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  875. }
  876. }
  877. });
  878. //获取实体和关系
  879. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  880. if (req.body[0]) {
  881. try {
  882. const { page = 1, limit = 10, documentId } = req.body[0];
  883. const offset = (page - 1) * limit;
  884. const response = await client.documents.listEntities({
  885. id: documentId,
  886. includeVectors: false,
  887. offset,
  888. limit,
  889. });
  890. const response2 = await client.documents.listRelationships({
  891. id: documentId,
  892. includeVectors: false,
  893. offset,
  894. limit,
  895. });
  896. res.status(200).json({ Entities: response, Relationships: response2 });
  897. } catch (error) {
  898. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  899. }
  900. }
  901. });
  902. //提取功能
  903. router.route("/extractFile2").all(async (req, res, next) => {
  904. try {
  905. let setting = {}
  906. if(req.body[0].prompt){
  907. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  908. }
  909. console.log(setting);
  910. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  911. console.log(response);
  912. res.status(200).json({ result: response.results });
  913. } catch (error) {
  914. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  915. }
  916. });
  917. //复制文件夹
  918. router.route("/copyFolder").all(async (req, res, next) => {
  919. if (req.body[0]) {
  920. const response = await client.collections.create({
  921. name: decodeURIComponent(req.body[0].n),
  922. description: decodeURIComponent(req.body[0].fd)
  923. });
  924. let folderid = response.results.id;
  925. p2 = [req.body[0].folderid]
  926. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  927. mysql.usselect(p2, async function (ret2) {
  928. let file = ret2[0]
  929. if (file.length) {
  930. let documentids = file.map(item => item.documentid);
  931. // 使用 Promise.all 确保所有添加操作都被执行
  932. await Promise.all(documentids.map(documentid =>
  933. client.collections.addDocument({
  934. id: folderid,
  935. documentId: documentid
  936. }).catch(error => {
  937. console.error("移动文件时发生错误:", error);
  938. })
  939. ));
  940. }
  941. let ids = file.map(item => item.fileid);
  942. console.log(ids);
  943. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  944. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  945. mysql.usselect(p, async function (ret) {
  946. res.end(JSON.stringify(ret));
  947. });
  948. });
  949. }
  950. });
  951. // 下载文件功能
  952. router.route("/downloadFile").all(async (req, res, next) => {
  953. var json = queryString(req.url);
  954. let params = {
  955. documentid: (json && json['documentid'] || req.body[0].documentid),
  956. filename: (json && json['filename'] || req.body[0].filename),
  957. };
  958. try {
  959. const response = await client.documents.download({ id: params.documentid });
  960. console.log(response);
  961. // 根据response的类型设置Content-Type
  962. const contentType = response.type || 'application/octet-stream';
  963. res.setHeader('Content-Type', contentType);
  964. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  965. // 直接发送blob数据
  966. res.send(Buffer.from(await response.arrayBuffer()));
  967. } catch (error) {
  968. console.error("下载文件时发生错误:", error);
  969. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  970. }
  971. });
  972. // 下载文件功能
  973. router.route("/downloadFile2").all(async (req, res, next) => {
  974. var json = queryString(req.url);
  975. let params = {
  976. documentid: (json && json['documentid'] || req.body[0].documentid),
  977. filename: (json && json['filename'] || req.body[0].filename),
  978. };
  979. try {
  980. const response = await client.documents.download({ id: params.documentid });
  981. console.log(response);
  982. // 直接返回blob数据
  983. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  984. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  985. const buffer = Buffer.from(await response.arrayBuffer());
  986. res.send(buffer); // 发送blob数据
  987. } catch (error) {
  988. console.error("下载文件时发生错误:", error);
  989. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  990. }
  991. });
  992. //召回测试
  993. router.route("/ragFolder").all(async (req, res, next) => {
  994. let reqJson = req.body[0]
  995. if (reqJson) {
  996. reqJson.query = decodeURIComponent(reqJson.query)
  997. try {
  998. let filters = {}
  999. if (reqJson.type == 1) {
  1000. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  1001. } else if (reqJson.type == 2) {
  1002. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  1003. } else if (reqJson.type == 3) {
  1004. filters = { "collection_ids": { "$eq": reqJson.ids } }
  1005. }
  1006. console.log(111);
  1007. const response = await client.retrieval.search({
  1008. query: reqJson.query,
  1009. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  1010. // searchMode: 'basic',
  1011. searchSettings: {
  1012. useHybridSearch: reqJson.useType == 3,
  1013. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1014. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  1015. hybrid_settings: {
  1016. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1017. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1018. },
  1019. filters: filters,
  1020. limit: reqJson.limit ? reqJson.limit : 100,
  1021. },
  1022. ragGenerationConfig: {
  1023. stream: false,
  1024. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1025. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1026. }
  1027. });
  1028. // res.status(200).json({ result: response.results });
  1029. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1030. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1031. mysql.usselect(p, async function (ret) {
  1032. res.status(200).json({ result: response.results });
  1033. });
  1034. } catch (error) {
  1035. console.log(error);
  1036. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1037. }
  1038. }
  1039. });
  1040. //召回测试
  1041. router.route("/searchFolder").all(async (req, res, next) => {
  1042. let reqJson = req.body[0]
  1043. if (reqJson) {
  1044. try {
  1045. let filters = {}
  1046. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1047. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1048. filters = {
  1049. "$or": [
  1050. { "collection_ids": { "$in": folderids } },
  1051. { "document_id": { "$in": documentids } }
  1052. ]
  1053. };
  1054. let params = {
  1055. query: reqJson.query,
  1056. searchSettings: {
  1057. useHybridSearch: reqJson.useType == 3,
  1058. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1059. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1060. hybrid_settings: {
  1061. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1062. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1063. },
  1064. filters: filters,
  1065. limit: reqJson.limit ? reqJson.limit : 100,
  1066. },
  1067. ragGenerationConfig: {
  1068. stream: false,
  1069. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1070. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1071. }
  1072. }
  1073. // console.log(params);
  1074. const response = await client.retrieval.search(params);
  1075. res.status(200).json({ result: response.results });
  1076. } catch (error) {
  1077. console.log(error);
  1078. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1079. }
  1080. }
  1081. });
  1082. //召回测试
  1083. router.route("/searchFolder2").all(async (req, res, next) => {
  1084. let reqJson = req.body
  1085. if (Object.keys(reqJson).length) {
  1086. try {
  1087. let filters = {}
  1088. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  1089. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  1090. filters = {
  1091. "$or": [
  1092. { "collection_ids": { "$in": folderids } },
  1093. { "document_id": { "$in": documentids } }
  1094. ]
  1095. };
  1096. if(!folderids.length && !documentids.length){
  1097. res.status(200).json({ result: [] });
  1098. return
  1099. }
  1100. let params = {
  1101. query: reqJson.query,
  1102. searchSettings: {
  1103. useHybridSearch: reqJson.useType == 3,
  1104. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1105. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1106. hybrid_settings: {
  1107. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1108. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1109. },
  1110. filters: filters,
  1111. limit: reqJson.limit ? reqJson.limit : 100,
  1112. },
  1113. ragGenerationConfig: {
  1114. stream: false,
  1115. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1116. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1117. }
  1118. }
  1119. // console.log(params);
  1120. const response = await client.retrieval.search(params);
  1121. res.status(200).json({ result: response.results });
  1122. } catch (error) {
  1123. console.log(error);
  1124. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1125. }
  1126. }else{
  1127. res.status(200).json({ result: "请传参" });
  1128. }
  1129. });
  1130. //召回测试
  1131. router.route("/searchFolder3").all(async (req, res, next) => {
  1132. let reqJson = req.body[0]
  1133. if (reqJson) {
  1134. reqJson.query = decodeURIComponent(reqJson.query)
  1135. try {
  1136. let filters = {}
  1137. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1138. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1139. filters = {
  1140. "$or": [
  1141. { "collection_ids": { "$in": folderids } },
  1142. { "document_id": { "$in": documentids } }
  1143. ]
  1144. };
  1145. if(!folderids.length && !documentids.length){
  1146. res.status(200).json({ result: [] });
  1147. return
  1148. }
  1149. let params = {
  1150. query: reqJson.query,
  1151. searchSettings: {
  1152. useHybridSearch: reqJson.useType == 3,
  1153. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1154. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1155. hybrid_settings: {
  1156. fullTextWeight: reqJson.fullTextWeight,
  1157. semanticWeight: reqJson.semanticWeight,
  1158. },
  1159. filters: filters,
  1160. limit: reqJson.limit ? reqJson.limit : 100,
  1161. },
  1162. ragGenerationConfig: {
  1163. stream: false,
  1164. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1165. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1166. }
  1167. }
  1168. // console.log(params);
  1169. const response = await client.retrieval.search(params);
  1170. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1171. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1172. mysql.usselect(p, async function (ret) {
  1173. res.status(200).json({ result: response.results });
  1174. });
  1175. } catch (error) {
  1176. console.log(error);
  1177. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1178. }
  1179. }
  1180. });
  1181. //召回测试
  1182. router.route("/searchFolder32").all(async (req, res, next) => {
  1183. let reqJson = req.body[0]
  1184. if (reqJson) {
  1185. reqJson.query = decodeURIComponent(reqJson.query)
  1186. try {
  1187. let filters = {}
  1188. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1189. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1190. filters = {
  1191. "$or": [
  1192. { "collection_ids": { "$in": folderids } },
  1193. { "document_id": { "$in": documentids } }
  1194. ]
  1195. };
  1196. if(!folderids.length && !documentids.length){
  1197. res.status(200).json({ result: [] });
  1198. return
  1199. }
  1200. let params = {
  1201. query: reqJson.query,
  1202. searchSettings: {
  1203. useHybridSearch: reqJson.useType == 3,
  1204. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1205. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1206. hybrid_settings: {
  1207. fullTextWeight: reqJson.fullTextWeight,
  1208. semanticWeight: reqJson.semanticWeight,
  1209. },
  1210. filters: filters,
  1211. limit: reqJson.limit ? reqJson.limit : 100,
  1212. },
  1213. ragGenerationConfig: {
  1214. stream: false,
  1215. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1216. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1217. }
  1218. }
  1219. console.log(params);
  1220. const response = await client.retrieval.search(params);
  1221. res.status(200).json({ result: response.results });
  1222. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1223. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1224. // mysql.usselect(p, async function (ret) {
  1225. // });
  1226. } catch (error) {
  1227. console.log(error);
  1228. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1229. }
  1230. }
  1231. });
  1232. //获取召回测试数据
  1233. router.route("/getRagData").all(async (req, res, next) => {
  1234. //mode的处理
  1235. if (req.body[0]) {
  1236. p = Object.values(req.body[0]);
  1237. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1238. mysql.usselect(p, async function (ret) {
  1239. res.status(200).json(ret);
  1240. });
  1241. }
  1242. });
  1243. //删除召回测试数据
  1244. router.route("/delRagData").all(async (req, res, next) => {
  1245. //mode的处理
  1246. if (req.body[0]) {
  1247. p = Object.values(req.body[0]);
  1248. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1249. mysql.usselect(p, async function (ret) {
  1250. res.status(200).json(ret);
  1251. });
  1252. }
  1253. });
  1254. function generateAPIKey(username) {
  1255. let timeStamp = new Date().getTime();
  1256. let hashStr = timeStamp + username;
  1257. let digest = crypto
  1258. .createHash("sha256")
  1259. .update(hashStr, "utf8")
  1260. .digest("hex");
  1261. return digest;
  1262. }
  1263. queryString = function (url) {
  1264. var a = url.split("?");
  1265. var json = querystring.parse(a[1]);
  1266. return json;
  1267. };
  1268. //post存儲過程處理
  1269. postmysql = function (req, res, functionname) {
  1270. //mode的處理
  1271. if (req.body[0]) {
  1272. p = Object.values(req.body[0]);
  1273. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1274. //執行存儲過程
  1275. mysql.usselect(p, function (ret) {
  1276. res.end(JSON.stringify(ret));
  1277. });
  1278. }
  1279. };
  1280. //post存儲過程處理
  1281. postmysql2 = function (req, res, functionname) {
  1282. //mode的處理
  1283. if (req.body) {
  1284. p = Object.values(req.body)[0].split(",");
  1285. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1286. //執行存儲過程
  1287. mysql.usselect(p, function (ret) {
  1288. res.end(JSON.stringify(ret));
  1289. });
  1290. }
  1291. };
  1292. //post存儲過程處理
  1293. postmysql3 = function (req, res, functionname) {
  1294. //mode的處理
  1295. if (req.body) {
  1296. console.log(req);
  1297. p = Object.values(req.body);
  1298. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1299. //執行存儲過程
  1300. mysql.usselect(p, function (ret) {
  1301. res.end(JSON.stringify(ret));
  1302. });
  1303. }
  1304. };
  1305. //get獲取數據庫信息
  1306. getmysql = function (req, res, functionname, ...p) {
  1307. p = p || [];
  1308. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1309. //執行存儲過程
  1310. mysql.usselect(p, function (ret) {
  1311. res.end(JSON.stringify(ret));
  1312. }); //
  1313. };
  1314. //get獲取數據庫信息
  1315. getmysql2 = function (req, res, functionname, ...p) {
  1316. p = p || [];
  1317. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1318. //執行存儲過程
  1319. mysql.usselect(p, function (ret) {
  1320. res.end(JSON.stringify(ret));
  1321. }); //
  1322. };
  1323. //get獲取數據庫信息
  1324. getmysql3 = function (req, res, functionname, ...p) {
  1325. p = p || [];
  1326. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1327. //執行存儲過程
  1328. mysql.usselect(p, function (ret) {
  1329. res.end(JSON.stringify(ret));
  1330. }); //
  1331. };
  1332. //post存储过程处理
  1333. userpostmysql = function (req, res, functionname) {
  1334. //mode的处理
  1335. if (req.body.mode) {
  1336. p = req.body.mode.split(",");
  1337. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1338. //执行存储过程
  1339. mysql.usselect(p, function (ret) {
  1340. res.end(JSON.stringify(ret));
  1341. });
  1342. }
  1343. };
  1344. //get获取数据库信息
  1345. usergetmysql = function (req, res, functionname, ...p) {
  1346. p = p || [];
  1347. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1348. //执行存储过程
  1349. mysql.usselect(p, function (ret) {
  1350. res.end(JSON.stringify(ret));
  1351. }); //
  1352. };
  1353. //post存儲過程處理
  1354. postmysqlLocal = function (req, res) {
  1355. //mode的處理
  1356. if (req.body[0]) {
  1357. p = Object.values(req.body[0]);
  1358. p.unshift(_localhost[0], _localhost[1]);
  1359. //執行存儲過程
  1360. mysql.usselect(p, function (ret) {
  1361. res.end(JSON.stringify(ret));
  1362. });
  1363. }
  1364. };
  1365. postmysqlOnline = function (req, res) {
  1366. //mode的處理
  1367. if (req.body[0]) {
  1368. p = Object.values(req.body[0]);
  1369. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1370. //執行存儲過程
  1371. mysql.usselect(p, function (ret) {
  1372. res.end(JSON.stringify(ret));
  1373. });
  1374. }
  1375. };
  1376. // 新增测试接口,直接返回成功
  1377. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1378. // console.log('收到的 req.body:', req.body);
  1379. // console.log('收到的 req.files:', req.files);
  1380. // console.log('收到的 headers:', req.headers);
  1381. const requestUrl = req.body.url; // 前端传入 url
  1382. if (!requestUrl) {
  1383. return res.status(400).json({error: 'Missing url'});
  1384. }
  1385. const browser = await initBrowser();
  1386. const page = await browser.newPage();
  1387. try {
  1388. console.log("开始抓取页面:", requestUrl);
  1389. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1390. await page.goto(requestUrl);
  1391. const html = await page.content();
  1392. // 新增:HTML 转 Markdown
  1393. const turndownService = new TurndownService();
  1394. // 用 cheerio 只提取正文部分
  1395. const cheerioDoc = cheerio.load(html);
  1396. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1397. // 用 cheerio 过滤无关标签
  1398. const cheerioMain = cheerio.load(mainHtml || '');
  1399. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1400. mainHtml = cheerioMain.html();
  1401. const markdownContent = turndownService.turndown(mainHtml);
  1402. // 新增:解析 <title> 标签内容
  1403. let title = '';
  1404. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1405. if (titleMatch && titleMatch[1]) {
  1406. title = titleMatch[1].trim();
  1407. }
  1408. let finalTitle;
  1409. if (title && title.trim()) {
  1410. finalTitle = title + '.md';
  1411. } else {
  1412. finalTitle = 'untitled-' + uuidv4() + '.md';
  1413. }
  1414. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1415. // 解析用户传入的参数
  1416. const {
  1417. collection_ids,
  1418. id: customId,
  1419. metadata: customMetadata,
  1420. ingestion_mode,
  1421. run_with_orchestration
  1422. } = req.body;
  1423. // 处理 metadata 字段
  1424. let finalMetadata = customMetadata;
  1425. if (typeof finalMetadata === 'string') {
  1426. try {
  1427. finalMetadata = JSON.parse(finalMetadata);
  1428. } catch (e) {
  1429. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1430. }
  1431. }
  1432. // if (!finalMetadata) {
  1433. // finalMetadata = {
  1434. // title: title || uuidv4(),
  1435. // // source_url: url,
  1436. // content_type: "markdown",
  1437. // // scraped_at: new Date().toISOString(),
  1438. // // upload_id: Date.now() + '-' + Math.random()
  1439. // };
  1440. // }
  1441. // 处理 collection_ids 字段
  1442. let finalCollectionIds = collection_ids;
  1443. if (typeof finalCollectionIds === 'string') {
  1444. try {
  1445. finalCollectionIds = JSON.parse(finalCollectionIds);
  1446. } catch (e) {
  1447. // 不是合法 JSON 字符串就当作普通字符串
  1448. }
  1449. }
  1450. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1451. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1452. finalMetadata.collection_ids = finalCollectionIds;
  1453. }
  1454. // 补全 finalMetadata.title 和 content_type
  1455. if (!finalMetadata.title) {
  1456. finalMetadata.title = finalTitle;
  1457. }
  1458. finalMetadata.content_type = 'markdown';
  1459. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1460. const FormData = require('form-data');
  1461. const fs = require('fs');
  1462. const path = require('path');
  1463. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1464. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1465. const form = new FormData();
  1466. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1467. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1468. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1469. }
  1470. if (customId) {
  1471. form.append('id', customId);
  1472. }
  1473. form.append('metadata', JSON.stringify(finalMetadata));
  1474. form.append('ingestion_mode', ingestion_mode || 'fast');
  1475. form.append('run_with_orchestration', 'false');
  1476. // 打印上传参数,方便排查问题
  1477. console.log('Request Headers:', form.getHeaders());
  1478. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1479. const response = await axios.post(url, form, {
  1480. headers: {
  1481. ...form.getHeaders(),
  1482. // 'User-Agent': 'NodeClient/1.0'
  1483. }
  1484. });
  1485. fs.unlinkSync(tmpFilePath);
  1486. result = {
  1487. results: response.data.results,
  1488. title: finalTitle
  1489. }
  1490. res.json(result);
  1491. } catch (error) {
  1492. console.error('Message:', error.message);
  1493. console.error('Stack:', error.stack);
  1494. res.status(500).json({error: error.message});
  1495. } finally {
  1496. await page.close(); // 确保页面总是被关闭
  1497. }
  1498. });
  1499. module.exports = router;