pbl.js 63 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742
  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 = ["183.36.25.92", "pbl"]; //edu數據庫信息
  10. // const _mysqluser = ["183.36.25.92", "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 _getmysqlLabor2 = ["172.16.12.5", "pbl"]; //edu數據庫信息
  20. // const _getmysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  21. const _getmysqluser = ["172.16.12.7", "cocorobouser"]; //edu數據庫信息
  22. const _localhost = ["127.0.0.1", "pbl"]; //edu數據庫信息
  23. const crypto = require("crypto");
  24. const https = require("https");
  25. const { Pool } = require('pg');
  26. const axios = require('axios');
  27. const multer = require("multer");
  28. const {chromium} = require('playwright');
  29. const TurndownService = require('turndown');
  30. const cheerio = require('cheerio');
  31. const { v4: uuidv4 } = require('uuid');
  32. const loginHandler = require('./login'); // 确保路径正确
  33. const { r2rClient } = require("r2r-js");
  34. const client = new r2rClient("https://r2rserver.cocorobo.cn/");
  35. // 配置 multer 存储选项
  36. const storage = multer.diskStorage({
  37. destination: function (req, file, cb) {
  38. req.setEncoding('utf-8');
  39. cb(null, 'uploads/'); // 文件上传目录
  40. },
  41. filename: function (req, file, cb) {
  42. req.setEncoding('utf-8');
  43. // 使用 Buffer 处理文件名,确保编码正确
  44. const fileExtension = file.originalname.split('.').pop(); // 获取文件后缀
  45. const fileNameWithoutExt = Buffer.from(file.originalname.slice(0, -(fileExtension.length + 1)), 'latin1').toString('utf8'); // 转换编码
  46. cb(null, fileNameWithoutExt + '-' + Date.now() + '.' + fileExtension); // 使用时间戳和原始文件名
  47. }
  48. });
  49. const upload = multer({
  50. storage: storage, limits: {
  51. fileSize: 100 * 1024 * 1024 // 100MB
  52. }
  53. });
  54. // 全局浏览器实例
  55. let globalBrowser = null;
  56. async function initBrowser() {
  57. if (!globalBrowser) {
  58. globalBrowser = await chromium.launch({headless: true});
  59. console.log('Browser initialized');
  60. }
  61. return globalBrowser;
  62. }
  63. router.route("/login").all(async (req, res, next) => {
  64. loginHandler(req, res)
  65. });
  66. router.route("/logout").all(async (req, res, next) => {
  67. try {
  68. const response = await client.users.logout();
  69. res.status(200).json({ result: response.results });
  70. } catch (error) {
  71. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  72. }
  73. });
  74. //获取文档列表
  75. router.route("/getDocument").all(async (req, res, next) => {
  76. try {
  77. const response = await client.documents.list({
  78. limit: 10,
  79. offset: 0,
  80. });
  81. console.log(response);
  82. res.status(200).json({ result: response.results });
  83. } catch (error) {
  84. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  85. }
  86. });
  87. //列出文档块
  88. router.route("/listChunks").all(async (req, res, next) => {
  89. try {
  90. const response = await client.documents.listChunks({
  91. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  92. });
  93. console.log(response);
  94. res.status(200).json({ result: response.results });
  95. } catch (error) {
  96. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  97. }
  98. });
  99. //提取实体和关系
  100. router.route("/extract").all(async (req, res, next) => {
  101. try {
  102. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  103. console.log(response);
  104. res.status(200).json({ result: response.results });
  105. } catch (error) {
  106. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  107. }
  108. });
  109. router.route("/selectUser").all((req, res, next) => {
  110. var json = queryString(req.url);
  111. getmysql(req, res, "selectUser", json["userid"]);
  112. });
  113. // 文件上传接口
  114. router.post('/upload', upload.single('file'), async (req, res) => {
  115. req.setEncoding('utf-8')
  116. try {
  117. console.log(req.file); // 打印上传的文件信息
  118. console.log(`File path: uploads/${req.file.filename}`);
  119. const path = require('path');
  120. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  121. console.log(filePath);
  122. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  123. const fs = require('fs');
  124. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  125. const response = await client.documents.create({
  126. file: file,
  127. metadata: { title: filename }
  128. });
  129. console.log(response);
  130. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  131. } catch (error) {
  132. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  133. }
  134. });
  135. //获取文档列表
  136. router.route("/getFile").all(async (req, res, next) => {
  137. //mode的处理
  138. if (req.body[0]) {
  139. p = Object.values(req.body[0]);
  140. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  141. mysql.usselect(p, async function (ret) {
  142. let data = ret[0];
  143. // const array = data.map(item => item.documentid);
  144. const array = data.map(item => item.documentid);
  145. try {
  146. const response = await client.documents.list({ ids: array });
  147. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  148. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  149. // console.log(response);
  150. let data2 = response.results.map(doc => {
  151. doc.documentid = doc.id;
  152. delete doc.id;
  153. return doc;
  154. });
  155. // console.log(data2);
  156. // 合并data和data2,根据documentid
  157. const mergedData = data.map(item => {
  158. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  159. return { ...item, ...correspondingDoc };
  160. });
  161. res.status(200).json({ result: mergedData });
  162. } catch (error) {
  163. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  164. }
  165. });
  166. }
  167. });
  168. //获取文档列表
  169. router.route("/getFileR2r").all(async (req, res, next) => {
  170. //mode的处理
  171. if (req.body[0]) {
  172. let fileid = decodeURIComponent(req.body[0].fileids).split(",");
  173. try {
  174. const response = await client.documents.list({ ids: fileid });
  175. res.status(200).json({ result: response.results });
  176. } catch (error) {
  177. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  178. }
  179. }
  180. });
  181. //获取文档列表
  182. router.route("/getFileD").all(async (req, res, next) => {
  183. //mode的处理
  184. if (req.body[0]) {
  185. p = Object.values(req.body[0]);
  186. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  187. mysql.usselect(p, async function (ret) {
  188. let data = ret[0];
  189. // const array = data.map(item => item.documentid);
  190. const array = data.map(item => item.documentid);
  191. try {
  192. const response = await client.documents.list({ ids: array });
  193. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  194. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  195. // console.log(response);
  196. let data2 = response.results.map(doc => {
  197. doc.documentid = doc.id;
  198. delete doc.id;
  199. return doc;
  200. });
  201. // console.log(data2);
  202. // 合并data和data2,根据documentid
  203. const mergedData = data.map(item => {
  204. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  205. return { ...item, ...correspondingDoc };
  206. });
  207. res.status(200).json({ result: mergedData });
  208. } catch (error) {
  209. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  210. }
  211. });
  212. }
  213. });
  214. //获取文档列表
  215. router.route("/getFileArray").all(async (req, res, next) => {
  216. //mode的处理
  217. if (req.body[0]) {
  218. // const array = data.map(item => item.documentid);
  219. try {
  220. let data = decodeURIComponent(req.body[0].array).split(",");
  221. const response = await client.documents.list({ ids: data });
  222. res.status(200).json({ result: response.results });
  223. } catch (error) {
  224. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  225. }
  226. }
  227. });
  228. //获取文档列表
  229. router.route("/getFileD2").all(async (req, res, next) => {
  230. //mode的处理
  231. if (req.body) {
  232. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  233. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  234. mysql.usselect(p, async function (ret) {
  235. let data = ret[0];
  236. // const array = data.map(item => item.documentid);
  237. const array = data.map(item => item.documentid);
  238. try {
  239. const response = await client.documents.list({ ids: array });
  240. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  241. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  242. // console.log(response);
  243. let data2 = response.results.map(doc => {
  244. doc.documentid = doc.id;
  245. delete doc.id;
  246. return doc;
  247. });
  248. // console.log(data2);
  249. // 合并data和data2,根据documentid
  250. const mergedData = data.map(item => {
  251. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  252. return { ...item, ...correspondingDoc };
  253. });
  254. res.status(200).json({ result: mergedData });
  255. } catch (error) {
  256. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  257. }
  258. });
  259. }
  260. });
  261. //获取文档列表
  262. router.route("/getFileD2").all(async (req, res, next) => {
  263. //mode的处理
  264. if (req.body) {
  265. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  266. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  267. mysql.usselect(p, async function (ret) {
  268. let data = ret[0];
  269. // const array = data.map(item => item.documentid);
  270. const array = data.map(item => item.documentid);
  271. try {
  272. const response = await client.documents.list({ ids: array });
  273. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  274. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  275. // console.log(response);
  276. let data2 = response.results.map(doc => {
  277. doc.documentid = doc.id;
  278. delete doc.id;
  279. return doc;
  280. });
  281. // console.log(data2);
  282. // 合并data和data2,根据documentid
  283. const mergedData = data.map(item => {
  284. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  285. return { ...item, ...correspondingDoc };
  286. });
  287. res.status(200).json({ result: mergedData });
  288. } catch (error) {
  289. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  290. }
  291. });
  292. }
  293. });
  294. //获取文档列表
  295. router.route("/getFileOpen").all(async (req, res, next) => {
  296. //mode的处理
  297. if (req.body[0]) {
  298. p = Object.values(req.body[0]);
  299. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  300. mysql.usselect(p, async function (ret) {
  301. let data = ret[0];
  302. let user = ret[1];
  303. // const array = data.map(item => item.documentid);
  304. const array = data.map(item => item.documentid);
  305. try {
  306. const response = await client.documents.list({ ids: array });
  307. let data2 = response.results.map(doc => {
  308. doc.documentid = doc.id;
  309. delete doc.id;
  310. return doc;
  311. });
  312. // 合并data和data2,根据documentid
  313. const mergedData = data.map(item => {
  314. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  315. return { ...item, ...correspondingDoc };
  316. });
  317. res.status(200).json({ result: mergedData, user: user });
  318. } catch (error) {
  319. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  320. }
  321. });
  322. }
  323. });
  324. //获取文档列表2
  325. router.route("/getFileP").all(async (req, res, next) => {
  326. //mode的处理
  327. if (req.body) {
  328. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  329. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  330. mysql.usselect(p, async function (ret) {
  331. let data = ret[0];
  332. // const array = data.map(item => item.documentid);
  333. try {
  334. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  335. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  336. // console.log(response);
  337. let data2 = response.results.map(doc => {
  338. doc.documentid = doc.id;
  339. delete doc.id;
  340. return doc;
  341. });
  342. // 合并data和data2,根据documentid
  343. const mergedData = data.map(item => {
  344. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  345. return { ...item, ...correspondingDoc };
  346. });
  347. res.status(200).json({ result: mergedData });
  348. } catch (error) {
  349. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  350. }
  351. });
  352. }
  353. });
  354. //获取能关联的文件
  355. router.route("/getFile2").all(async (req, res, next) => {
  356. //mode的处理
  357. if (req.body[0]) {
  358. p = Object.values(req.body[0]);
  359. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  360. mysql.usselect(p, async function (ret) {
  361. let data = ret[0];
  362. const array = data.map(item => item.documentid);
  363. try {
  364. const response = await client.documents.list({ ids: array });
  365. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  366. // console.log(response);
  367. let data2 = response.results.map(doc => {
  368. doc.documentid = doc.id;
  369. delete doc.id;
  370. return doc;
  371. });
  372. // 合并data和data2,根据documentid
  373. const mergedData = data.map(item => {
  374. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  375. return { ...item, ...correspondingDoc };
  376. });
  377. res.status(200).json({ result: mergedData });
  378. } catch (error) {
  379. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  380. }
  381. });
  382. }
  383. });
  384. //添加文件夹
  385. router.route("/addFolder").all(async (req, res, next) => {
  386. //mode的处理
  387. if (req.body[0]) {
  388. try {
  389. const response = await client.collections.create({
  390. name: decodeURIComponent(req.body[0].n),
  391. description: decodeURIComponent(req.body[0].fd)
  392. });
  393. p = [...Object.values(req.body[0]), response.results.id];
  394. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  395. mysql.usselect(p, async function (ret) {
  396. res.end(JSON.stringify(ret));
  397. });
  398. } catch (error) {
  399. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  400. }
  401. }
  402. });
  403. //添加文件夹
  404. router.route("/addFolderP").all(async (req, res, next) => {
  405. //mode的处理
  406. if (req.body) {
  407. try {
  408. const response = await client.collections.create({
  409. name: decodeURIComponent(req.body.n),
  410. description: decodeURIComponent(req.body.fd)
  411. });
  412. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  413. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  414. mysql.usselect(p, async function (ret) {
  415. res.end(JSON.stringify(ret));
  416. });
  417. } catch (error) {
  418. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  419. }
  420. }
  421. });
  422. //删除文件夹
  423. router.route("/deleteFolder").all(async (req, res, next) => {
  424. //mode的处理
  425. if (req.body[0]) {
  426. try {
  427. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  428. const ids = req.body[0].ids;
  429. // 并行处理文件夹下的文件删除和文件夹删除
  430. await Promise.all(folderids.map(async folderid => {
  431. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  432. await Promise.all(files.results.map(file =>
  433. client.collections.removeDocument({ id: folderid, documentId: file.id })
  434. ));
  435. }));
  436. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  437. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  438. mysql.usselect(params, async function (ret) {
  439. res.end(JSON.stringify(ret));
  440. });
  441. } catch (error) {
  442. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  443. }
  444. }
  445. });
  446. //修改文件夹
  447. router.route("/updateFolder").all(async (req, res, next) => {
  448. //mode的处理
  449. if (req.body[0]) {
  450. try {
  451. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  452. const response = await client.collections.update({
  453. id: folderid,
  454. name: decodeURIComponent(req.body[0].n),
  455. description: decodeURIComponent(req.body[0].d)
  456. });
  457. p = Object.values(folderData); // 获取剩余属性的键
  458. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  459. mysql.usselect(p, async function (ret) {
  460. res.end(JSON.stringify(ret));
  461. });
  462. } catch (error) {
  463. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  464. }
  465. }
  466. });
  467. //修改文件夹
  468. router.route("/updateFolder2").all(async (req, res, next) => {
  469. //mode的处理
  470. if (req.body[0]) {
  471. try {
  472. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  473. const response = await client.collections.update({
  474. id: folderid,
  475. name: decodeURIComponent(req.body[0].n),
  476. description: decodeURIComponent(req.body[0].d)
  477. });
  478. p = Object.values(folderData); // 获取剩余属性的键
  479. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  480. mysql.usselect(p, async function (ret) {
  481. res.end(JSON.stringify(ret));
  482. });
  483. } catch (error) {
  484. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  485. }
  486. }
  487. });
  488. //修改文件夹
  489. router.route("/updateFolder3").all(async (req, res, next) => {
  490. //mode的处理
  491. if (req.body[0]) {
  492. try {
  493. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  494. const response = await client.collections.update({
  495. id: folderid,
  496. name: decodeURIComponent(req.body[0].n),
  497. description: decodeURIComponent(req.body[0].d)
  498. });
  499. p = Object.values(folderData); // 获取剩余属性的键
  500. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder4");
  501. mysql.usselect(p, async function (ret) {
  502. res.end(JSON.stringify(ret));
  503. });
  504. } catch (error) {
  505. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  506. }
  507. }
  508. });
  509. //移除文件
  510. router.route("/removeFile").all(async (req, res, next) => {
  511. if (req.body[0]) {
  512. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  513. let folderid = req.body[0].folderid;
  514. let ids = req.body[0].ids;
  515. // 使用 Promise.all 确保所有删除操作都被执行
  516. const removePromises = documentids.map(documentid =>
  517. client.collections.removeDocument({ id: folderid, documentId: documentid })
  518. );
  519. // 不管client.documents报错没有也要执行removeFile
  520. Promise.all(removePromises).catch(error => {
  521. console.error("移除文件时发生错误:", error);
  522. });
  523. p = [ids, folderid];
  524. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  525. mysql.usselect(p, async function (ret) {
  526. res.end(JSON.stringify(ret));
  527. });
  528. }
  529. });
  530. //移入文件
  531. router.route("/moveFile").all(async (req, res, next) => {
  532. if (req.body[0]) {
  533. try {
  534. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  535. let folderid = req.body[0].folderid
  536. let ids = decodeURIComponent(req.body[0].ids).split(",")
  537. // 使用 Promise.all 确保所有添加操作都被执行
  538. await Promise.all(documentids.map(documentid =>
  539. client.collections.addDocument({
  540. id: folderid,
  541. documentId: documentid
  542. }).catch(error => {
  543. console.error("移动文件时发生错误:", error);
  544. })
  545. ));
  546. for (const id of ids) {
  547. p = [id, folderid];
  548. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  549. await mysql.usselect(p, async function (ret) { });
  550. }
  551. res.status(200).json({ messages: "移动成功" });
  552. } catch (error) {
  553. // 这里不返回错误,确保deleteFile执行完
  554. console.error("处理过程中发生错误:", error);
  555. }
  556. }
  557. });
  558. //删除文件
  559. router.route("/deleteFile").all(async (req, res, next) => {
  560. if (req.body[0]) {
  561. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  562. let ids = req.body[0].ids;
  563. // 使用 Promise.all 确保所有删除操作都被执行
  564. Promise.all(documentids.map(documentid =>
  565. client.documents.delete({ id: documentid }).catch(error => {
  566. console.error("删除文件时发生错误:", error);
  567. })
  568. ));
  569. p = [ids];
  570. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  571. mysql.usselect(p, async function (ret) {
  572. res.end(JSON.stringify(ret));
  573. });
  574. }
  575. });
  576. //删除文件
  577. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  578. if (req.body[0]) {
  579. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  580. let ids = req.body[0].ids;
  581. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  582. console.log("文件删除成功");
  583. }).catch(error => {
  584. console.error("删除文件时发生错误:", error);
  585. })
  586. p = [ids];
  587. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  588. mysql.usselect(p, async function (ret) {
  589. res.end(JSON.stringify(ret));
  590. });
  591. }
  592. });
  593. router.route("/getFolder").all((req, res, next) => {
  594. getmysql3(req, res, "getFolder", req.body.uid);
  595. });
  596. router.route("/getFolder2").all((req, res, next) => {
  597. var json = queryString(req.url);
  598. getmysql3(req, res, "getFolder", json['uid']);
  599. });
  600. router.route("/getFolderPage").all((req, res, next) => {
  601. getmysql3(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);
  602. });
  603. router.route("/getFolderPage3").all((req, res, next) => {
  604. getmysql3(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);
  605. });
  606. router.route("/getFolderPage32").all((req, res, next) => {
  607. var json = queryString(req.url);
  608. getmysql3(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  609. });
  610. router.route("/getFolderPageOpen").all((req, res, next) => {
  611. getmysql3(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);
  612. });
  613. router.route("/getFolderPageOpen2").all((req, res, next) => {
  614. getmysql3(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);
  615. });
  616. router.route("/haveFile").all((req, res, next) => {
  617. postmysql(req, res, "haveFile");
  618. });
  619. router.route("/haveFolder").all((req, res, next) => {
  620. postmysql(req, res, "haveFolder");
  621. });
  622. //search文件
  623. router.route("/searchFile").all(async (req, res, next) => {
  624. if (req.body) {
  625. try {
  626. let filters = {}
  627. if (req.body.type == 1) {
  628. filters = { "document_id": { "$in": req.body.ids } }
  629. } else if (req.body.type == 2) {
  630. filters = { "collection_ids": { "$in": req.body.ids } }
  631. }
  632. const response = await client.retrieval.search({
  633. query: req.body.query,
  634. search_settings: {
  635. filters: filters,
  636. useSemanticSearch: true,
  637. chunkSettings: {
  638. limit: req.body.limit ? req.body.limit : 10,
  639. enabled: true
  640. },
  641. }
  642. });
  643. res.status(200).json({ result: response.results });
  644. } catch (error) {
  645. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  646. }
  647. }
  648. });
  649. //提取文件实体
  650. router.route("/extractFile").all(async (req, res, next) => {
  651. if (req.body) {
  652. try {
  653. const response = client.documents.extract({
  654. id: req.body.id
  655. })
  656. res.status(200).json({ result: response.results });
  657. } catch (error) {
  658. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  659. }
  660. }
  661. });
  662. //添加上传好的知识库文件
  663. router.route("/addFile").all((req, res, next) => {
  664. getmysql3(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  665. })
  666. //获取文件详情
  667. router.route("/getFileDetail").all(async (req, res, next) => {
  668. if (req.body[0]) {
  669. p = Object.values(req.body[0]);
  670. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  671. mysql.usselect(p, async function (ret) {
  672. let data = ret[0][0];
  673. // const array = data.map(item => item.documentid);
  674. try {
  675. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  676. // console.log(response);
  677. let data2 = response.results[0]
  678. // 合并data和data2,根据documentid
  679. const mergedData = { ...data, ...data2 };
  680. res.status(200).json({ result: mergedData });
  681. } catch (error) {
  682. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  683. }
  684. });
  685. }
  686. });
  687. //获取切片
  688. router.route("/getChunks").all(async (req, res, next) => {
  689. if (req.body[0]) {
  690. try {
  691. const { page = 1, limit = 10, documentId } = req.body[0];
  692. const offset = (page - 1) * limit;
  693. const response = await client.documents.listChunks({
  694. id: documentId,
  695. includeVectors: false,
  696. offset,
  697. limit,
  698. });
  699. res.status(200).json({ result: response });
  700. } catch (error) {
  701. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  702. }
  703. }
  704. });
  705. //获取实体
  706. router.route("/getEntities").all(async (req, res, next) => {
  707. if (req.body[0]) {
  708. try {
  709. const { page = 1, limit = 10, documentId } = req.body[0];
  710. const offset = (page - 1) * limit;
  711. const response = await client.documents.listEntities({
  712. id: documentId,
  713. includeVectors: false,
  714. offset,
  715. limit,
  716. });
  717. res.status(200).json({ result: response });
  718. } catch (error) {
  719. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  720. }
  721. }
  722. });
  723. //获取关系
  724. router.route("/getRelationships").all(async (req, res, next) => {
  725. if (req.body[0]) {
  726. try {
  727. const { page = 1, limit = 10, documentId } = req.body[0];
  728. const offset = (page - 1) * limit;
  729. const response = await client.documents.listRelationships({
  730. id: documentId,
  731. includeVectors: false,
  732. offset,
  733. limit,
  734. });
  735. res.status(200).json({ result: response });
  736. } catch (error) {
  737. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  738. }
  739. }
  740. });
  741. //提取整个文件夹的关系
  742. router.route("/resetGraphs").all(async (req, res, next) => {
  743. if (req.body[0]) {
  744. try {
  745. // const response = await client.graphs.pull({
  746. // collectionId: req.body[0].collectionId,
  747. // });
  748. const response = await client.graphs.reset({
  749. collectionId: req.body[0].collectionId,
  750. });
  751. res.status(200).json({ result: response });
  752. } catch (error) {
  753. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  754. }
  755. }
  756. });
  757. //获取整个文件夹的关系
  758. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  759. if (req.body[0]) {
  760. try {
  761. await client.graphs.pull({
  762. collectionId: req.body[0].collectionId,
  763. });
  764. const response = await client.graphs.listRelationships({
  765. collectionId: req.body[0].collectionId,
  766. offset: req.body[0].page ? req.body[0].page : 0,
  767. limit: 1000
  768. });
  769. res.status(200).json({ result: response });
  770. } catch (error) {
  771. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  772. }
  773. }
  774. });
  775. //提取知识图谱
  776. router.route("/getGraphs").all(async (req, res, next) => {
  777. console.log(req.body);
  778. //mode的处理
  779. if (req.body[0]) {
  780. try {
  781. //创建临时文件夹
  782. const folder = await client.collections.create({
  783. name: "临时文件夹",
  784. description: "临时文件夹"
  785. });
  786. let folderid = folder.results.id
  787. console.log(folder.results.id);
  788. //将要提取的文件添加到临时文件夹
  789. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  790. await Promise.all(documentids.map(documentid =>
  791. client.documents.extract({
  792. id: documentid,
  793. runWithOrchestration: false
  794. }).catch(error => {
  795. console.error("提取文件发生错误:", error);
  796. })
  797. ));
  798. // 使用 Promise.all 确保所有添加操作都被执行
  799. await Promise.all(documentids.map(documentid =>
  800. client.collections.addDocument({
  801. id: folderid,
  802. documentId: documentid
  803. }).catch(error => {
  804. console.error("移动文件时发生错误:", error);
  805. })
  806. ));
  807. //将临时文件夹提取知识图谱
  808. await client.graphs.pull({
  809. collectionId: folderid
  810. });
  811. let offset = 0;
  812. let results = [];
  813. let totalEntries = 0;
  814. do {
  815. const response = await client.graphs.listRelationships({
  816. collectionId: folderid,
  817. offset: offset,
  818. limit: 1000
  819. });
  820. results = results.concat(response.results);
  821. totalEntries = response.totalEntries;
  822. offset += 1000;
  823. } while (offset < totalEntries);
  824. let offset2 = 0;
  825. let results2 = [];
  826. let totalEntries2 = 0;
  827. do {
  828. const response = await client.graphs.listEntities({
  829. collectionId: folderid,
  830. offset: offset2,
  831. limit: 1000
  832. });
  833. results2 = results2.concat(response.results);
  834. totalEntries2 = response.totalEntries;
  835. offset2 += 1000;
  836. } while (offset2 < totalEntries2);
  837. //移除临时文件夹的文件
  838. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  839. await Promise.all(files.results.map(file =>
  840. client.collections.removeDocument({ id: folderid, documentId: file.id })
  841. ));
  842. //删除临时文件夹
  843. await client.collections.delete({ id: folderid });
  844. res.status(200).json({ result: results, result2: results2 });
  845. } catch (error) {
  846. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  847. }
  848. }
  849. });
  850. //提取知识图谱
  851. router.route("/getGraphs2").all(async (req, res, next) => {
  852. console.log(req.body);
  853. //mode的处理
  854. if (req.body[0]) {
  855. try {
  856. let documentIds = decodeURIComponent(req.body[0].fileids).split(",");
  857. // 使用 Promise.all 并通过 map 返回新的 results 数组
  858. let results = await Promise.all(documentIds.map(async documentId => {
  859. try {
  860. const response = await client.documents.listRelationships({
  861. id: documentId,
  862. includeVectors: false,
  863. offset: 0,
  864. limit: 1000, // 修正这里的参数
  865. });
  866. return response.results;
  867. } catch (error) {
  868. console.error(`Error fetching relationships for documentId ${documentId}:`, error);
  869. return []; // 返回空数组
  870. }
  871. }));
  872. // 使用 flat() 合并所有数组
  873. results = results.flat();
  874. res.status(200).json({ result: results });
  875. } catch (error) {
  876. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  877. }
  878. }
  879. });
  880. //获取实体和关系
  881. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  882. if (req.body[0]) {
  883. try {
  884. const { page = 1, limit = 10, documentId } = req.body[0];
  885. const offset = (page - 1) * limit;
  886. const response = await client.documents.listEntities({
  887. id: documentId,
  888. includeVectors: false,
  889. offset,
  890. limit,
  891. });
  892. const response2 = await client.documents.listRelationships({
  893. id: documentId,
  894. includeVectors: false,
  895. offset,
  896. limit,
  897. });
  898. res.status(200).json({ Entities: response, Relationships: response2 });
  899. } catch (error) {
  900. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  901. }
  902. }
  903. });
  904. //提取功能
  905. router.route("/extractFile2").all(async (req, res, next) => {
  906. try {
  907. let setting = {}
  908. if(req.body[0].prompt){
  909. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  910. }
  911. console.log(setting);
  912. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  913. console.log(response);
  914. res.status(200).json({ result: response.results });
  915. } catch (error) {
  916. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  917. }
  918. });
  919. //复制文件夹
  920. router.route("/copyFolder").all(async (req, res, next) => {
  921. if (req.body[0]) {
  922. const response = await client.collections.create({
  923. name: decodeURIComponent(req.body[0].n),
  924. description: decodeURIComponent(req.body[0].fd)
  925. });
  926. let folderid = response.results.id;
  927. p2 = [req.body[0].folderid]
  928. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  929. mysql.usselect(p2, async function (ret2) {
  930. let file = ret2[0]
  931. if (file.length) {
  932. let documentids = file.map(item => item.documentid);
  933. // 使用 Promise.all 确保所有添加操作都被执行
  934. await Promise.all(documentids.map(documentid =>
  935. client.collections.addDocument({
  936. id: folderid,
  937. documentId: documentid
  938. }).catch(error => {
  939. console.error("移动文件时发生错误:", error);
  940. })
  941. ));
  942. }
  943. let ids = file.map(item => item.fileid);
  944. console.log(ids);
  945. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  946. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  947. mysql.usselect(p, async function (ret) {
  948. res.end(JSON.stringify(ret));
  949. });
  950. });
  951. }
  952. });
  953. // 下载文件功能
  954. router.route("/downloadFile").all(async (req, res, next) => {
  955. var json = queryString(req.url);
  956. let params = {
  957. documentid: (json && json['documentid'] || req.body[0].documentid),
  958. filename: (json && json['filename'] || req.body[0].filename),
  959. };
  960. try {
  961. const response = await client.documents.download({ id: params.documentid });
  962. console.log(response);
  963. // 根据response的类型设置Content-Type
  964. const contentType = response.type || 'application/octet-stream';
  965. res.setHeader('Content-Type', contentType);
  966. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  967. // 直接发送blob数据
  968. res.send(Buffer.from(await response.arrayBuffer()));
  969. } catch (error) {
  970. console.error("下载文件时发生错误:", error);
  971. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  972. }
  973. });
  974. // 下载文件功能
  975. router.route("/downloadFile2").all(async (req, res, next) => {
  976. var json = queryString(req.url);
  977. let params = {
  978. documentid: (json && json['documentid'] || req.body[0].documentid),
  979. filename: (json && json['filename'] || req.body[0].filename),
  980. };
  981. try {
  982. const response = await client.documents.download({ id: params.documentid });
  983. console.log(response);
  984. // 直接返回blob数据
  985. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  986. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  987. const buffer = Buffer.from(await response.arrayBuffer());
  988. res.send(buffer); // 发送blob数据
  989. } catch (error) {
  990. console.error("下载文件时发生错误:", error);
  991. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  992. }
  993. });
  994. //召回测试
  995. router.route("/ragFolder").all(async (req, res, next) => {
  996. let reqJson = req.body[0]
  997. if (reqJson) {
  998. reqJson.query = decodeURIComponent(reqJson.query)
  999. try {
  1000. let filters = {}
  1001. if (reqJson.type == 1) {
  1002. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  1003. } else if (reqJson.type == 2) {
  1004. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  1005. } else if (reqJson.type == 3) {
  1006. filters = { "collection_ids": { "$eq": reqJson.ids } }
  1007. }
  1008. console.log(111);
  1009. const response = await client.retrieval.search({
  1010. query: reqJson.query,
  1011. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  1012. // searchMode: 'basic',
  1013. searchSettings: {
  1014. useHybridSearch: reqJson.useType == 3,
  1015. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1016. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  1017. hybrid_settings: {
  1018. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1019. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1020. },
  1021. filters: filters,
  1022. limit: reqJson.limit ? reqJson.limit : 100,
  1023. },
  1024. ragGenerationConfig: {
  1025. stream: false,
  1026. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1027. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1028. }
  1029. });
  1030. // res.status(200).json({ result: response.results });
  1031. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1032. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1033. mysql.usselect(p, async function (ret) {
  1034. res.status(200).json({ result: response.results });
  1035. });
  1036. } catch (error) {
  1037. console.log(error);
  1038. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1039. }
  1040. }
  1041. });
  1042. //召回测试
  1043. router.route("/searchFolder").all(async (req, res, next) => {
  1044. let reqJson = req.body[0]
  1045. if (reqJson) {
  1046. try {
  1047. let filters = {}
  1048. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1049. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1050. filters = {
  1051. "$or": [
  1052. { "collection_ids": { "$in": folderids } },
  1053. { "document_id": { "$in": documentids } }
  1054. ]
  1055. };
  1056. let params = {
  1057. query: reqJson.query,
  1058. searchSettings: {
  1059. useHybridSearch: reqJson.useType == 3,
  1060. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1061. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1062. hybrid_settings: {
  1063. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1064. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1065. },
  1066. filters: filters,
  1067. limit: reqJson.limit ? reqJson.limit : 100,
  1068. },
  1069. ragGenerationConfig: {
  1070. stream: false,
  1071. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1072. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1073. }
  1074. }
  1075. // console.log(params);
  1076. const response = await client.retrieval.search(params);
  1077. res.status(200).json({ result: response.results });
  1078. } catch (error) {
  1079. console.log(error);
  1080. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1081. }
  1082. }
  1083. });
  1084. //召回测试
  1085. router.route("/searchFolder2").all(async (req, res, next) => {
  1086. let reqJson = req.body
  1087. if (Object.keys(reqJson).length) {
  1088. try {
  1089. let filters = {}
  1090. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  1091. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  1092. filters = {
  1093. "$or": [
  1094. { "collection_ids": { "$in": folderids } },
  1095. { "document_id": { "$in": documentids } }
  1096. ]
  1097. };
  1098. if(!folderids.length && !documentids.length){
  1099. res.status(200).json({ result: [] });
  1100. return
  1101. }
  1102. let params = {
  1103. query: reqJson.query,
  1104. searchSettings: {
  1105. useHybridSearch: reqJson.useType == 3,
  1106. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1107. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1108. hybrid_settings: {
  1109. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1110. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1111. },
  1112. filters: filters,
  1113. limit: reqJson.limit ? reqJson.limit : 100,
  1114. },
  1115. ragGenerationConfig: {
  1116. stream: false,
  1117. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1118. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1119. }
  1120. }
  1121. // console.log(params);
  1122. const response = await client.retrieval.search(params);
  1123. res.status(200).json({ result: response.results });
  1124. } catch (error) {
  1125. console.log(error);
  1126. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1127. }
  1128. }else{
  1129. res.status(200).json({ result: "请传参" });
  1130. }
  1131. });
  1132. //召回测试
  1133. router.route("/searchFolder3").all(async (req, res, next) => {
  1134. let reqJson = req.body[0]
  1135. if (reqJson) {
  1136. reqJson.query = decodeURIComponent(reqJson.query)
  1137. try {
  1138. let filters = {}
  1139. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1140. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1141. filters = {
  1142. "$or": [
  1143. { "collection_ids": { "$in": folderids } },
  1144. { "document_id": { "$in": documentids } }
  1145. ]
  1146. };
  1147. if(!folderids.length && !documentids.length){
  1148. res.status(200).json({ result: [] });
  1149. return
  1150. }
  1151. let params = {
  1152. query: reqJson.query,
  1153. searchSettings: {
  1154. useHybridSearch: reqJson.useType == 3,
  1155. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1156. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1157. hybrid_settings: {
  1158. fullTextWeight: reqJson.fullTextWeight,
  1159. semanticWeight: reqJson.semanticWeight,
  1160. },
  1161. filters: filters,
  1162. limit: reqJson.limit ? reqJson.limit : 100,
  1163. },
  1164. ragGenerationConfig: {
  1165. stream: false,
  1166. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1167. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1168. }
  1169. }
  1170. // console.log(params);
  1171. const response = await client.retrieval.search(params);
  1172. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1173. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1174. mysql.usselect(p, async function (ret) {
  1175. res.status(200).json({ result: response.results });
  1176. });
  1177. } catch (error) {
  1178. console.log(error);
  1179. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1180. }
  1181. }
  1182. });
  1183. //召回测试
  1184. router.route("/searchFolder32").all(async (req, res, next) => {
  1185. let reqJson = req.body[0]
  1186. if (reqJson) {
  1187. reqJson.query = decodeURIComponent(reqJson.query)
  1188. try {
  1189. let filters = {}
  1190. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1191. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1192. filters = {
  1193. "$or": [
  1194. { "collection_ids": { "$in": folderids } },
  1195. { "document_id": { "$in": documentids } }
  1196. ]
  1197. };
  1198. if(!folderids.length && !documentids.length){
  1199. res.status(200).json({ result: [] });
  1200. return
  1201. }
  1202. let params = {
  1203. query: reqJson.query,
  1204. searchSettings: {
  1205. useHybridSearch: reqJson.useType == 3,
  1206. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1207. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1208. hybrid_settings: {
  1209. fullTextWeight: reqJson.fullTextWeight,
  1210. semanticWeight: reqJson.semanticWeight,
  1211. },
  1212. filters: filters,
  1213. limit: reqJson.limit ? reqJson.limit : 100,
  1214. },
  1215. ragGenerationConfig: {
  1216. stream: false,
  1217. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1218. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1219. }
  1220. }
  1221. console.log(params);
  1222. const response = await client.retrieval.search(params);
  1223. res.status(200).json({ result: response.results });
  1224. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1225. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1226. // mysql.usselect(p, async function (ret) {
  1227. // });
  1228. } catch (error) {
  1229. console.log(error);
  1230. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1231. }
  1232. }
  1233. });
  1234. //获取召回测试数据
  1235. router.route("/getRagData").all(async (req, res, next) => {
  1236. //mode的处理
  1237. if (req.body[0]) {
  1238. p = Object.values(req.body[0]);
  1239. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1240. mysql.usselect(p, async function (ret) {
  1241. res.status(200).json(ret);
  1242. });
  1243. }
  1244. });
  1245. //删除召回测试数据
  1246. router.route("/delRagData").all(async (req, res, next) => {
  1247. //mode的处理
  1248. if (req.body[0]) {
  1249. p = Object.values(req.body[0]);
  1250. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1251. mysql.usselect(p, async function (ret) {
  1252. res.status(200).json(ret);
  1253. });
  1254. }
  1255. });
  1256. function generateAPIKey(username) {
  1257. let timeStamp = new Date().getTime();
  1258. let hashStr = timeStamp + username;
  1259. let digest = crypto
  1260. .createHash("sha256")
  1261. .update(hashStr, "utf8")
  1262. .digest("hex");
  1263. return digest;
  1264. }
  1265. queryString = function (url) {
  1266. var a = url.split("?");
  1267. var json = querystring.parse(a[1]);
  1268. return json;
  1269. };
  1270. //post存儲過程處理
  1271. postmysql = function (req, res, functionname) {
  1272. //mode的處理
  1273. if (req.body[0]) {
  1274. p = Object.values(req.body[0]);
  1275. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1276. //執行存儲過程
  1277. mysql.usselect(p, function (ret) {
  1278. res.end(JSON.stringify(ret));
  1279. });
  1280. }
  1281. };
  1282. //post存儲過程處理
  1283. postmysql2 = function (req, res, functionname) {
  1284. //mode的處理
  1285. if (req.body) {
  1286. p = Object.values(req.body)[0].split(",");
  1287. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1288. //執行存儲過程
  1289. mysql.usselect(p, function (ret) {
  1290. res.end(JSON.stringify(ret));
  1291. });
  1292. }
  1293. };
  1294. //post存儲過程處理
  1295. postmysql3 = function (req, res, functionname) {
  1296. //mode的處理
  1297. if (req.body) {
  1298. console.log(req);
  1299. p = Object.values(req.body);
  1300. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1301. //執行存儲過程
  1302. mysql.usselect(p, function (ret) {
  1303. res.end(JSON.stringify(ret));
  1304. });
  1305. }
  1306. };
  1307. //get獲取數據庫信息
  1308. getmysql = function (req, res, functionname, ...p) {
  1309. p = p || [];
  1310. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1311. //執行存儲過程
  1312. mysql.usselect(p, function (ret) {
  1313. res.end(JSON.stringify(ret));
  1314. }); //
  1315. };
  1316. //get獲取數據庫信息
  1317. getmysql2 = function (req, res, functionname, ...p) {
  1318. p = p || [];
  1319. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1320. //執行存儲過程
  1321. mysql.usselect(p, function (ret) {
  1322. res.end(JSON.stringify(ret));
  1323. }); //
  1324. };
  1325. //get獲取數據庫信息
  1326. getmysql3 = function (req, res, functionname, ...p) {
  1327. p = p || [];
  1328. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1329. //執行存儲過程
  1330. mysql.usselect(p, function (ret) {
  1331. res.end(JSON.stringify(ret));
  1332. }); //
  1333. };
  1334. //post存储过程处理
  1335. userpostmysql = function (req, res, functionname) {
  1336. //mode的处理
  1337. if (req.body.mode) {
  1338. p = req.body.mode.split(",");
  1339. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1340. //执行存储过程
  1341. mysql.usselect(p, function (ret) {
  1342. res.end(JSON.stringify(ret));
  1343. });
  1344. }
  1345. };
  1346. //get获取数据库信息
  1347. usergetmysql = function (req, res, functionname, ...p) {
  1348. p = p || [];
  1349. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1350. //执行存储过程
  1351. mysql.usselect(p, function (ret) {
  1352. res.end(JSON.stringify(ret));
  1353. }); //
  1354. };
  1355. //post存儲過程處理
  1356. postmysqlLocal = function (req, res) {
  1357. //mode的處理
  1358. if (req.body[0]) {
  1359. p = Object.values(req.body[0]);
  1360. p.unshift(_localhost[0], _localhost[1]);
  1361. //執行存儲過程
  1362. mysql.usselect(p, function (ret) {
  1363. res.end(JSON.stringify(ret));
  1364. });
  1365. }
  1366. };
  1367. postmysqlOnline = function (req, res) {
  1368. //mode的處理
  1369. if (req.body[0]) {
  1370. p = Object.values(req.body[0]);
  1371. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1372. //執行存儲過程
  1373. mysql.usselect(p, function (ret) {
  1374. res.end(JSON.stringify(ret));
  1375. });
  1376. }
  1377. };
  1378. // 新增测试接口,直接返回成功
  1379. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1380. // console.log('收到的 req.body:', req.body);
  1381. // console.log('收到的 req.files:', req.files);
  1382. // console.log('收到的 headers:', req.headers);
  1383. const requestUrl = req.body.url; // 前端传入 url
  1384. if (!requestUrl) {
  1385. return res.status(400).json({error: 'Missing url'});
  1386. }
  1387. const browser = await initBrowser();
  1388. const page = await browser.newPage();
  1389. try {
  1390. console.log("开始抓取页面:", requestUrl);
  1391. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1392. await page.goto(requestUrl);
  1393. const html = await page.content();
  1394. // 新增:HTML 转 Markdown
  1395. const turndownService = new TurndownService();
  1396. // 用 cheerio 只提取正文部分
  1397. const cheerioDoc = cheerio.load(html);
  1398. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1399. // 用 cheerio 过滤无关标签
  1400. const cheerioMain = cheerio.load(mainHtml || '');
  1401. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1402. mainHtml = cheerioMain.html();
  1403. const markdownContent = turndownService.turndown(mainHtml);
  1404. // 新增:解析 <title> 标签内容
  1405. let title = '';
  1406. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1407. if (titleMatch && titleMatch[1]) {
  1408. title = titleMatch[1].trim();
  1409. }
  1410. let finalTitle;
  1411. if (title && title.trim()) {
  1412. finalTitle = title + '.md';
  1413. } else {
  1414. finalTitle = 'untitled-' + uuidv4() + '.md';
  1415. }
  1416. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1417. // 解析用户传入的参数
  1418. const {
  1419. collection_ids,
  1420. id: customId,
  1421. metadata: customMetadata,
  1422. ingestion_mode,
  1423. run_with_orchestration
  1424. } = req.body;
  1425. // 处理 metadata 字段
  1426. let finalMetadata = customMetadata;
  1427. if (typeof finalMetadata === 'string') {
  1428. try {
  1429. finalMetadata = JSON.parse(finalMetadata);
  1430. } catch (e) {
  1431. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1432. }
  1433. }
  1434. // if (!finalMetadata) {
  1435. // finalMetadata = {
  1436. // title: title || uuidv4(),
  1437. // // source_url: url,
  1438. // content_type: "markdown",
  1439. // // scraped_at: new Date().toISOString(),
  1440. // // upload_id: Date.now() + '-' + Math.random()
  1441. // };
  1442. // }
  1443. // 处理 collection_ids 字段
  1444. let finalCollectionIds = collection_ids;
  1445. if (typeof finalCollectionIds === 'string') {
  1446. try {
  1447. finalCollectionIds = JSON.parse(finalCollectionIds);
  1448. } catch (e) {
  1449. // 不是合法 JSON 字符串就当作普通字符串
  1450. }
  1451. }
  1452. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1453. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1454. finalMetadata.collection_ids = finalCollectionIds;
  1455. }
  1456. // 补全 finalMetadata.title 和 content_type
  1457. if (!finalMetadata.title) {
  1458. finalMetadata.title = finalTitle;
  1459. }
  1460. finalMetadata.content_type = 'markdown';
  1461. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1462. const FormData = require('form-data');
  1463. const fs = require('fs');
  1464. const path = require('path');
  1465. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1466. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1467. const form = new FormData();
  1468. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1469. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1470. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1471. }
  1472. if (customId) {
  1473. form.append('id', customId);
  1474. }
  1475. form.append('metadata', JSON.stringify(finalMetadata));
  1476. form.append('ingestion_mode', ingestion_mode || 'fast');
  1477. form.append('run_with_orchestration', 'false');
  1478. // 打印上传参数,方便排查问题
  1479. console.log('Request Headers:', form.getHeaders());
  1480. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1481. const response = await axios.post(url, form, {
  1482. headers: {
  1483. ...form.getHeaders(),
  1484. // 'User-Agent': 'NodeClient/1.0'
  1485. }
  1486. });
  1487. fs.unlinkSync(tmpFilePath);
  1488. result = {
  1489. results: response.data.results,
  1490. title: finalTitle
  1491. }
  1492. res.json(result);
  1493. } catch (error) {
  1494. console.error('Message:', error.message);
  1495. console.error('Stack:', error.stack);
  1496. res.status(500).json({error: error.message});
  1497. } finally {
  1498. await page.close(); // 确保页面总是被关闭
  1499. }
  1500. });
  1501. module.exports = router;