pbl.js 52 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430
  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. router.route("/login").all(async (req, res, next) => {
  53. loginHandler(req, res)
  54. });
  55. router.route("/logout").all(async (req, res, next) => {
  56. try {
  57. const response = await client.users.logout();
  58. res.status(200).json({ result: response.results });
  59. } catch (error) {
  60. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  61. }
  62. });
  63. //获取文档列表
  64. router.route("/getDocument").all(async (req, res, next) => {
  65. try {
  66. const response = await client.documents.list({
  67. limit: 10,
  68. offset: 0,
  69. });
  70. console.log(response);
  71. res.status(200).json({ result: response.results });
  72. } catch (error) {
  73. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  74. }
  75. });
  76. //列出文档块
  77. router.route("/listChunks").all(async (req, res, next) => {
  78. try {
  79. const response = await client.documents.listChunks({
  80. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  81. });
  82. console.log(response);
  83. res.status(200).json({ result: response.results });
  84. } catch (error) {
  85. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  86. }
  87. });
  88. //提取实体和关系
  89. router.route("/extract").all(async (req, res, next) => {
  90. try {
  91. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  92. console.log(response);
  93. res.status(200).json({ result: response.results });
  94. } catch (error) {
  95. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  96. }
  97. });
  98. router.route("/selectUser").all((req, res, next) => {
  99. var json = queryString(req.url);
  100. getmysql(req, res, "selectUser", json["userid"]);
  101. });
  102. // 文件上传接口
  103. router.post('/upload', upload.single('file'), async (req, res) => {
  104. req.setEncoding('utf-8')
  105. try {
  106. console.log(req.file); // 打印上传的文件信息
  107. console.log(`File path: uploads/${req.file.filename}`);
  108. const path = require('path');
  109. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  110. console.log(filePath);
  111. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  112. const fs = require('fs');
  113. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  114. const response = await client.documents.create({
  115. file: file,
  116. metadata: { title: filename }
  117. });
  118. console.log(response);
  119. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  120. } catch (error) {
  121. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  122. }
  123. });
  124. //获取文档列表
  125. router.route("/getFile").all(async (req, res, next) => {
  126. //mode的处理
  127. if (req.body[0]) {
  128. p = Object.values(req.body[0]);
  129. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  130. mysql.usselect(p, async function (ret) {
  131. let data = ret[0];
  132. // const array = data.map(item => item.documentid);
  133. const array = data.map(item => item.documentid);
  134. try {
  135. const response = await client.documents.list({ ids: array });
  136. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  137. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  138. // console.log(response);
  139. let data2 = response.results.map(doc => {
  140. doc.documentid = doc.id;
  141. delete doc.id;
  142. return doc;
  143. });
  144. // console.log(data2);
  145. // 合并data和data2,根据documentid
  146. const mergedData = data.map(item => {
  147. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  148. return { ...item, ...correspondingDoc };
  149. });
  150. res.status(200).json({ result: mergedData });
  151. } catch (error) {
  152. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  153. }
  154. });
  155. }
  156. });
  157. //获取文档列表
  158. router.route("/getFileD").all(async (req, res, next) => {
  159. //mode的处理
  160. if (req.body[0]) {
  161. p = Object.values(req.body[0]);
  162. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  163. mysql.usselect(p, async function (ret) {
  164. let data = ret[0];
  165. // const array = data.map(item => item.documentid);
  166. const array = data.map(item => item.documentid);
  167. try {
  168. const response = await client.documents.list({ ids: array });
  169. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  170. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  171. // console.log(response);
  172. let data2 = response.results.map(doc => {
  173. doc.documentid = doc.id;
  174. delete doc.id;
  175. return doc;
  176. });
  177. // console.log(data2);
  178. // 合并data和data2,根据documentid
  179. const mergedData = data.map(item => {
  180. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  181. return { ...item, ...correspondingDoc };
  182. });
  183. res.status(200).json({ result: mergedData });
  184. } catch (error) {
  185. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  186. }
  187. });
  188. }
  189. });
  190. //获取文档列表
  191. router.route("/getFileD2").all(async (req, res, next) => {
  192. //mode的处理
  193. if (req.body) {
  194. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  195. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  196. mysql.usselect(p, async function (ret) {
  197. let data = ret[0];
  198. // const array = data.map(item => item.documentid);
  199. const array = data.map(item => item.documentid);
  200. try {
  201. const response = await client.documents.list({ ids: array });
  202. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  203. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  204. // console.log(response);
  205. let data2 = response.results.map(doc => {
  206. doc.documentid = doc.id;
  207. delete doc.id;
  208. return doc;
  209. });
  210. // console.log(data2);
  211. // 合并data和data2,根据documentid
  212. const mergedData = data.map(item => {
  213. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  214. return { ...item, ...correspondingDoc };
  215. });
  216. res.status(200).json({ result: mergedData });
  217. } catch (error) {
  218. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  219. }
  220. });
  221. }
  222. });
  223. //获取文档列表
  224. router.route("/getFileOpen").all(async (req, res, next) => {
  225. //mode的处理
  226. if (req.body[0]) {
  227. p = Object.values(req.body[0]);
  228. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  229. mysql.usselect(p, async function (ret) {
  230. let data = ret[0];
  231. let user = ret[1];
  232. // const array = data.map(item => item.documentid);
  233. const array = data.map(item => item.documentid);
  234. try {
  235. const response = await client.documents.list({ ids: array });
  236. let data2 = response.results.map(doc => {
  237. doc.documentid = doc.id;
  238. delete doc.id;
  239. return doc;
  240. });
  241. // 合并data和data2,根据documentid
  242. const mergedData = data.map(item => {
  243. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  244. return { ...item, ...correspondingDoc };
  245. });
  246. res.status(200).json({ result: mergedData, user: user });
  247. } catch (error) {
  248. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  249. }
  250. });
  251. }
  252. });
  253. //获取文档列表2
  254. router.route("/getFileP").all(async (req, res, next) => {
  255. //mode的处理
  256. if (req.body) {
  257. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  258. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  259. mysql.usselect(p, async function (ret) {
  260. let data = ret[0];
  261. // const array = data.map(item => item.documentid);
  262. try {
  263. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  264. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  265. // console.log(response);
  266. let data2 = response.results.map(doc => {
  267. doc.documentid = doc.id;
  268. delete doc.id;
  269. return doc;
  270. });
  271. // 合并data和data2,根据documentid
  272. const mergedData = data.map(item => {
  273. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  274. return { ...item, ...correspondingDoc };
  275. });
  276. res.status(200).json({ result: mergedData });
  277. } catch (error) {
  278. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  279. }
  280. });
  281. }
  282. });
  283. //获取能关联的文件
  284. router.route("/getFile2").all(async (req, res, next) => {
  285. //mode的处理
  286. if (req.body[0]) {
  287. p = Object.values(req.body[0]);
  288. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  289. mysql.usselect(p, async function (ret) {
  290. let data = ret[0];
  291. const array = data.map(item => item.documentid);
  292. try {
  293. const response = await client.documents.list({ ids: array });
  294. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  295. // console.log(response);
  296. let data2 = response.results.map(doc => {
  297. doc.documentid = doc.id;
  298. delete doc.id;
  299. return doc;
  300. });
  301. // 合并data和data2,根据documentid
  302. const mergedData = data.map(item => {
  303. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  304. return { ...item, ...correspondingDoc };
  305. });
  306. res.status(200).json({ result: mergedData });
  307. } catch (error) {
  308. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  309. }
  310. });
  311. }
  312. });
  313. //添加文件夹
  314. router.route("/addFolder").all(async (req, res, next) => {
  315. //mode的处理
  316. if (req.body[0]) {
  317. try {
  318. const response = await client.collections.create({
  319. name: decodeURIComponent(req.body[0].n),
  320. description: decodeURIComponent(req.body[0].fd)
  321. });
  322. p = [...Object.values(req.body[0]), response.results.id];
  323. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  324. mysql.usselect(p, async function (ret) {
  325. res.end(JSON.stringify(ret));
  326. });
  327. } catch (error) {
  328. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  329. }
  330. }
  331. });
  332. //添加文件夹
  333. router.route("/addFolderP").all(async (req, res, next) => {
  334. //mode的处理
  335. if (req.body) {
  336. try {
  337. const response = await client.collections.create({
  338. name: decodeURIComponent(req.body.n),
  339. description: decodeURIComponent(req.body.fd)
  340. });
  341. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  342. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  343. mysql.usselect(p, async function (ret) {
  344. res.end(JSON.stringify(ret));
  345. });
  346. } catch (error) {
  347. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  348. }
  349. }
  350. });
  351. //删除文件夹
  352. router.route("/deleteFolder").all(async (req, res, next) => {
  353. //mode的处理
  354. if (req.body[0]) {
  355. try {
  356. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  357. const ids = req.body[0].ids;
  358. // 并行处理文件夹下的文件删除和文件夹删除
  359. await Promise.all(folderids.map(async folderid => {
  360. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  361. await Promise.all(files.results.map(file =>
  362. client.collections.removeDocument({ id: folderid, documentId: file.id })
  363. ));
  364. }));
  365. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  366. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  367. mysql.usselect(params, async function (ret) {
  368. res.end(JSON.stringify(ret));
  369. });
  370. } catch (error) {
  371. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  372. }
  373. }
  374. });
  375. //修改文件夹
  376. router.route("/updateFolder").all(async (req, res, next) => {
  377. //mode的处理
  378. if (req.body[0]) {
  379. try {
  380. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  381. const response = await client.collections.update({
  382. id: folderid,
  383. name: decodeURIComponent(req.body[0].n),
  384. description: decodeURIComponent(req.body[0].d)
  385. });
  386. p = Object.values(folderData); // 获取剩余属性的键
  387. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  388. mysql.usselect(p, async function (ret) {
  389. res.end(JSON.stringify(ret));
  390. });
  391. } catch (error) {
  392. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  393. }
  394. }
  395. });
  396. //修改文件夹
  397. router.route("/updateFolder2").all(async (req, res, next) => {
  398. //mode的处理
  399. if (req.body[0]) {
  400. try {
  401. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  402. const response = await client.collections.update({
  403. id: folderid,
  404. name: decodeURIComponent(req.body[0].n),
  405. description: decodeURIComponent(req.body[0].d)
  406. });
  407. p = Object.values(folderData); // 获取剩余属性的键
  408. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  409. mysql.usselect(p, async function (ret) {
  410. res.end(JSON.stringify(ret));
  411. });
  412. } catch (error) {
  413. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  414. }
  415. }
  416. });
  417. //移除文件
  418. router.route("/removeFile").all(async (req, res, next) => {
  419. if (req.body[0]) {
  420. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  421. let folderid = req.body[0].folderid;
  422. let ids = req.body[0].ids;
  423. // 使用 Promise.all 确保所有删除操作都被执行
  424. const removePromises = documentids.map(documentid =>
  425. client.collections.removeDocument({ id: folderid, documentId: documentid })
  426. );
  427. // 不管client.documents报错没有也要执行removeFile
  428. Promise.all(removePromises).catch(error => {
  429. console.error("移除文件时发生错误:", error);
  430. });
  431. p = [ids, folderid];
  432. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  433. mysql.usselect(p, async function (ret) {
  434. res.end(JSON.stringify(ret));
  435. });
  436. }
  437. });
  438. //移入文件
  439. router.route("/moveFile").all(async (req, res, next) => {
  440. if (req.body[0]) {
  441. try {
  442. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  443. let folderid = req.body[0].folderid
  444. let ids = decodeURIComponent(req.body[0].ids).split(",")
  445. // 使用 Promise.all 确保所有添加操作都被执行
  446. await Promise.all(documentids.map(documentid =>
  447. client.collections.addDocument({
  448. id: folderid,
  449. documentId: documentid
  450. }).catch(error => {
  451. console.error("移动文件时发生错误:", error);
  452. })
  453. ));
  454. for (const id of ids) {
  455. p = [id, folderid];
  456. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  457. await mysql.usselect(p, async function (ret) { });
  458. }
  459. res.status(200).json({ messages: "移动成功" });
  460. } catch (error) {
  461. // 这里不返回错误,确保deleteFile执行完
  462. console.error("处理过程中发生错误:", error);
  463. }
  464. }
  465. });
  466. //删除文件
  467. router.route("/deleteFile").all(async (req, res, next) => {
  468. if (req.body[0]) {
  469. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  470. let ids = req.body[0].ids;
  471. // 使用 Promise.all 确保所有删除操作都被执行
  472. Promise.all(documentids.map(documentid =>
  473. client.documents.delete({ id: documentid }).catch(error => {
  474. console.error("删除文件时发生错误:", error);
  475. })
  476. ));
  477. p = [ids];
  478. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  479. mysql.usselect(p, async function (ret) {
  480. res.end(JSON.stringify(ret));
  481. });
  482. }
  483. });
  484. //删除文件
  485. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  486. if (req.body[0]) {
  487. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  488. let ids = req.body[0].ids;
  489. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  490. console.log("文件删除成功");
  491. }).catch(error => {
  492. console.error("删除文件时发生错误:", error);
  493. })
  494. p = [ids];
  495. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  496. mysql.usselect(p, async function (ret) {
  497. res.end(JSON.stringify(ret));
  498. });
  499. }
  500. });
  501. router.route("/getFolder").all((req, res, next) => {
  502. getmysql2(req, res, "getFolder", req.body.uid);
  503. });
  504. router.route("/getFolder2").all((req, res, next) => {
  505. var json = queryString(req.url);
  506. getmysql2(req, res, "getFolder", json['uid']);
  507. });
  508. router.route("/getFolderPage").all((req, res, next) => {
  509. 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);
  510. });
  511. router.route("/getFolderPage3").all((req, res, next) => {
  512. 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);
  513. });
  514. router.route("/getFolderPage32").all((req, res, next) => {
  515. var json = queryString(req.url);
  516. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  517. });
  518. router.route("/getFolderPageOpen").all((req, res, next) => {
  519. 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);
  520. });
  521. router.route("/haveFile").all((req, res, next) => {
  522. postmysql(req, res, "haveFile");
  523. });
  524. router.route("/haveFolder").all((req, res, next) => {
  525. postmysql(req, res, "haveFolder");
  526. });
  527. //search文件
  528. router.route("/searchFile").all(async (req, res, next) => {
  529. if (req.body) {
  530. try {
  531. let filters = {}
  532. if (req.body.type == 1) {
  533. filters = { "document_id": { "$in": req.body.ids } }
  534. } else if (req.body.type == 2) {
  535. filters = { "collection_ids": { "$in": req.body.ids } }
  536. }
  537. const response = await client.retrieval.search({
  538. query: req.body.query,
  539. search_settings: {
  540. filters: filters,
  541. useSemanticSearch: true,
  542. chunkSettings: {
  543. limit: req.body.limit ? req.body.limit : 10,
  544. enabled: true
  545. },
  546. }
  547. });
  548. res.status(200).json({ result: response.results });
  549. } catch (error) {
  550. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  551. }
  552. }
  553. });
  554. //提取文件实体
  555. router.route("/extractFile").all(async (req, res, next) => {
  556. if (req.body) {
  557. try {
  558. const response = client.documents.extract({
  559. id: req.body.id
  560. })
  561. res.status(200).json({ result: response.results });
  562. } catch (error) {
  563. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  564. }
  565. }
  566. });
  567. //添加上传好的知识库文件
  568. router.route("/addFile").all((req, res, next) => {
  569. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  570. })
  571. //获取文件详情
  572. router.route("/getFileDetail").all(async (req, res, next) => {
  573. if (req.body[0]) {
  574. p = Object.values(req.body[0]);
  575. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  576. mysql.usselect(p, async function (ret) {
  577. let data = ret[0][0];
  578. // const array = data.map(item => item.documentid);
  579. try {
  580. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  581. // console.log(response);
  582. let data2 = response.results[0]
  583. // 合并data和data2,根据documentid
  584. const mergedData = { ...data, ...data2 };
  585. res.status(200).json({ result: mergedData });
  586. } catch (error) {
  587. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  588. }
  589. });
  590. }
  591. });
  592. //获取切片
  593. router.route("/getChunks").all(async (req, res, next) => {
  594. if (req.body[0]) {
  595. try {
  596. const { page = 1, limit = 10, documentId } = req.body[0];
  597. const offset = (page - 1) * limit;
  598. const response = await client.documents.listChunks({
  599. id: documentId,
  600. includeVectors: false,
  601. offset,
  602. limit,
  603. });
  604. res.status(200).json({ result: response });
  605. } catch (error) {
  606. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  607. }
  608. }
  609. });
  610. //获取实体
  611. router.route("/getEntities").all(async (req, res, next) => {
  612. if (req.body[0]) {
  613. try {
  614. const { page = 1, limit = 10, documentId } = req.body[0];
  615. const offset = (page - 1) * limit;
  616. const response = await client.documents.listEntities({
  617. id: documentId,
  618. includeVectors: false,
  619. offset,
  620. limit,
  621. });
  622. res.status(200).json({ result: response });
  623. } catch (error) {
  624. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  625. }
  626. }
  627. });
  628. //获取关系
  629. router.route("/getRelationships").all(async (req, res, next) => {
  630. if (req.body[0]) {
  631. try {
  632. const { page = 1, limit = 10, documentId } = req.body[0];
  633. const offset = (page - 1) * limit;
  634. const response = await client.documents.listRelationships({
  635. id: documentId,
  636. includeVectors: false,
  637. offset,
  638. limit,
  639. });
  640. res.status(200).json({ result: response });
  641. } catch (error) {
  642. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  643. }
  644. }
  645. });
  646. //获取实体和关系
  647. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  648. if (req.body[0]) {
  649. try {
  650. const { documentId } = req.body[0];
  651. const offset = (page - 1) * limit;
  652. const response = await client.documents.listEntities({
  653. id: documentId,
  654. includeVectors: false,
  655. });
  656. const response2 = await client.documents.listRelationships({
  657. id: documentId,
  658. includeVectors: false,
  659. });
  660. res.status(200).json({ Entities: response, Relationships: response2 });
  661. } catch (error) {
  662. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  663. }
  664. }
  665. });
  666. //提取功能
  667. router.route("/extractFile2").all(async (req, res, next) => {
  668. try {
  669. const response = await client.documents.extract({ id: req.body[0].documentid });
  670. console.log(response);
  671. res.status(200).json({ result: response.results });
  672. } catch (error) {
  673. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  674. }
  675. });
  676. //复制文件夹
  677. router.route("/copyFolder").all(async (req, res, next) => {
  678. if (req.body[0]) {
  679. const response = await client.collections.create({
  680. name: decodeURIComponent(req.body[0].n),
  681. description: decodeURIComponent(req.body[0].fd)
  682. });
  683. let folderid = response.results.id;
  684. p2 = [req.body[0].folderid]
  685. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  686. mysql.usselect(p2, async function (ret2) {
  687. let file = ret2[0]
  688. if (file.length) {
  689. let documentids = file.map(item => item.documentid);
  690. // 使用 Promise.all 确保所有添加操作都被执行
  691. await Promise.all(documentids.map(documentid =>
  692. client.collections.addDocument({
  693. id: folderid,
  694. documentId: documentid
  695. }).catch(error => {
  696. console.error("移动文件时发生错误:", error);
  697. })
  698. ));
  699. }
  700. let ids = file.map(item => item.fileid);
  701. console.log(ids);
  702. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  703. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  704. mysql.usselect(p, async function (ret) {
  705. res.end(JSON.stringify(ret));
  706. });
  707. });
  708. }
  709. });
  710. // 下载文件功能
  711. router.route("/downloadFile").all(async (req, res, next) => {
  712. var json = queryString(req.url);
  713. let params = {
  714. documentid: (json && json['documentid'] || req.body[0].documentid),
  715. filename: (json && json['filename'] || req.body[0].filename),
  716. };
  717. try {
  718. const response = await client.documents.download({ id: params.documentid });
  719. console.log(response);
  720. // 根据response的类型设置Content-Type
  721. const contentType = response.type || 'application/octet-stream';
  722. res.setHeader('Content-Type', contentType);
  723. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  724. // 直接发送blob数据
  725. res.send(Buffer.from(await response.arrayBuffer()));
  726. } catch (error) {
  727. console.error("下载文件时发生错误:", error);
  728. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  729. }
  730. });
  731. // 下载文件功能
  732. router.route("/downloadFile2").all(async (req, res, next) => {
  733. var json = queryString(req.url);
  734. let params = {
  735. documentid: (json && json['documentid'] || req.body[0].documentid),
  736. filename: (json && json['filename'] || req.body[0].filename),
  737. };
  738. try {
  739. const response = await client.documents.download({ id: params.documentid });
  740. console.log(response);
  741. // 直接返回blob数据
  742. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  743. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  744. const buffer = Buffer.from(await response.arrayBuffer());
  745. res.send(buffer); // 发送blob数据
  746. } catch (error) {
  747. console.error("下载文件时发生错误:", error);
  748. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  749. }
  750. });
  751. //召回测试
  752. router.route("/ragFolder").all(async (req, res, next) => {
  753. let reqJson = req.body[0]
  754. if (reqJson) {
  755. reqJson.query = decodeURIComponent(reqJson.query)
  756. try {
  757. let filters = {}
  758. if (reqJson.type == 1) {
  759. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  760. } else if (reqJson.type == 2) {
  761. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  762. } else if (reqJson.type == 3) {
  763. filters = { "collection_ids": { "$eq": reqJson.ids } }
  764. }
  765. console.log(111);
  766. const response = await client.retrieval.search({
  767. query: reqJson.query,
  768. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  769. // searchMode: 'basic',
  770. searchSettings: {
  771. useHybridSearch: reqJson.useType == 3,
  772. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  773. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  774. hybrid_settings: {
  775. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  776. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  777. },
  778. filters: filters,
  779. limit: reqJson.limit ? reqJson.limit : 100,
  780. },
  781. ragGenerationConfig: {
  782. stream: false,
  783. temperature: reqJson.temperature ? reqJson.temperature : 0,
  784. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  785. }
  786. });
  787. // res.status(200).json({ result: response.results });
  788. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  789. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  790. mysql.usselect(p, async function (ret) {
  791. res.status(200).json({ result: response.results });
  792. });
  793. } catch (error) {
  794. console.log(error);
  795. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  796. }
  797. }
  798. });
  799. //召回测试
  800. router.route("/searchFolder").all(async (req, res, next) => {
  801. let reqJson = req.body[0]
  802. if (reqJson) {
  803. try {
  804. let filters = {}
  805. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  806. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  807. filters = {
  808. "$or": [
  809. { "collection_ids": { "$in": folderids } },
  810. { "document_id": { "$in": documentids } }
  811. ]
  812. };
  813. let params = {
  814. query: reqJson.query,
  815. searchSettings: {
  816. useHybridSearch: reqJson.useType == 3,
  817. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  818. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  819. hybrid_settings: {
  820. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  821. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  822. },
  823. filters: filters,
  824. limit: reqJson.limit ? reqJson.limit : 100,
  825. },
  826. ragGenerationConfig: {
  827. stream: false,
  828. temperature: reqJson.temperature ? reqJson.temperature : 0,
  829. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  830. }
  831. }
  832. // console.log(params);
  833. const response = await client.retrieval.search(params);
  834. res.status(200).json({ result: response.results });
  835. } catch (error) {
  836. console.log(error);
  837. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  838. }
  839. }
  840. });
  841. //召回测试
  842. router.route("/searchFolder2").all(async (req, res, next) => {
  843. let reqJson = req.body
  844. if (reqJson) {
  845. try {
  846. let filters = {}
  847. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  848. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  849. filters = {
  850. "$or": [
  851. { "collection_ids": { "$in": folderids } },
  852. { "document_id": { "$in": documentids } }
  853. ]
  854. };
  855. let params = {
  856. query: reqJson.query,
  857. searchSettings: {
  858. useHybridSearch: reqJson.useType == 3,
  859. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  860. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  861. hybrid_settings: {
  862. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  863. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  864. },
  865. filters: filters,
  866. limit: reqJson.limit ? reqJson.limit : 100,
  867. },
  868. ragGenerationConfig: {
  869. stream: false,
  870. temperature: reqJson.temperature ? reqJson.temperature : 0,
  871. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  872. }
  873. }
  874. // console.log(params);
  875. const response = await client.retrieval.search(params);
  876. res.status(200).json({ result: response.results });
  877. } catch (error) {
  878. console.log(error);
  879. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  880. }
  881. }
  882. });
  883. //召回测试
  884. router.route("/searchFolder3").all(async (req, res, next) => {
  885. let reqJson = req.body[0]
  886. if (reqJson) {
  887. reqJson.query = decodeURIComponent(reqJson.query)
  888. try {
  889. let filters = {}
  890. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  891. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  892. filters = {
  893. "$or": [
  894. { "collection_ids": { "$in": folderids } },
  895. { "document_id": { "$in": documentids } }
  896. ]
  897. };
  898. let params = {
  899. query: reqJson.query,
  900. searchSettings: {
  901. useHybridSearch: reqJson.useType == 3,
  902. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  903. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  904. hybrid_settings: {
  905. fullTextWeight: reqJson.fullTextWeight,
  906. semanticWeight: reqJson.semanticWeight,
  907. },
  908. filters: filters,
  909. limit: reqJson.limit ? reqJson.limit : 100,
  910. },
  911. ragGenerationConfig: {
  912. stream: false,
  913. temperature: reqJson.temperature ? reqJson.temperature : 0,
  914. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  915. }
  916. }
  917. // console.log(params);
  918. const response = await client.retrieval.search(params);
  919. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  920. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  921. mysql.usselect(p, async function (ret) {
  922. res.status(200).json({ result: response.results });
  923. });
  924. } catch (error) {
  925. console.log(error);
  926. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  927. }
  928. }
  929. });
  930. //召回测试
  931. router.route("/searchFolder32").all(async (req, res, next) => {
  932. let reqJson = req.body[0]
  933. if (reqJson) {
  934. reqJson.query = decodeURIComponent(reqJson.query)
  935. try {
  936. let filters = {}
  937. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  938. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  939. filters = {
  940. "$or": [
  941. { "collection_ids": { "$in": folderids } },
  942. { "document_id": { "$in": documentids } }
  943. ]
  944. };
  945. let params = {
  946. query: reqJson.query,
  947. searchSettings: {
  948. useHybridSearch: reqJson.useType == 3,
  949. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  950. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  951. hybrid_settings: {
  952. fullTextWeight: reqJson.fullTextWeight,
  953. semanticWeight: reqJson.semanticWeight,
  954. },
  955. filters: filters,
  956. limit: reqJson.limit ? reqJson.limit : 100,
  957. },
  958. ragGenerationConfig: {
  959. stream: false,
  960. temperature: reqJson.temperature ? reqJson.temperature : 0,
  961. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  962. }
  963. }
  964. // console.log(params);
  965. const response = await client.retrieval.search(params);
  966. res.status(200).json({ result: response.results });
  967. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  968. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  969. // mysql.usselect(p, async function (ret) {
  970. // });
  971. } catch (error) {
  972. console.log(error);
  973. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  974. }
  975. }
  976. });
  977. //获取召回测试数据
  978. router.route("/getRagData").all(async (req, res, next) => {
  979. //mode的处理
  980. if (req.body[0]) {
  981. p = Object.values(req.body[0]);
  982. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  983. mysql.usselect(p, async function (ret) {
  984. res.status(200).json(ret);
  985. });
  986. }
  987. });
  988. //删除召回测试数据
  989. router.route("/delRagData").all(async (req, res, next) => {
  990. //mode的处理
  991. if (req.body[0]) {
  992. p = Object.values(req.body[0]);
  993. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  994. mysql.usselect(p, async function (ret) {
  995. res.status(200).json(ret);
  996. });
  997. }
  998. });
  999. function generateAPIKey(username) {
  1000. let timeStamp = new Date().getTime();
  1001. let hashStr = timeStamp + username;
  1002. let digest = crypto
  1003. .createHash("sha256")
  1004. .update(hashStr, "utf8")
  1005. .digest("hex");
  1006. return digest;
  1007. }
  1008. queryString = function (url) {
  1009. var a = url.split("?");
  1010. var json = querystring.parse(a[1]);
  1011. return json;
  1012. };
  1013. //post存儲過程處理
  1014. postmysql = function (req, res, functionname) {
  1015. //mode的處理
  1016. if (req.body[0]) {
  1017. p = Object.values(req.body[0]);
  1018. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1019. //執行存儲過程
  1020. mysql.usselect(p, function (ret) {
  1021. res.end(JSON.stringify(ret));
  1022. });
  1023. }
  1024. };
  1025. //post存儲過程處理
  1026. postmysql2 = function (req, res, functionname) {
  1027. //mode的處理
  1028. if (req.body) {
  1029. p = Object.values(req.body)[0].split(",");
  1030. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1031. //執行存儲過程
  1032. mysql.usselect(p, function (ret) {
  1033. res.end(JSON.stringify(ret));
  1034. });
  1035. }
  1036. };
  1037. //post存儲過程處理
  1038. postmysql3 = function (req, res, functionname) {
  1039. //mode的處理
  1040. if (req.body) {
  1041. console.log(req);
  1042. p = Object.values(req.body);
  1043. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1044. //執行存儲過程
  1045. mysql.usselect(p, function (ret) {
  1046. res.end(JSON.stringify(ret));
  1047. });
  1048. }
  1049. };
  1050. //get獲取數據庫信息
  1051. getmysql = function (req, res, functionname, ...p) {
  1052. p = p || [];
  1053. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1054. //執行存儲過程
  1055. mysql.usselect(p, function (ret) {
  1056. res.end(JSON.stringify(ret));
  1057. }); //
  1058. };
  1059. //get獲取數據庫信息
  1060. getmysql2 = function (req, res, functionname, ...p) {
  1061. p = p || [];
  1062. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1063. //執行存儲過程
  1064. mysql.usselect(p, function (ret) {
  1065. res.end(JSON.stringify(ret));
  1066. }); //
  1067. };
  1068. //get獲取數據庫信息
  1069. getmysql3 = function (req, res, functionname, ...p) {
  1070. p = p || [];
  1071. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1072. //執行存儲過程
  1073. mysql.usselect(p, function (ret) {
  1074. res.end(JSON.stringify(ret));
  1075. }); //
  1076. };
  1077. //post存储过程处理
  1078. userpostmysql = function (req, res, functionname) {
  1079. //mode的处理
  1080. if (req.body.mode) {
  1081. p = req.body.mode.split(",");
  1082. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1083. //执行存储过程
  1084. mysql.usselect(p, function (ret) {
  1085. res.end(JSON.stringify(ret));
  1086. });
  1087. }
  1088. };
  1089. //get获取数据库信息
  1090. usergetmysql = function (req, res, functionname, ...p) {
  1091. p = p || [];
  1092. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1093. //执行存储过程
  1094. mysql.usselect(p, function (ret) {
  1095. res.end(JSON.stringify(ret));
  1096. }); //
  1097. };
  1098. //post存儲過程處理
  1099. postmysqlLocal = function (req, res) {
  1100. //mode的處理
  1101. if (req.body[0]) {
  1102. p = Object.values(req.body[0]);
  1103. p.unshift(_localhost[0], _localhost[1]);
  1104. //執行存儲過程
  1105. mysql.usselect(p, function (ret) {
  1106. res.end(JSON.stringify(ret));
  1107. });
  1108. }
  1109. };
  1110. postmysqlOnline = function (req, res) {
  1111. //mode的處理
  1112. if (req.body[0]) {
  1113. p = Object.values(req.body[0]);
  1114. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1115. //執行存儲過程
  1116. mysql.usselect(p, function (ret) {
  1117. res.end(JSON.stringify(ret));
  1118. });
  1119. }
  1120. };
  1121. // 新增测试接口,直接返回成功
  1122. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1123. // console.log('收到的 req.body:', req.body);
  1124. // console.log('收到的 req.files:', req.files);
  1125. // console.log('收到的 headers:', req.headers);
  1126. const requestUrl = req.body.url; // 前端传入 url
  1127. if (!requestUrl) {
  1128. return res.status(400).json({error: 'Missing url'});
  1129. }
  1130. try {
  1131. const browser = await chromium.launch({headless: true});
  1132. const page = await browser.newPage();
  1133. await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1134. const html = await page.content();
  1135. await browser.close();
  1136. // 新增:HTML 转 Markdown
  1137. const turndownService = new TurndownService();
  1138. // 用 cheerio 只提取正文部分
  1139. const cheerioDoc = cheerio.load(html);
  1140. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1141. // 用 cheerio 过滤无关标签
  1142. const cheerioMain = cheerio.load(mainHtml || '');
  1143. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1144. mainHtml = cheerioMain.html();
  1145. const markdownContent = turndownService.turndown(mainHtml);
  1146. // 新增:解析 <title> 标签内容
  1147. let title = '';
  1148. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1149. if (titleMatch && titleMatch[1]) {
  1150. title = titleMatch[1].trim();
  1151. }
  1152. let finalTitle;
  1153. if (title && title.trim()) {
  1154. finalTitle = title + '.md';
  1155. } else {
  1156. finalTitle = 'untitled-' + uuidv4() + '.md';
  1157. }
  1158. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1159. // 解析用户传入的参数
  1160. const {
  1161. collection_ids,
  1162. id: customId,
  1163. metadata: customMetadata,
  1164. ingestion_mode,
  1165. run_with_orchestration
  1166. } = req.body;
  1167. // 处理 metadata 字段
  1168. let finalMetadata = customMetadata;
  1169. if (typeof finalMetadata === 'string') {
  1170. try {
  1171. finalMetadata = JSON.parse(finalMetadata);
  1172. } catch (e) {
  1173. return res.status(400).json({error: 'metadata 不是合法的 JSON 字符串'});
  1174. }
  1175. }
  1176. // if (!finalMetadata) {
  1177. // finalMetadata = {
  1178. // title: title || uuidv4(),
  1179. // // source_url: url,
  1180. // content_type: "markdown",
  1181. // // scraped_at: new Date().toISOString(),
  1182. // // upload_id: Date.now() + '-' + Math.random()
  1183. // };
  1184. // }
  1185. // 处理 collection_ids 字段
  1186. let finalCollectionIds = collection_ids;
  1187. if (typeof finalCollectionIds === 'string') {
  1188. try {
  1189. finalCollectionIds = JSON.parse(finalCollectionIds);
  1190. } catch (e) {
  1191. // 不是合法 JSON 字符串就当作普通字符串
  1192. }
  1193. }
  1194. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1195. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1196. finalMetadata.collection_ids = finalCollectionIds;
  1197. }
  1198. // 补全 finalMetadata.title 和 content_type
  1199. if (!finalMetadata.title) {
  1200. finalMetadata.title = finalTitle;
  1201. }
  1202. finalMetadata.content_type = 'markdown';
  1203. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1204. const FormData = require('form-data');
  1205. const fs = require('fs');
  1206. const path = require('path');
  1207. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1208. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1209. const form = new FormData();
  1210. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1211. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1212. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1213. }
  1214. if (customId) {
  1215. form.append('id', customId);
  1216. }
  1217. form.append('metadata', JSON.stringify(finalMetadata));
  1218. form.append('ingestion_mode', ingestion_mode || 'fast');
  1219. form.append('run_with_orchestration', 'false');
  1220. // 打印上传参数,方便排查问题
  1221. console.log('Request Headers:', form.getHeaders());
  1222. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1223. const response = await axios.post(url, form, {
  1224. headers: {
  1225. ...form.getHeaders(),
  1226. // 'User-Agent': 'NodeClient/1.0'
  1227. }
  1228. });
  1229. fs.unlinkSync(tmpFilePath);
  1230. result = {
  1231. results: response.data.results,
  1232. title: finalTitle
  1233. }
  1234. res.json(result);
  1235. } catch (error) {
  1236. console.error('Error occurred:', error);
  1237. res.status(500).json({error: error.message});
  1238. }
  1239. });
  1240. module.exports = router;