pbl.js 58 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582
  1. // ROUTES FOR OUR APIselectClassStudent
  2. // =============================================================================
  3. const express = require("express");
  4. const request = require("request");
  5. const bcrypt = require("bcryptjs");
  6. const router = express.Router(); // get an instance of the express Router
  7. const querystring = require("querystring");
  8. var mysql = require("./mysql");
  9. // const _mysqlLabor = ["183.36.25.93", "pbl"]; //edu數據庫信息
  10. // const _mysqluser = ["183.36.25.93", "cocorobouser"]; //用户数据库信息
  11. // const _getmysqlLabor = ["123.58.32.151", "pbl"]; //edu數據庫信息f
  12. // const _getmysqluser = ["123.58.32.151", "cocorobouser"]; //edu數據庫信息
  13. const _mysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  14. const _mysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  15. // const _getmysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  16. // const _getmysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  17. const _getmysqlLabor2 = ["172.16.12.7", "pbl"]; //edu數據庫信息
  18. const _getmysqlLabor = ["172.16.12.7", "pbl"]; //edu數據庫信息
  19. const _getmysqluser = ["172.16.12.7", "cocorobouser"]; //edu數據庫信息
  20. const _localhost = ["127.0.0.1", "pbl"]; //edu數據庫信息
  21. const crypto = require("crypto");
  22. const https = require("https");
  23. const { Pool } = require('pg');
  24. const axios = require('axios');
  25. const multer = require("multer");
  26. const {chromium} = require('playwright');
  27. const TurndownService = require('turndown');
  28. const cheerio = require('cheerio');
  29. const { v4: uuidv4 } = require('uuid');
  30. const loginHandler = require('./login'); // 确保路径正确
  31. const { r2rClient } = require("r2r-js");
  32. const client = new r2rClient("https://r2rserver.cocorobo.cn/");
  33. // 配置 multer 存储选项
  34. const storage = multer.diskStorage({
  35. destination: function (req, file, cb) {
  36. req.setEncoding('utf-8');
  37. cb(null, 'uploads/'); // 文件上传目录
  38. },
  39. filename: function (req, file, cb) {
  40. req.setEncoding('utf-8');
  41. // 使用 Buffer 处理文件名,确保编码正确
  42. const fileExtension = file.originalname.split('.').pop(); // 获取文件后缀
  43. const fileNameWithoutExt = Buffer.from(file.originalname.slice(0, -(fileExtension.length + 1)), 'latin1').toString('utf8'); // 转换编码
  44. cb(null, fileNameWithoutExt + '-' + Date.now() + '.' + fileExtension); // 使用时间戳和原始文件名
  45. }
  46. });
  47. const upload = multer({
  48. storage: storage, limits: {
  49. fileSize: 100 * 1024 * 1024 // 100MB
  50. }
  51. });
  52. // 全局浏览器实例
  53. let globalBrowser = null;
  54. async function initBrowser() {
  55. if (!globalBrowser) {
  56. globalBrowser = await chromium.launch({headless: true});
  57. console.log('Browser initialized');
  58. }
  59. return globalBrowser;
  60. }
  61. router.route("/login").all(async (req, res, next) => {
  62. loginHandler(req, res)
  63. });
  64. router.route("/logout").all(async (req, res, next) => {
  65. try {
  66. const response = await client.users.logout();
  67. res.status(200).json({ result: response.results });
  68. } catch (error) {
  69. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  70. }
  71. });
  72. //获取文档列表
  73. router.route("/getDocument").all(async (req, res, next) => {
  74. try {
  75. const response = await client.documents.list({
  76. limit: 10,
  77. offset: 0,
  78. });
  79. console.log(response);
  80. res.status(200).json({ result: response.results });
  81. } catch (error) {
  82. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  83. }
  84. });
  85. //列出文档块
  86. router.route("/listChunks").all(async (req, res, next) => {
  87. try {
  88. const response = await client.documents.listChunks({
  89. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  90. });
  91. console.log(response);
  92. res.status(200).json({ result: response.results });
  93. } catch (error) {
  94. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  95. }
  96. });
  97. //提取实体和关系
  98. router.route("/extract").all(async (req, res, next) => {
  99. try {
  100. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  101. console.log(response);
  102. res.status(200).json({ result: response.results });
  103. } catch (error) {
  104. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  105. }
  106. });
  107. router.route("/selectUser").all((req, res, next) => {
  108. var json = queryString(req.url);
  109. getmysql(req, res, "selectUser", json["userid"]);
  110. });
  111. // 文件上传接口
  112. router.post('/upload', upload.single('file'), async (req, res) => {
  113. req.setEncoding('utf-8')
  114. try {
  115. console.log(req.file); // 打印上传的文件信息
  116. console.log(`File path: uploads/${req.file.filename}`);
  117. const path = require('path');
  118. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  119. console.log(filePath);
  120. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  121. const fs = require('fs');
  122. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  123. const response = await client.documents.create({
  124. file: file,
  125. metadata: { title: filename }
  126. });
  127. console.log(response);
  128. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  129. } catch (error) {
  130. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  131. }
  132. });
  133. //获取文档列表
  134. router.route("/getFile").all(async (req, res, next) => {
  135. //mode的处理
  136. if (req.body[0]) {
  137. p = Object.values(req.body[0]);
  138. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  139. mysql.usselect(p, async function (ret) {
  140. let data = ret[0];
  141. // const array = data.map(item => item.documentid);
  142. const array = data.map(item => item.documentid);
  143. try {
  144. const response = await client.documents.list({ ids: array });
  145. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  146. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  147. // console.log(response);
  148. let data2 = response.results.map(doc => {
  149. doc.documentid = doc.id;
  150. delete doc.id;
  151. return doc;
  152. });
  153. // console.log(data2);
  154. // 合并data和data2,根据documentid
  155. const mergedData = data.map(item => {
  156. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  157. return { ...item, ...correspondingDoc };
  158. });
  159. res.status(200).json({ result: mergedData });
  160. } catch (error) {
  161. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  162. }
  163. });
  164. }
  165. });
  166. //获取文档列表
  167. router.route("/getFileR2r").all(async (req, res, next) => {
  168. //mode的处理
  169. if (req.body[0]) {
  170. let fileid = decodeURIComponent(req.body[0].fileids).split(",");
  171. try {
  172. const response = await client.documents.list({ ids: fileid });
  173. res.status(200).json({ result: response.results });
  174. } catch (error) {
  175. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  176. }
  177. }
  178. });
  179. //获取文档列表
  180. router.route("/getFileD").all(async (req, res, next) => {
  181. //mode的处理
  182. if (req.body[0]) {
  183. p = Object.values(req.body[0]);
  184. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  185. mysql.usselect(p, async function (ret) {
  186. let data = ret[0];
  187. // const array = data.map(item => item.documentid);
  188. const array = data.map(item => item.documentid);
  189. try {
  190. const response = await client.documents.list({ ids: array });
  191. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  192. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  193. // console.log(response);
  194. let data2 = response.results.map(doc => {
  195. doc.documentid = doc.id;
  196. delete doc.id;
  197. return doc;
  198. });
  199. // console.log(data2);
  200. // 合并data和data2,根据documentid
  201. const mergedData = data.map(item => {
  202. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  203. return { ...item, ...correspondingDoc };
  204. });
  205. res.status(200).json({ result: mergedData });
  206. } catch (error) {
  207. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  208. }
  209. });
  210. }
  211. });
  212. //获取文档列表
  213. router.route("/getFileD2").all(async (req, res, next) => {
  214. //mode的处理
  215. if (req.body) {
  216. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  217. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  218. mysql.usselect(p, async function (ret) {
  219. let data = ret[0];
  220. // const array = data.map(item => item.documentid);
  221. const array = data.map(item => item.documentid);
  222. try {
  223. const response = await client.documents.list({ ids: array });
  224. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  225. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  226. // console.log(response);
  227. let data2 = response.results.map(doc => {
  228. doc.documentid = doc.id;
  229. delete doc.id;
  230. return doc;
  231. });
  232. // console.log(data2);
  233. // 合并data和data2,根据documentid
  234. const mergedData = data.map(item => {
  235. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  236. return { ...item, ...correspondingDoc };
  237. });
  238. res.status(200).json({ result: mergedData });
  239. } catch (error) {
  240. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  241. }
  242. });
  243. }
  244. });
  245. //获取文档列表
  246. router.route("/getFileOpen").all(async (req, res, next) => {
  247. //mode的处理
  248. if (req.body[0]) {
  249. p = Object.values(req.body[0]);
  250. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  251. mysql.usselect(p, async function (ret) {
  252. let data = ret[0];
  253. let user = ret[1];
  254. // const array = data.map(item => item.documentid);
  255. const array = data.map(item => item.documentid);
  256. try {
  257. const response = await client.documents.list({ ids: array });
  258. let data2 = response.results.map(doc => {
  259. doc.documentid = doc.id;
  260. delete doc.id;
  261. return doc;
  262. });
  263. // 合并data和data2,根据documentid
  264. const mergedData = data.map(item => {
  265. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  266. return { ...item, ...correspondingDoc };
  267. });
  268. res.status(200).json({ result: mergedData, user: user });
  269. } catch (error) {
  270. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  271. }
  272. });
  273. }
  274. });
  275. //获取文档列表2
  276. router.route("/getFileP").all(async (req, res, next) => {
  277. //mode的处理
  278. if (req.body) {
  279. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  280. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  281. mysql.usselect(p, async function (ret) {
  282. let data = ret[0];
  283. // const array = data.map(item => item.documentid);
  284. try {
  285. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  286. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  287. // console.log(response);
  288. let data2 = response.results.map(doc => {
  289. doc.documentid = doc.id;
  290. delete doc.id;
  291. return doc;
  292. });
  293. // 合并data和data2,根据documentid
  294. const mergedData = data.map(item => {
  295. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  296. return { ...item, ...correspondingDoc };
  297. });
  298. res.status(200).json({ result: mergedData });
  299. } catch (error) {
  300. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  301. }
  302. });
  303. }
  304. });
  305. //获取能关联的文件
  306. router.route("/getFile2").all(async (req, res, next) => {
  307. //mode的处理
  308. if (req.body[0]) {
  309. p = Object.values(req.body[0]);
  310. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  311. mysql.usselect(p, async function (ret) {
  312. let data = ret[0];
  313. const array = data.map(item => item.documentid);
  314. try {
  315. const response = await client.documents.list({ ids: array });
  316. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  317. // console.log(response);
  318. let data2 = response.results.map(doc => {
  319. doc.documentid = doc.id;
  320. delete doc.id;
  321. return doc;
  322. });
  323. // 合并data和data2,根据documentid
  324. const mergedData = data.map(item => {
  325. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  326. return { ...item, ...correspondingDoc };
  327. });
  328. res.status(200).json({ result: mergedData });
  329. } catch (error) {
  330. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  331. }
  332. });
  333. }
  334. });
  335. //添加文件夹
  336. router.route("/addFolder").all(async (req, res, next) => {
  337. //mode的处理
  338. if (req.body[0]) {
  339. try {
  340. const response = await client.collections.create({
  341. name: decodeURIComponent(req.body[0].n),
  342. description: decodeURIComponent(req.body[0].fd)
  343. });
  344. p = [...Object.values(req.body[0]), response.results.id];
  345. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  346. mysql.usselect(p, async function (ret) {
  347. res.end(JSON.stringify(ret));
  348. });
  349. } catch (error) {
  350. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  351. }
  352. }
  353. });
  354. //添加文件夹
  355. router.route("/addFolderP").all(async (req, res, next) => {
  356. //mode的处理
  357. if (req.body) {
  358. try {
  359. const response = await client.collections.create({
  360. name: decodeURIComponent(req.body.n),
  361. description: decodeURIComponent(req.body.fd)
  362. });
  363. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  364. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  365. mysql.usselect(p, async function (ret) {
  366. res.end(JSON.stringify(ret));
  367. });
  368. } catch (error) {
  369. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  370. }
  371. }
  372. });
  373. //删除文件夹
  374. router.route("/deleteFolder").all(async (req, res, next) => {
  375. //mode的处理
  376. if (req.body[0]) {
  377. try {
  378. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  379. const ids = req.body[0].ids;
  380. // 并行处理文件夹下的文件删除和文件夹删除
  381. await Promise.all(folderids.map(async folderid => {
  382. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  383. await Promise.all(files.results.map(file =>
  384. client.collections.removeDocument({ id: folderid, documentId: file.id })
  385. ));
  386. }));
  387. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  388. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  389. mysql.usselect(params, async function (ret) {
  390. res.end(JSON.stringify(ret));
  391. });
  392. } catch (error) {
  393. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  394. }
  395. }
  396. });
  397. //修改文件夹
  398. router.route("/updateFolder").all(async (req, res, next) => {
  399. //mode的处理
  400. if (req.body[0]) {
  401. try {
  402. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  403. const response = await client.collections.update({
  404. id: folderid,
  405. name: decodeURIComponent(req.body[0].n),
  406. description: decodeURIComponent(req.body[0].d)
  407. });
  408. p = Object.values(folderData); // 获取剩余属性的键
  409. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  410. mysql.usselect(p, async function (ret) {
  411. res.end(JSON.stringify(ret));
  412. });
  413. } catch (error) {
  414. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  415. }
  416. }
  417. });
  418. //修改文件夹
  419. router.route("/updateFolder2").all(async (req, res, next) => {
  420. //mode的处理
  421. if (req.body[0]) {
  422. try {
  423. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  424. const response = await client.collections.update({
  425. id: folderid,
  426. name: decodeURIComponent(req.body[0].n),
  427. description: decodeURIComponent(req.body[0].d)
  428. });
  429. p = Object.values(folderData); // 获取剩余属性的键
  430. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  431. mysql.usselect(p, async function (ret) {
  432. res.end(JSON.stringify(ret));
  433. });
  434. } catch (error) {
  435. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  436. }
  437. }
  438. });
  439. //移除文件
  440. router.route("/removeFile").all(async (req, res, next) => {
  441. if (req.body[0]) {
  442. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  443. let folderid = req.body[0].folderid;
  444. let ids = req.body[0].ids;
  445. // 使用 Promise.all 确保所有删除操作都被执行
  446. const removePromises = documentids.map(documentid =>
  447. client.collections.removeDocument({ id: folderid, documentId: documentid })
  448. );
  449. // 不管client.documents报错没有也要执行removeFile
  450. Promise.all(removePromises).catch(error => {
  451. console.error("移除文件时发生错误:", error);
  452. });
  453. p = [ids, folderid];
  454. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  455. mysql.usselect(p, async function (ret) {
  456. res.end(JSON.stringify(ret));
  457. });
  458. }
  459. });
  460. //移入文件
  461. router.route("/moveFile").all(async (req, res, next) => {
  462. if (req.body[0]) {
  463. try {
  464. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  465. let folderid = req.body[0].folderid
  466. let ids = decodeURIComponent(req.body[0].ids).split(",")
  467. // 使用 Promise.all 确保所有添加操作都被执行
  468. await Promise.all(documentids.map(documentid =>
  469. client.collections.addDocument({
  470. id: folderid,
  471. documentId: documentid
  472. }).catch(error => {
  473. console.error("移动文件时发生错误:", error);
  474. })
  475. ));
  476. for (const id of ids) {
  477. p = [id, folderid];
  478. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  479. await mysql.usselect(p, async function (ret) { });
  480. }
  481. res.status(200).json({ messages: "移动成功" });
  482. } catch (error) {
  483. // 这里不返回错误,确保deleteFile执行完
  484. console.error("处理过程中发生错误:", error);
  485. }
  486. }
  487. });
  488. //删除文件
  489. router.route("/deleteFile").all(async (req, res, next) => {
  490. if (req.body[0]) {
  491. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  492. let ids = req.body[0].ids;
  493. // 使用 Promise.all 确保所有删除操作都被执行
  494. Promise.all(documentids.map(documentid =>
  495. client.documents.delete({ id: documentid }).catch(error => {
  496. console.error("删除文件时发生错误:", error);
  497. })
  498. ));
  499. p = [ids];
  500. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  501. mysql.usselect(p, async function (ret) {
  502. res.end(JSON.stringify(ret));
  503. });
  504. }
  505. });
  506. //删除文件
  507. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  508. if (req.body[0]) {
  509. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  510. let ids = req.body[0].ids;
  511. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  512. console.log("文件删除成功");
  513. }).catch(error => {
  514. console.error("删除文件时发生错误:", error);
  515. })
  516. p = [ids];
  517. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  518. mysql.usselect(p, async function (ret) {
  519. res.end(JSON.stringify(ret));
  520. });
  521. }
  522. });
  523. router.route("/getFolder").all((req, res, next) => {
  524. getmysql2(req, res, "getFolder", req.body.uid);
  525. });
  526. router.route("/getFolder2").all((req, res, next) => {
  527. var json = queryString(req.url);
  528. getmysql2(req, res, "getFolder", json['uid']);
  529. });
  530. router.route("/getFolderPage").all((req, res, next) => {
  531. 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);
  532. });
  533. router.route("/getFolderPage3").all((req, res, next) => {
  534. 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);
  535. });
  536. router.route("/getFolderPage32").all((req, res, next) => {
  537. var json = queryString(req.url);
  538. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  539. });
  540. router.route("/getFolderPageOpen").all((req, res, next) => {
  541. 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);
  542. });
  543. router.route("/haveFile").all((req, res, next) => {
  544. postmysql(req, res, "haveFile");
  545. });
  546. router.route("/haveFolder").all((req, res, next) => {
  547. postmysql(req, res, "haveFolder");
  548. });
  549. //search文件
  550. router.route("/searchFile").all(async (req, res, next) => {
  551. if (req.body) {
  552. try {
  553. let filters = {}
  554. if (req.body.type == 1) {
  555. filters = { "document_id": { "$in": req.body.ids } }
  556. } else if (req.body.type == 2) {
  557. filters = { "collection_ids": { "$in": req.body.ids } }
  558. }
  559. const response = await client.retrieval.search({
  560. query: req.body.query,
  561. search_settings: {
  562. filters: filters,
  563. useSemanticSearch: true,
  564. chunkSettings: {
  565. limit: req.body.limit ? req.body.limit : 10,
  566. enabled: true
  567. },
  568. }
  569. });
  570. res.status(200).json({ result: response.results });
  571. } catch (error) {
  572. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  573. }
  574. }
  575. });
  576. //提取文件实体
  577. router.route("/extractFile").all(async (req, res, next) => {
  578. if (req.body) {
  579. try {
  580. const response = client.documents.extract({
  581. id: req.body.id
  582. })
  583. res.status(200).json({ result: response.results });
  584. } catch (error) {
  585. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  586. }
  587. }
  588. });
  589. //添加上传好的知识库文件
  590. router.route("/addFile").all((req, res, next) => {
  591. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  592. })
  593. //获取文件详情
  594. router.route("/getFileDetail").all(async (req, res, next) => {
  595. if (req.body[0]) {
  596. p = Object.values(req.body[0]);
  597. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  598. mysql.usselect(p, async function (ret) {
  599. let data = ret[0][0];
  600. // const array = data.map(item => item.documentid);
  601. try {
  602. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  603. // console.log(response);
  604. let data2 = response.results[0]
  605. // 合并data和data2,根据documentid
  606. const mergedData = { ...data, ...data2 };
  607. res.status(200).json({ result: mergedData });
  608. } catch (error) {
  609. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  610. }
  611. });
  612. }
  613. });
  614. //获取切片
  615. router.route("/getChunks").all(async (req, res, next) => {
  616. if (req.body[0]) {
  617. try {
  618. const { page = 1, limit = 10, documentId } = req.body[0];
  619. const offset = (page - 1) * limit;
  620. const response = await client.documents.listChunks({
  621. id: documentId,
  622. includeVectors: false,
  623. offset,
  624. limit,
  625. });
  626. res.status(200).json({ result: response });
  627. } catch (error) {
  628. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  629. }
  630. }
  631. });
  632. //获取实体
  633. router.route("/getEntities").all(async (req, res, next) => {
  634. if (req.body[0]) {
  635. try {
  636. const { page = 1, limit = 10, documentId } = req.body[0];
  637. const offset = (page - 1) * limit;
  638. const response = await client.documents.listEntities({
  639. id: documentId,
  640. includeVectors: false,
  641. offset,
  642. limit,
  643. });
  644. res.status(200).json({ result: response });
  645. } catch (error) {
  646. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  647. }
  648. }
  649. });
  650. //获取关系
  651. router.route("/getRelationships").all(async (req, res, next) => {
  652. if (req.body[0]) {
  653. try {
  654. const { page = 1, limit = 10, documentId } = req.body[0];
  655. const offset = (page - 1) * limit;
  656. const response = await client.documents.listRelationships({
  657. id: documentId,
  658. includeVectors: false,
  659. offset,
  660. limit,
  661. });
  662. res.status(200).json({ result: response });
  663. } catch (error) {
  664. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  665. }
  666. }
  667. });
  668. //提取整个文件夹的关系
  669. router.route("/resetGraphs").all(async (req, res, next) => {
  670. if (req.body[0]) {
  671. try {
  672. // const response = await client.graphs.pull({
  673. // collectionId: req.body[0].collectionId,
  674. // });
  675. const response = await client.graphs.reset({
  676. collectionId: req.body[0].collectionId,
  677. });
  678. res.status(200).json({ result: response });
  679. } catch (error) {
  680. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  681. }
  682. }
  683. });
  684. //获取整个文件夹的关系
  685. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  686. if (req.body[0]) {
  687. try {
  688. await client.graphs.pull({
  689. collectionId: req.body[0].collectionId,
  690. });
  691. const response = await client.graphs.listRelationships({
  692. collectionId: req.body[0].collectionId,
  693. offset: req.body[0].page ? req.body[0].page : 0,
  694. limit: 1000
  695. });
  696. res.status(200).json({ result: response });
  697. } catch (error) {
  698. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  699. }
  700. }
  701. });
  702. //提取知识图谱
  703. router.route("/getGraphs").all(async (req, res, next) => {
  704. console.log(req.body);
  705. //mode的处理
  706. if (req.body[0]) {
  707. try {
  708. //创建临时文件夹
  709. const folder = await client.collections.create({
  710. name: "临时文件夹",
  711. description: "临时文件夹"
  712. });
  713. let folderid = folder.results.id
  714. console.log(folder.results.id);
  715. //将要提取的文件添加到临时文件夹
  716. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  717. // 使用 Promise.all 确保所有添加操作都被执行
  718. await Promise.all(documentids.map(documentid =>
  719. client.collections.addDocument({
  720. id: folderid,
  721. documentId: documentid
  722. }).catch(error => {
  723. console.error("移动文件时发生错误:", error);
  724. })
  725. ));
  726. //将临时文件夹提取知识图谱
  727. await client.graphs.pull({
  728. collectionId: folderid
  729. });
  730. let offset = 0;
  731. let results = [];
  732. let totalEntries = 0;
  733. do {
  734. const response = await client.graphs.listRelationships({
  735. collectionId: folderid,
  736. offset: offset,
  737. limit: 1000
  738. });
  739. results = results.concat(response.results);
  740. totalEntries = response.totalEntries;
  741. offset += 1000;
  742. } while (offset < totalEntries);
  743. //移除临时文件夹的文件
  744. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  745. await Promise.all(files.results.map(file =>
  746. client.collections.removeDocument({ id: folderid, documentId: file.id })
  747. ));
  748. //删除临时文件夹
  749. await client.collections.delete({ id: folderid });
  750. res.status(200).json({ result: results });
  751. } catch (error) {
  752. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  753. }
  754. }
  755. });
  756. //获取实体和关系
  757. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  758. if (req.body[0]) {
  759. try {
  760. const { documentId } = req.body[0];
  761. const offset = (page - 1) * limit;
  762. const response = await client.documents.listEntities({
  763. id: documentId,
  764. includeVectors: false,
  765. });
  766. const response2 = await client.documents.listRelationships({
  767. id: documentId,
  768. includeVectors: false,
  769. });
  770. res.status(200).json({ Entities: response, Relationships: response2 });
  771. } catch (error) {
  772. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  773. }
  774. }
  775. });
  776. //提取功能
  777. router.route("/extractFile2").all(async (req, res, next) => {
  778. try {
  779. let setting = {}
  780. if(req.body[0].prompt){
  781. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  782. }
  783. console.log(setting);
  784. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  785. console.log(response);
  786. res.status(200).json({ result: response.results });
  787. } catch (error) {
  788. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  789. }
  790. });
  791. //复制文件夹
  792. router.route("/copyFolder").all(async (req, res, next) => {
  793. if (req.body[0]) {
  794. const response = await client.collections.create({
  795. name: decodeURIComponent(req.body[0].n),
  796. description: decodeURIComponent(req.body[0].fd)
  797. });
  798. let folderid = response.results.id;
  799. p2 = [req.body[0].folderid]
  800. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  801. mysql.usselect(p2, async function (ret2) {
  802. let file = ret2[0]
  803. if (file.length) {
  804. let documentids = file.map(item => item.documentid);
  805. // 使用 Promise.all 确保所有添加操作都被执行
  806. await Promise.all(documentids.map(documentid =>
  807. client.collections.addDocument({
  808. id: folderid,
  809. documentId: documentid
  810. }).catch(error => {
  811. console.error("移动文件时发生错误:", error);
  812. })
  813. ));
  814. }
  815. let ids = file.map(item => item.fileid);
  816. console.log(ids);
  817. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  818. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  819. mysql.usselect(p, async function (ret) {
  820. res.end(JSON.stringify(ret));
  821. });
  822. });
  823. }
  824. });
  825. // 下载文件功能
  826. router.route("/downloadFile").all(async (req, res, next) => {
  827. var json = queryString(req.url);
  828. let params = {
  829. documentid: (json && json['documentid'] || req.body[0].documentid),
  830. filename: (json && json['filename'] || req.body[0].filename),
  831. };
  832. try {
  833. const response = await client.documents.download({ id: params.documentid });
  834. console.log(response);
  835. // 根据response的类型设置Content-Type
  836. const contentType = response.type || 'application/octet-stream';
  837. res.setHeader('Content-Type', contentType);
  838. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  839. // 直接发送blob数据
  840. res.send(Buffer.from(await response.arrayBuffer()));
  841. } catch (error) {
  842. console.error("下载文件时发生错误:", error);
  843. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  844. }
  845. });
  846. // 下载文件功能
  847. router.route("/downloadFile2").all(async (req, res, next) => {
  848. var json = queryString(req.url);
  849. let params = {
  850. documentid: (json && json['documentid'] || req.body[0].documentid),
  851. filename: (json && json['filename'] || req.body[0].filename),
  852. };
  853. try {
  854. const response = await client.documents.download({ id: params.documentid });
  855. console.log(response);
  856. // 直接返回blob数据
  857. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  858. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  859. const buffer = Buffer.from(await response.arrayBuffer());
  860. res.send(buffer); // 发送blob数据
  861. } catch (error) {
  862. console.error("下载文件时发生错误:", error);
  863. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  864. }
  865. });
  866. //召回测试
  867. router.route("/ragFolder").all(async (req, res, next) => {
  868. let reqJson = req.body[0]
  869. if (reqJson) {
  870. reqJson.query = decodeURIComponent(reqJson.query)
  871. try {
  872. let filters = {}
  873. if (reqJson.type == 1) {
  874. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  875. } else if (reqJson.type == 2) {
  876. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  877. } else if (reqJson.type == 3) {
  878. filters = { "collection_ids": { "$eq": reqJson.ids } }
  879. }
  880. console.log(111);
  881. const response = await client.retrieval.search({
  882. query: reqJson.query,
  883. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  884. // searchMode: 'basic',
  885. searchSettings: {
  886. useHybridSearch: reqJson.useType == 3,
  887. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  888. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  889. hybrid_settings: {
  890. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  891. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  892. },
  893. filters: filters,
  894. limit: reqJson.limit ? reqJson.limit : 100,
  895. },
  896. ragGenerationConfig: {
  897. stream: false,
  898. temperature: reqJson.temperature ? reqJson.temperature : 0,
  899. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  900. }
  901. });
  902. // res.status(200).json({ result: response.results });
  903. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  904. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  905. mysql.usselect(p, async function (ret) {
  906. res.status(200).json({ result: response.results });
  907. });
  908. } catch (error) {
  909. console.log(error);
  910. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  911. }
  912. }
  913. });
  914. //召回测试
  915. router.route("/searchFolder").all(async (req, res, next) => {
  916. let reqJson = req.body[0]
  917. if (reqJson) {
  918. try {
  919. let filters = {}
  920. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  921. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  922. filters = {
  923. "$or": [
  924. { "collection_ids": { "$in": folderids } },
  925. { "document_id": { "$in": documentids } }
  926. ]
  927. };
  928. let params = {
  929. query: reqJson.query,
  930. searchSettings: {
  931. useHybridSearch: reqJson.useType == 3,
  932. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  933. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  934. hybrid_settings: {
  935. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  936. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  937. },
  938. filters: filters,
  939. limit: reqJson.limit ? reqJson.limit : 100,
  940. },
  941. ragGenerationConfig: {
  942. stream: false,
  943. temperature: reqJson.temperature ? reqJson.temperature : 0,
  944. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  945. }
  946. }
  947. // console.log(params);
  948. const response = await client.retrieval.search(params);
  949. res.status(200).json({ result: response.results });
  950. } catch (error) {
  951. console.log(error);
  952. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  953. }
  954. }
  955. });
  956. //召回测试
  957. router.route("/searchFolder2").all(async (req, res, next) => {
  958. let reqJson = req.body
  959. if (Object.keys(reqJson).length) {
  960. try {
  961. let filters = {}
  962. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  963. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  964. filters = {
  965. "$or": [
  966. { "collection_ids": { "$in": folderids } },
  967. { "document_id": { "$in": documentids } }
  968. ]
  969. };
  970. if(!folderids.length && !documentids.length){
  971. res.status(200).json({ result: [] });
  972. return
  973. }
  974. let params = {
  975. query: reqJson.query,
  976. searchSettings: {
  977. useHybridSearch: reqJson.useType == 3,
  978. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  979. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  980. hybrid_settings: {
  981. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  982. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  983. },
  984. filters: filters,
  985. limit: reqJson.limit ? reqJson.limit : 100,
  986. },
  987. ragGenerationConfig: {
  988. stream: false,
  989. temperature: reqJson.temperature ? reqJson.temperature : 0,
  990. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  991. }
  992. }
  993. // console.log(params);
  994. const response = await client.retrieval.search(params);
  995. res.status(200).json({ result: response.results });
  996. } catch (error) {
  997. console.log(error);
  998. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  999. }
  1000. }else{
  1001. res.status(200).json({ result: "请传参" });
  1002. }
  1003. });
  1004. //召回测试
  1005. router.route("/searchFolder3").all(async (req, res, next) => {
  1006. let reqJson = req.body[0]
  1007. if (reqJson) {
  1008. reqJson.query = decodeURIComponent(reqJson.query)
  1009. try {
  1010. let filters = {}
  1011. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1012. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1013. filters = {
  1014. "$or": [
  1015. { "collection_ids": { "$in": folderids } },
  1016. { "document_id": { "$in": documentids } }
  1017. ]
  1018. };
  1019. if(!folderids.length && !documentids.length){
  1020. res.status(200).json({ result: [] });
  1021. return
  1022. }
  1023. let params = {
  1024. query: reqJson.query,
  1025. searchSettings: {
  1026. useHybridSearch: reqJson.useType == 3,
  1027. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1028. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1029. hybrid_settings: {
  1030. fullTextWeight: reqJson.fullTextWeight,
  1031. semanticWeight: reqJson.semanticWeight,
  1032. },
  1033. filters: filters,
  1034. limit: reqJson.limit ? reqJson.limit : 100,
  1035. },
  1036. ragGenerationConfig: {
  1037. stream: false,
  1038. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1039. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1040. }
  1041. }
  1042. // console.log(params);
  1043. const response = await client.retrieval.search(params);
  1044. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1045. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1046. mysql.usselect(p, async function (ret) {
  1047. res.status(200).json({ result: response.results });
  1048. });
  1049. } catch (error) {
  1050. console.log(error);
  1051. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1052. }
  1053. }
  1054. });
  1055. //召回测试
  1056. router.route("/searchFolder32").all(async (req, res, next) => {
  1057. let reqJson = req.body[0]
  1058. if (reqJson) {
  1059. reqJson.query = decodeURIComponent(reqJson.query)
  1060. try {
  1061. let filters = {}
  1062. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1063. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1064. filters = {
  1065. "$or": [
  1066. { "collection_ids": { "$in": folderids } },
  1067. { "document_id": { "$in": documentids } }
  1068. ]
  1069. };
  1070. if(!folderids.length && !documentids.length){
  1071. res.status(200).json({ result: [] });
  1072. return
  1073. }
  1074. let params = {
  1075. query: reqJson.query,
  1076. searchSettings: {
  1077. useHybridSearch: reqJson.useType == 3,
  1078. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1079. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1080. hybrid_settings: {
  1081. fullTextWeight: reqJson.fullTextWeight,
  1082. semanticWeight: reqJson.semanticWeight,
  1083. },
  1084. filters: filters,
  1085. limit: reqJson.limit ? reqJson.limit : 100,
  1086. },
  1087. ragGenerationConfig: {
  1088. stream: false,
  1089. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1090. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1091. }
  1092. }
  1093. console.log(params);
  1094. const response = await client.retrieval.search(params);
  1095. res.status(200).json({ result: response.results });
  1096. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1097. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1098. // mysql.usselect(p, async function (ret) {
  1099. // });
  1100. } catch (error) {
  1101. console.log(error);
  1102. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1103. }
  1104. }
  1105. });
  1106. //获取召回测试数据
  1107. router.route("/getRagData").all(async (req, res, next) => {
  1108. //mode的处理
  1109. if (req.body[0]) {
  1110. p = Object.values(req.body[0]);
  1111. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1112. mysql.usselect(p, async function (ret) {
  1113. res.status(200).json(ret);
  1114. });
  1115. }
  1116. });
  1117. //删除召回测试数据
  1118. router.route("/delRagData").all(async (req, res, next) => {
  1119. //mode的处理
  1120. if (req.body[0]) {
  1121. p = Object.values(req.body[0]);
  1122. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1123. mysql.usselect(p, async function (ret) {
  1124. res.status(200).json(ret);
  1125. });
  1126. }
  1127. });
  1128. function generateAPIKey(username) {
  1129. let timeStamp = new Date().getTime();
  1130. let hashStr = timeStamp + username;
  1131. let digest = crypto
  1132. .createHash("sha256")
  1133. .update(hashStr, "utf8")
  1134. .digest("hex");
  1135. return digest;
  1136. }
  1137. queryString = function (url) {
  1138. var a = url.split("?");
  1139. var json = querystring.parse(a[1]);
  1140. return json;
  1141. };
  1142. //post存儲過程處理
  1143. postmysql = function (req, res, functionname) {
  1144. //mode的處理
  1145. if (req.body[0]) {
  1146. p = Object.values(req.body[0]);
  1147. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1148. //執行存儲過程
  1149. mysql.usselect(p, function (ret) {
  1150. res.end(JSON.stringify(ret));
  1151. });
  1152. }
  1153. };
  1154. //post存儲過程處理
  1155. postmysql2 = function (req, res, functionname) {
  1156. //mode的處理
  1157. if (req.body) {
  1158. p = Object.values(req.body)[0].split(",");
  1159. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1160. //執行存儲過程
  1161. mysql.usselect(p, function (ret) {
  1162. res.end(JSON.stringify(ret));
  1163. });
  1164. }
  1165. };
  1166. //post存儲過程處理
  1167. postmysql3 = function (req, res, functionname) {
  1168. //mode的處理
  1169. if (req.body) {
  1170. console.log(req);
  1171. p = Object.values(req.body);
  1172. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1173. //執行存儲過程
  1174. mysql.usselect(p, function (ret) {
  1175. res.end(JSON.stringify(ret));
  1176. });
  1177. }
  1178. };
  1179. //get獲取數據庫信息
  1180. getmysql = function (req, res, functionname, ...p) {
  1181. p = p || [];
  1182. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1183. //執行存儲過程
  1184. mysql.usselect(p, function (ret) {
  1185. res.end(JSON.stringify(ret));
  1186. }); //
  1187. };
  1188. //get獲取數據庫信息
  1189. getmysql2 = function (req, res, functionname, ...p) {
  1190. p = p || [];
  1191. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1192. //執行存儲過程
  1193. mysql.usselect(p, function (ret) {
  1194. res.end(JSON.stringify(ret));
  1195. }); //
  1196. };
  1197. //get獲取數據庫信息
  1198. getmysql3 = function (req, res, functionname, ...p) {
  1199. p = p || [];
  1200. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1201. //執行存儲過程
  1202. mysql.usselect(p, function (ret) {
  1203. res.end(JSON.stringify(ret));
  1204. }); //
  1205. };
  1206. //post存储过程处理
  1207. userpostmysql = function (req, res, functionname) {
  1208. //mode的处理
  1209. if (req.body.mode) {
  1210. p = req.body.mode.split(",");
  1211. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1212. //执行存储过程
  1213. mysql.usselect(p, function (ret) {
  1214. res.end(JSON.stringify(ret));
  1215. });
  1216. }
  1217. };
  1218. //get获取数据库信息
  1219. usergetmysql = function (req, res, functionname, ...p) {
  1220. p = p || [];
  1221. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1222. //执行存储过程
  1223. mysql.usselect(p, function (ret) {
  1224. res.end(JSON.stringify(ret));
  1225. }); //
  1226. };
  1227. //post存儲過程處理
  1228. postmysqlLocal = function (req, res) {
  1229. //mode的處理
  1230. if (req.body[0]) {
  1231. p = Object.values(req.body[0]);
  1232. p.unshift(_localhost[0], _localhost[1]);
  1233. //執行存儲過程
  1234. mysql.usselect(p, function (ret) {
  1235. res.end(JSON.stringify(ret));
  1236. });
  1237. }
  1238. };
  1239. postmysqlOnline = function (req, res) {
  1240. //mode的處理
  1241. if (req.body[0]) {
  1242. p = Object.values(req.body[0]);
  1243. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1244. //執行存儲過程
  1245. mysql.usselect(p, function (ret) {
  1246. res.end(JSON.stringify(ret));
  1247. });
  1248. }
  1249. };
  1250. // 新增测试接口,直接返回成功
  1251. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1252. // console.log('收到的 req.body:', req.body);
  1253. // console.log('收到的 req.files:', req.files);
  1254. // console.log('收到的 headers:', req.headers);
  1255. const requestUrl = req.body.url; // 前端传入 url
  1256. if (!requestUrl) {
  1257. return res.status(400).json({error: 'Missing url'});
  1258. }
  1259. const browser = await initBrowser();
  1260. const page = await browser.newPage();
  1261. try {
  1262. console.log("开始抓取页面:", requestUrl);
  1263. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1264. await page.goto(requestUrl);
  1265. const html = await page.content();
  1266. // 新增:HTML 转 Markdown
  1267. const turndownService = new TurndownService();
  1268. // 用 cheerio 只提取正文部分
  1269. const cheerioDoc = cheerio.load(html);
  1270. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1271. // 用 cheerio 过滤无关标签
  1272. const cheerioMain = cheerio.load(mainHtml || '');
  1273. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1274. mainHtml = cheerioMain.html();
  1275. const markdownContent = turndownService.turndown(mainHtml);
  1276. // 新增:解析 <title> 标签内容
  1277. let title = '';
  1278. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1279. if (titleMatch && titleMatch[1]) {
  1280. title = titleMatch[1].trim();
  1281. }
  1282. let finalTitle;
  1283. if (title && title.trim()) {
  1284. finalTitle = title + '.md';
  1285. } else {
  1286. finalTitle = 'untitled-' + uuidv4() + '.md';
  1287. }
  1288. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1289. // 解析用户传入的参数
  1290. const {
  1291. collection_ids,
  1292. id: customId,
  1293. metadata: customMetadata,
  1294. ingestion_mode,
  1295. run_with_orchestration
  1296. } = req.body;
  1297. // 处理 metadata 字段
  1298. let finalMetadata = customMetadata;
  1299. if (typeof finalMetadata === 'string') {
  1300. try {
  1301. finalMetadata = JSON.parse(finalMetadata);
  1302. } catch (e) {
  1303. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1304. }
  1305. }
  1306. // if (!finalMetadata) {
  1307. // finalMetadata = {
  1308. // title: title || uuidv4(),
  1309. // // source_url: url,
  1310. // content_type: "markdown",
  1311. // // scraped_at: new Date().toISOString(),
  1312. // // upload_id: Date.now() + '-' + Math.random()
  1313. // };
  1314. // }
  1315. // 处理 collection_ids 字段
  1316. let finalCollectionIds = collection_ids;
  1317. if (typeof finalCollectionIds === 'string') {
  1318. try {
  1319. finalCollectionIds = JSON.parse(finalCollectionIds);
  1320. } catch (e) {
  1321. // 不是合法 JSON 字符串就当作普通字符串
  1322. }
  1323. }
  1324. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1325. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1326. finalMetadata.collection_ids = finalCollectionIds;
  1327. }
  1328. // 补全 finalMetadata.title 和 content_type
  1329. if (!finalMetadata.title) {
  1330. finalMetadata.title = finalTitle;
  1331. }
  1332. finalMetadata.content_type = 'markdown';
  1333. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1334. const FormData = require('form-data');
  1335. const fs = require('fs');
  1336. const path = require('path');
  1337. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1338. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1339. const form = new FormData();
  1340. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1341. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1342. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1343. }
  1344. if (customId) {
  1345. form.append('id', customId);
  1346. }
  1347. form.append('metadata', JSON.stringify(finalMetadata));
  1348. form.append('ingestion_mode', ingestion_mode || 'fast');
  1349. form.append('run_with_orchestration', 'false');
  1350. // 打印上传参数,方便排查问题
  1351. console.log('Request Headers:', form.getHeaders());
  1352. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1353. const response = await axios.post(url, form, {
  1354. headers: {
  1355. ...form.getHeaders(),
  1356. // 'User-Agent': 'NodeClient/1.0'
  1357. }
  1358. });
  1359. fs.unlinkSync(tmpFilePath);
  1360. result = {
  1361. results: response.data.results,
  1362. title: finalTitle
  1363. }
  1364. res.json(result);
  1365. } catch (error) {
  1366. console.error('Message:', error.message);
  1367. console.error('Stack:', error.stack);
  1368. res.status(500).json({error: error.message});
  1369. } finally {
  1370. await page.close(); // 确保页面总是被关闭
  1371. }
  1372. });
  1373. module.exports = router;