pbl.js 61 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674
  1. // ROUTES FOR OUR APIselectClassStudent
  2. // =============================================================================
  3. const express = require("express");
  4. const request = require("request");
  5. const bcrypt = require("bcryptjs");
  6. const router = express.Router(); // get an instance of the express Router
  7. const querystring = require("querystring");
  8. var mysql = require("./mysql");
  9. // const _mysqlLabor = ["183.36.25.93", "pbl"]; //edu數據庫信息
  10. // const _mysqluser = ["183.36.25.93", "cocorobouser"]; //用户数据库信息
  11. // const _getmysqlLabor = ["123.58.32.151", "pbl"]; //edu數據庫信息f
  12. // const _getmysqluser = ["123.58.32.151", "cocorobouser"]; //edu數據庫信息
  13. const _mysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  14. const _mysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  15. // const _getmysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  16. // const _getmysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  17. const _getmysqlLabor2 = ["172.16.12.7", "pbl"]; //edu數據庫信息
  18. const _getmysqlLabor = ["172.16.12.7", "pbl"]; //edu數據庫信息
  19. const _getmysqluser = ["172.16.12.7", "cocorobouser"]; //edu數據庫信息
  20. const _localhost = ["127.0.0.1", "pbl"]; //edu數據庫信息
  21. const crypto = require("crypto");
  22. const https = require("https");
  23. const { Pool } = require('pg');
  24. const axios = require('axios');
  25. const multer = require("multer");
  26. const {chromium} = require('playwright');
  27. const TurndownService = require('turndown');
  28. const cheerio = require('cheerio');
  29. const { v4: uuidv4 } = require('uuid');
  30. const loginHandler = require('./login'); // 确保路径正确
  31. const { r2rClient } = require("r2r-js");
  32. const client = new r2rClient("https://r2rserver.cocorobo.cn/");
  33. // 配置 multer 存储选项
  34. const storage = multer.diskStorage({
  35. destination: function (req, file, cb) {
  36. req.setEncoding('utf-8');
  37. cb(null, 'uploads/'); // 文件上传目录
  38. },
  39. filename: function (req, file, cb) {
  40. req.setEncoding('utf-8');
  41. // 使用 Buffer 处理文件名,确保编码正确
  42. const fileExtension = file.originalname.split('.').pop(); // 获取文件后缀
  43. const fileNameWithoutExt = Buffer.from(file.originalname.slice(0, -(fileExtension.length + 1)), 'latin1').toString('utf8'); // 转换编码
  44. cb(null, fileNameWithoutExt + '-' + Date.now() + '.' + fileExtension); // 使用时间戳和原始文件名
  45. }
  46. });
  47. const upload = multer({
  48. storage: storage, limits: {
  49. fileSize: 100 * 1024 * 1024 // 100MB
  50. }
  51. });
  52. // 全局浏览器实例
  53. let globalBrowser = null;
  54. async function initBrowser() {
  55. if (!globalBrowser) {
  56. globalBrowser = await chromium.launch({headless: true});
  57. console.log('Browser initialized');
  58. }
  59. return globalBrowser;
  60. }
  61. router.route("/login").all(async (req, res, next) => {
  62. loginHandler(req, res)
  63. });
  64. router.route("/logout").all(async (req, res, next) => {
  65. try {
  66. const response = await client.users.logout();
  67. res.status(200).json({ result: response.results });
  68. } catch (error) {
  69. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  70. }
  71. });
  72. //获取文档列表
  73. router.route("/getDocument").all(async (req, res, next) => {
  74. try {
  75. const response = await client.documents.list({
  76. limit: 10,
  77. offset: 0,
  78. });
  79. console.log(response);
  80. res.status(200).json({ result: response.results });
  81. } catch (error) {
  82. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  83. }
  84. });
  85. //列出文档块
  86. router.route("/listChunks").all(async (req, res, next) => {
  87. try {
  88. const response = await client.documents.listChunks({
  89. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  90. });
  91. console.log(response);
  92. res.status(200).json({ result: response.results });
  93. } catch (error) {
  94. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  95. }
  96. });
  97. //提取实体和关系
  98. router.route("/extract").all(async (req, res, next) => {
  99. try {
  100. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  101. console.log(response);
  102. res.status(200).json({ result: response.results });
  103. } catch (error) {
  104. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  105. }
  106. });
  107. router.route("/selectUser").all((req, res, next) => {
  108. var json = queryString(req.url);
  109. getmysql(req, res, "selectUser", json["userid"]);
  110. });
  111. // 文件上传接口
  112. router.post('/upload', upload.single('file'), async (req, res) => {
  113. req.setEncoding('utf-8')
  114. try {
  115. console.log(req.file); // 打印上传的文件信息
  116. console.log(`File path: uploads/${req.file.filename}`);
  117. const path = require('path');
  118. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  119. console.log(filePath);
  120. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  121. const fs = require('fs');
  122. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  123. const response = await client.documents.create({
  124. file: file,
  125. metadata: { title: filename }
  126. });
  127. console.log(response);
  128. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  129. } catch (error) {
  130. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  131. }
  132. });
  133. //获取文档列表
  134. router.route("/getFile").all(async (req, res, next) => {
  135. //mode的处理
  136. if (req.body[0]) {
  137. p = Object.values(req.body[0]);
  138. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  139. mysql.usselect(p, async function (ret) {
  140. let data = ret[0];
  141. // const array = data.map(item => item.documentid);
  142. const array = data.map(item => item.documentid);
  143. try {
  144. const response = await client.documents.list({ ids: array });
  145. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  146. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  147. // console.log(response);
  148. let data2 = response.results.map(doc => {
  149. doc.documentid = doc.id;
  150. delete doc.id;
  151. return doc;
  152. });
  153. // console.log(data2);
  154. // 合并data和data2,根据documentid
  155. const mergedData = data.map(item => {
  156. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  157. return { ...item, ...correspondingDoc };
  158. });
  159. res.status(200).json({ result: mergedData });
  160. } catch (error) {
  161. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  162. }
  163. });
  164. }
  165. });
  166. //获取文档列表
  167. router.route("/getFileR2r").all(async (req, res, next) => {
  168. //mode的处理
  169. if (req.body[0]) {
  170. let fileid = decodeURIComponent(req.body[0].fileids).split(",");
  171. try {
  172. const response = await client.documents.list({ ids: fileid });
  173. res.status(200).json({ result: response.results });
  174. } catch (error) {
  175. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  176. }
  177. }
  178. });
  179. //获取文档列表
  180. router.route("/getFileD").all(async (req, res, next) => {
  181. //mode的处理
  182. if (req.body[0]) {
  183. p = Object.values(req.body[0]);
  184. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  185. mysql.usselect(p, async function (ret) {
  186. let data = ret[0];
  187. // const array = data.map(item => item.documentid);
  188. const array = data.map(item => item.documentid);
  189. try {
  190. const response = await client.documents.list({ ids: array });
  191. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  192. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  193. // console.log(response);
  194. let data2 = response.results.map(doc => {
  195. doc.documentid = doc.id;
  196. delete doc.id;
  197. return doc;
  198. });
  199. // console.log(data2);
  200. // 合并data和data2,根据documentid
  201. const mergedData = data.map(item => {
  202. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  203. return { ...item, ...correspondingDoc };
  204. });
  205. res.status(200).json({ result: mergedData });
  206. } catch (error) {
  207. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  208. }
  209. });
  210. }
  211. });
  212. //获取文档列表
  213. router.route("/getFileArray").all(async (req, res, next) => {
  214. //mode的处理
  215. if (req.body[0]) {
  216. let data = req.body[0].array.split(",");
  217. // const array = data.map(item => item.documentid);
  218. const array = data;
  219. try {
  220. const response = await client.documents.list({ ids: array });
  221. res.status(200).json({ result: response.results });
  222. } catch (error) {
  223. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  224. }
  225. }
  226. });
  227. //获取文档列表
  228. router.route("/getFileD2").all(async (req, res, next) => {
  229. //mode的处理
  230. if (req.body) {
  231. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  232. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  233. mysql.usselect(p, async function (ret) {
  234. let data = ret[0];
  235. // const array = data.map(item => item.documentid);
  236. const array = data.map(item => item.documentid);
  237. try {
  238. const response = await client.documents.list({ ids: array });
  239. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  240. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  241. // console.log(response);
  242. let data2 = response.results.map(doc => {
  243. doc.documentid = doc.id;
  244. delete doc.id;
  245. return doc;
  246. });
  247. // console.log(data2);
  248. // 合并data和data2,根据documentid
  249. const mergedData = data.map(item => {
  250. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  251. return { ...item, ...correspondingDoc };
  252. });
  253. res.status(200).json({ result: mergedData });
  254. } catch (error) {
  255. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  256. }
  257. });
  258. }
  259. });
  260. //获取文档列表
  261. router.route("/getFileD2").all(async (req, res, next) => {
  262. //mode的处理
  263. if (req.body) {
  264. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  265. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  266. mysql.usselect(p, async function (ret) {
  267. let data = ret[0];
  268. // const array = data.map(item => item.documentid);
  269. const array = data.map(item => item.documentid);
  270. try {
  271. const response = await client.documents.list({ ids: array });
  272. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  273. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  274. // console.log(response);
  275. let data2 = response.results.map(doc => {
  276. doc.documentid = doc.id;
  277. delete doc.id;
  278. return doc;
  279. });
  280. // console.log(data2);
  281. // 合并data和data2,根据documentid
  282. const mergedData = data.map(item => {
  283. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  284. return { ...item, ...correspondingDoc };
  285. });
  286. res.status(200).json({ result: mergedData });
  287. } catch (error) {
  288. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  289. }
  290. });
  291. }
  292. });
  293. //获取文档列表
  294. router.route("/getFileOpen").all(async (req, res, next) => {
  295. //mode的处理
  296. if (req.body[0]) {
  297. p = Object.values(req.body[0]);
  298. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  299. mysql.usselect(p, async function (ret) {
  300. let data = ret[0];
  301. let user = ret[1];
  302. // const array = data.map(item => item.documentid);
  303. const array = data.map(item => item.documentid);
  304. try {
  305. const response = await client.documents.list({ ids: array });
  306. let data2 = response.results.map(doc => {
  307. doc.documentid = doc.id;
  308. delete doc.id;
  309. return doc;
  310. });
  311. // 合并data和data2,根据documentid
  312. const mergedData = data.map(item => {
  313. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  314. return { ...item, ...correspondingDoc };
  315. });
  316. res.status(200).json({ result: mergedData, user: user });
  317. } catch (error) {
  318. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  319. }
  320. });
  321. }
  322. });
  323. //获取文档列表2
  324. router.route("/getFileP").all(async (req, res, next) => {
  325. //mode的处理
  326. if (req.body) {
  327. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  328. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  329. mysql.usselect(p, async function (ret) {
  330. let data = ret[0];
  331. // const array = data.map(item => item.documentid);
  332. try {
  333. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  334. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  335. // console.log(response);
  336. let data2 = response.results.map(doc => {
  337. doc.documentid = doc.id;
  338. delete doc.id;
  339. return doc;
  340. });
  341. // 合并data和data2,根据documentid
  342. const mergedData = data.map(item => {
  343. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  344. return { ...item, ...correspondingDoc };
  345. });
  346. res.status(200).json({ result: mergedData });
  347. } catch (error) {
  348. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  349. }
  350. });
  351. }
  352. });
  353. //获取能关联的文件
  354. router.route("/getFile2").all(async (req, res, next) => {
  355. //mode的处理
  356. if (req.body[0]) {
  357. p = Object.values(req.body[0]);
  358. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  359. mysql.usselect(p, async function (ret) {
  360. let data = ret[0];
  361. const array = data.map(item => item.documentid);
  362. try {
  363. const response = await client.documents.list({ ids: array });
  364. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  365. // console.log(response);
  366. let data2 = response.results.map(doc => {
  367. doc.documentid = doc.id;
  368. delete doc.id;
  369. return doc;
  370. });
  371. // 合并data和data2,根据documentid
  372. const mergedData = data.map(item => {
  373. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  374. return { ...item, ...correspondingDoc };
  375. });
  376. res.status(200).json({ result: mergedData });
  377. } catch (error) {
  378. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  379. }
  380. });
  381. }
  382. });
  383. //添加文件夹
  384. router.route("/addFolder").all(async (req, res, next) => {
  385. //mode的处理
  386. if (req.body[0]) {
  387. try {
  388. const response = await client.collections.create({
  389. name: decodeURIComponent(req.body[0].n),
  390. description: decodeURIComponent(req.body[0].fd)
  391. });
  392. p = [...Object.values(req.body[0]), response.results.id];
  393. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  394. mysql.usselect(p, async function (ret) {
  395. res.end(JSON.stringify(ret));
  396. });
  397. } catch (error) {
  398. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  399. }
  400. }
  401. });
  402. //添加文件夹
  403. router.route("/addFolderP").all(async (req, res, next) => {
  404. //mode的处理
  405. if (req.body) {
  406. try {
  407. const response = await client.collections.create({
  408. name: decodeURIComponent(req.body.n),
  409. description: decodeURIComponent(req.body.fd)
  410. });
  411. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  412. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  413. mysql.usselect(p, async function (ret) {
  414. res.end(JSON.stringify(ret));
  415. });
  416. } catch (error) {
  417. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  418. }
  419. }
  420. });
  421. //删除文件夹
  422. router.route("/deleteFolder").all(async (req, res, next) => {
  423. //mode的处理
  424. if (req.body[0]) {
  425. try {
  426. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  427. const ids = req.body[0].ids;
  428. // 并行处理文件夹下的文件删除和文件夹删除
  429. await Promise.all(folderids.map(async folderid => {
  430. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  431. await Promise.all(files.results.map(file =>
  432. client.collections.removeDocument({ id: folderid, documentId: file.id })
  433. ));
  434. }));
  435. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  436. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  437. mysql.usselect(params, async function (ret) {
  438. res.end(JSON.stringify(ret));
  439. });
  440. } catch (error) {
  441. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  442. }
  443. }
  444. });
  445. //修改文件夹
  446. router.route("/updateFolder").all(async (req, res, next) => {
  447. //mode的处理
  448. if (req.body[0]) {
  449. try {
  450. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  451. const response = await client.collections.update({
  452. id: folderid,
  453. name: decodeURIComponent(req.body[0].n),
  454. description: decodeURIComponent(req.body[0].d)
  455. });
  456. p = Object.values(folderData); // 获取剩余属性的键
  457. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  458. mysql.usselect(p, async function (ret) {
  459. res.end(JSON.stringify(ret));
  460. });
  461. } catch (error) {
  462. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  463. }
  464. }
  465. });
  466. //修改文件夹
  467. router.route("/updateFolder2").all(async (req, res, next) => {
  468. //mode的处理
  469. if (req.body[0]) {
  470. try {
  471. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  472. const response = await client.collections.update({
  473. id: folderid,
  474. name: decodeURIComponent(req.body[0].n),
  475. description: decodeURIComponent(req.body[0].d)
  476. });
  477. p = Object.values(folderData); // 获取剩余属性的键
  478. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  479. mysql.usselect(p, async function (ret) {
  480. res.end(JSON.stringify(ret));
  481. });
  482. } catch (error) {
  483. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  484. }
  485. }
  486. });
  487. //移除文件
  488. router.route("/removeFile").all(async (req, res, next) => {
  489. if (req.body[0]) {
  490. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  491. let folderid = req.body[0].folderid;
  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. const removePromises = documentids.map(documentid =>
  495. client.collections.removeDocument({ id: folderid, documentId: documentid })
  496. );
  497. // 不管client.documents报错没有也要执行removeFile
  498. Promise.all(removePromises).catch(error => {
  499. console.error("移除文件时发生错误:", error);
  500. });
  501. p = [ids, folderid];
  502. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  503. mysql.usselect(p, async function (ret) {
  504. res.end(JSON.stringify(ret));
  505. });
  506. }
  507. });
  508. //移入文件
  509. router.route("/moveFile").all(async (req, res, next) => {
  510. if (req.body[0]) {
  511. try {
  512. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  513. let folderid = req.body[0].folderid
  514. let ids = decodeURIComponent(req.body[0].ids).split(",")
  515. // 使用 Promise.all 确保所有添加操作都被执行
  516. await Promise.all(documentids.map(documentid =>
  517. client.collections.addDocument({
  518. id: folderid,
  519. documentId: documentid
  520. }).catch(error => {
  521. console.error("移动文件时发生错误:", error);
  522. })
  523. ));
  524. for (const id of ids) {
  525. p = [id, folderid];
  526. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  527. await mysql.usselect(p, async function (ret) { });
  528. }
  529. res.status(200).json({ messages: "移动成功" });
  530. } catch (error) {
  531. // 这里不返回错误,确保deleteFile执行完
  532. console.error("处理过程中发生错误:", error);
  533. }
  534. }
  535. });
  536. //删除文件
  537. router.route("/deleteFile").all(async (req, res, next) => {
  538. if (req.body[0]) {
  539. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  540. let ids = req.body[0].ids;
  541. // 使用 Promise.all 确保所有删除操作都被执行
  542. Promise.all(documentids.map(documentid =>
  543. client.documents.delete({ id: documentid }).catch(error => {
  544. console.error("删除文件时发生错误:", error);
  545. })
  546. ));
  547. p = [ids];
  548. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  549. mysql.usselect(p, async function (ret) {
  550. res.end(JSON.stringify(ret));
  551. });
  552. }
  553. });
  554. //删除文件
  555. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  556. if (req.body[0]) {
  557. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  558. let ids = req.body[0].ids;
  559. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  560. console.log("文件删除成功");
  561. }).catch(error => {
  562. console.error("删除文件时发生错误:", error);
  563. })
  564. p = [ids];
  565. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  566. mysql.usselect(p, async function (ret) {
  567. res.end(JSON.stringify(ret));
  568. });
  569. }
  570. });
  571. router.route("/getFolder").all((req, res, next) => {
  572. getmysql2(req, res, "getFolder", req.body.uid);
  573. });
  574. router.route("/getFolder2").all((req, res, next) => {
  575. var json = queryString(req.url);
  576. getmysql2(req, res, "getFolder", json['uid']);
  577. });
  578. router.route("/getFolderPage").all((req, res, next) => {
  579. 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);
  580. });
  581. router.route("/getFolderPage3").all((req, res, next) => {
  582. 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);
  583. });
  584. router.route("/getFolderPage32").all((req, res, next) => {
  585. var json = queryString(req.url);
  586. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  587. });
  588. router.route("/getFolderPageOpen").all((req, res, next) => {
  589. 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);
  590. });
  591. router.route("/haveFile").all((req, res, next) => {
  592. postmysql(req, res, "haveFile");
  593. });
  594. router.route("/haveFolder").all((req, res, next) => {
  595. postmysql(req, res, "haveFolder");
  596. });
  597. //search文件
  598. router.route("/searchFile").all(async (req, res, next) => {
  599. if (req.body) {
  600. try {
  601. let filters = {}
  602. if (req.body.type == 1) {
  603. filters = { "document_id": { "$in": req.body.ids } }
  604. } else if (req.body.type == 2) {
  605. filters = { "collection_ids": { "$in": req.body.ids } }
  606. }
  607. const response = await client.retrieval.search({
  608. query: req.body.query,
  609. search_settings: {
  610. filters: filters,
  611. useSemanticSearch: true,
  612. chunkSettings: {
  613. limit: req.body.limit ? req.body.limit : 10,
  614. enabled: true
  615. },
  616. }
  617. });
  618. res.status(200).json({ result: response.results });
  619. } catch (error) {
  620. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  621. }
  622. }
  623. });
  624. //提取文件实体
  625. router.route("/extractFile").all(async (req, res, next) => {
  626. if (req.body) {
  627. try {
  628. const response = client.documents.extract({
  629. id: req.body.id
  630. })
  631. res.status(200).json({ result: response.results });
  632. } catch (error) {
  633. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  634. }
  635. }
  636. });
  637. //添加上传好的知识库文件
  638. router.route("/addFile").all((req, res, next) => {
  639. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  640. })
  641. //获取文件详情
  642. router.route("/getFileDetail").all(async (req, res, next) => {
  643. if (req.body[0]) {
  644. p = Object.values(req.body[0]);
  645. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  646. mysql.usselect(p, async function (ret) {
  647. let data = ret[0][0];
  648. // const array = data.map(item => item.documentid);
  649. try {
  650. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  651. // console.log(response);
  652. let data2 = response.results[0]
  653. // 合并data和data2,根据documentid
  654. const mergedData = { ...data, ...data2 };
  655. res.status(200).json({ result: mergedData });
  656. } catch (error) {
  657. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  658. }
  659. });
  660. }
  661. });
  662. //获取切片
  663. router.route("/getChunks").all(async (req, res, next) => {
  664. if (req.body[0]) {
  665. try {
  666. const { page = 1, limit = 10, documentId } = req.body[0];
  667. const offset = (page - 1) * limit;
  668. const response = await client.documents.listChunks({
  669. id: documentId,
  670. includeVectors: false,
  671. offset,
  672. limit,
  673. });
  674. res.status(200).json({ result: response });
  675. } catch (error) {
  676. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  677. }
  678. }
  679. });
  680. //获取实体
  681. router.route("/getEntities").all(async (req, res, next) => {
  682. if (req.body[0]) {
  683. try {
  684. const { page = 1, limit = 10, documentId } = req.body[0];
  685. const offset = (page - 1) * limit;
  686. const response = await client.documents.listEntities({
  687. id: documentId,
  688. includeVectors: false,
  689. offset,
  690. limit,
  691. });
  692. res.status(200).json({ result: response });
  693. } catch (error) {
  694. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  695. }
  696. }
  697. });
  698. //获取关系
  699. router.route("/getRelationships").all(async (req, res, next) => {
  700. if (req.body[0]) {
  701. try {
  702. const { page = 1, limit = 10, documentId } = req.body[0];
  703. const offset = (page - 1) * limit;
  704. const response = await client.documents.listRelationships({
  705. id: documentId,
  706. includeVectors: false,
  707. offset,
  708. limit,
  709. });
  710. res.status(200).json({ result: response });
  711. } catch (error) {
  712. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  713. }
  714. }
  715. });
  716. //提取整个文件夹的关系
  717. router.route("/resetGraphs").all(async (req, res, next) => {
  718. if (req.body[0]) {
  719. try {
  720. // const response = await client.graphs.pull({
  721. // collectionId: req.body[0].collectionId,
  722. // });
  723. const response = await client.graphs.reset({
  724. collectionId: req.body[0].collectionId,
  725. });
  726. res.status(200).json({ result: response });
  727. } catch (error) {
  728. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  729. }
  730. }
  731. });
  732. //获取整个文件夹的关系
  733. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  734. if (req.body[0]) {
  735. try {
  736. await client.graphs.pull({
  737. collectionId: req.body[0].collectionId,
  738. });
  739. const response = await client.graphs.listRelationships({
  740. collectionId: req.body[0].collectionId,
  741. offset: req.body[0].page ? req.body[0].page : 0,
  742. limit: 1000
  743. });
  744. res.status(200).json({ result: response });
  745. } catch (error) {
  746. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  747. }
  748. }
  749. });
  750. //提取知识图谱
  751. router.route("/getGraphs").all(async (req, res, next) => {
  752. console.log(req.body);
  753. //mode的处理
  754. if (req.body[0]) {
  755. try {
  756. //创建临时文件夹
  757. const folder = await client.collections.create({
  758. name: "临时文件夹",
  759. description: "临时文件夹"
  760. });
  761. let folderid = folder.results.id
  762. console.log(folder.results.id);
  763. //将要提取的文件添加到临时文件夹
  764. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  765. // 使用 Promise.all 确保所有添加操作都被执行
  766. await Promise.all(documentids.map(documentid =>
  767. client.collections.addDocument({
  768. id: folderid,
  769. documentId: documentid
  770. }).catch(error => {
  771. console.error("移动文件时发生错误:", error);
  772. })
  773. ));
  774. //将临时文件夹提取知识图谱
  775. await client.graphs.pull({
  776. collectionId: folderid
  777. });
  778. let offset = 0;
  779. let results = [];
  780. let totalEntries = 0;
  781. do {
  782. const response = await client.graphs.listRelationships({
  783. collectionId: folderid,
  784. offset: offset,
  785. limit: 1000
  786. });
  787. results = results.concat(response.results);
  788. totalEntries = response.totalEntries;
  789. offset += 1000;
  790. } while (offset < totalEntries);
  791. //移除临时文件夹的文件
  792. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  793. await Promise.all(files.results.map(file =>
  794. client.collections.removeDocument({ id: folderid, documentId: file.id })
  795. ));
  796. //删除临时文件夹
  797. await client.collections.delete({ id: folderid });
  798. res.status(200).json({ result: results });
  799. } catch (error) {
  800. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  801. }
  802. }
  803. });
  804. //提取知识图谱
  805. router.route("/getGraphs2").all(async (req, res, next) => {
  806. console.log(req.body);
  807. //mode的处理
  808. if (req.body[0]) {
  809. try {
  810. let documentIds = decodeURIComponent(req.body[0].fileids).split(",");
  811. // 使用 Promise.all 并通过 map 返回新的 results 数组
  812. let results = await Promise.all(documentIds.map(async documentId => {
  813. const response = await client.documents.listRelationships({
  814. id: documentId,
  815. includeVectors: false,
  816. limit: 1000, // 修正这里的参数
  817. });
  818. return response.results;
  819. }));
  820. // 使用 flat() 合并所有数组
  821. results = results.flat();
  822. res.status(200).json({ result: results });
  823. } catch (error) {
  824. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  825. }
  826. }
  827. });
  828. //获取实体和关系
  829. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  830. if (req.body[0]) {
  831. try {
  832. const { documentId } = req.body[0];
  833. const offset = (page - 1) * limit;
  834. const response = await client.documents.listEntities({
  835. id: documentId,
  836. includeVectors: false,
  837. });
  838. const response2 = await client.documents.listRelationships({
  839. id: documentId,
  840. includeVectors: false,
  841. });
  842. res.status(200).json({ Entities: response, Relationships: response2 });
  843. } catch (error) {
  844. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  845. }
  846. }
  847. });
  848. //提取功能
  849. router.route("/extractFile2").all(async (req, res, next) => {
  850. try {
  851. let setting = {}
  852. if(req.body[0].prompt){
  853. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  854. }
  855. console.log(setting);
  856. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  857. console.log(response);
  858. res.status(200).json({ result: response.results });
  859. } catch (error) {
  860. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  861. }
  862. });
  863. //复制文件夹
  864. router.route("/copyFolder").all(async (req, res, next) => {
  865. if (req.body[0]) {
  866. const response = await client.collections.create({
  867. name: decodeURIComponent(req.body[0].n),
  868. description: decodeURIComponent(req.body[0].fd)
  869. });
  870. let folderid = response.results.id;
  871. p2 = [req.body[0].folderid]
  872. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  873. mysql.usselect(p2, async function (ret2) {
  874. let file = ret2[0]
  875. if (file.length) {
  876. let documentids = file.map(item => item.documentid);
  877. // 使用 Promise.all 确保所有添加操作都被执行
  878. await Promise.all(documentids.map(documentid =>
  879. client.collections.addDocument({
  880. id: folderid,
  881. documentId: documentid
  882. }).catch(error => {
  883. console.error("移动文件时发生错误:", error);
  884. })
  885. ));
  886. }
  887. let ids = file.map(item => item.fileid);
  888. console.log(ids);
  889. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  890. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  891. mysql.usselect(p, async function (ret) {
  892. res.end(JSON.stringify(ret));
  893. });
  894. });
  895. }
  896. });
  897. // 下载文件功能
  898. router.route("/downloadFile").all(async (req, res, next) => {
  899. var json = queryString(req.url);
  900. let params = {
  901. documentid: (json && json['documentid'] || req.body[0].documentid),
  902. filename: (json && json['filename'] || req.body[0].filename),
  903. };
  904. try {
  905. const response = await client.documents.download({ id: params.documentid });
  906. console.log(response);
  907. // 根据response的类型设置Content-Type
  908. const contentType = response.type || 'application/octet-stream';
  909. res.setHeader('Content-Type', contentType);
  910. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  911. // 直接发送blob数据
  912. res.send(Buffer.from(await response.arrayBuffer()));
  913. } catch (error) {
  914. console.error("下载文件时发生错误:", error);
  915. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  916. }
  917. });
  918. // 下载文件功能
  919. router.route("/downloadFile2").all(async (req, res, next) => {
  920. var json = queryString(req.url);
  921. let params = {
  922. documentid: (json && json['documentid'] || req.body[0].documentid),
  923. filename: (json && json['filename'] || req.body[0].filename),
  924. };
  925. try {
  926. const response = await client.documents.download({ id: params.documentid });
  927. console.log(response);
  928. // 直接返回blob数据
  929. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  930. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  931. const buffer = Buffer.from(await response.arrayBuffer());
  932. res.send(buffer); // 发送blob数据
  933. } catch (error) {
  934. console.error("下载文件时发生错误:", error);
  935. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  936. }
  937. });
  938. //召回测试
  939. router.route("/ragFolder").all(async (req, res, next) => {
  940. let reqJson = req.body[0]
  941. if (reqJson) {
  942. reqJson.query = decodeURIComponent(reqJson.query)
  943. try {
  944. let filters = {}
  945. if (reqJson.type == 1) {
  946. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  947. } else if (reqJson.type == 2) {
  948. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  949. } else if (reqJson.type == 3) {
  950. filters = { "collection_ids": { "$eq": reqJson.ids } }
  951. }
  952. console.log(111);
  953. const response = await client.retrieval.search({
  954. query: reqJson.query,
  955. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  956. // searchMode: 'basic',
  957. searchSettings: {
  958. useHybridSearch: reqJson.useType == 3,
  959. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  960. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  961. hybrid_settings: {
  962. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  963. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  964. },
  965. filters: filters,
  966. limit: reqJson.limit ? reqJson.limit : 100,
  967. },
  968. ragGenerationConfig: {
  969. stream: false,
  970. temperature: reqJson.temperature ? reqJson.temperature : 0,
  971. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  972. }
  973. });
  974. // res.status(200).json({ result: response.results });
  975. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  976. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  977. mysql.usselect(p, async function (ret) {
  978. res.status(200).json({ result: response.results });
  979. });
  980. } catch (error) {
  981. console.log(error);
  982. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  983. }
  984. }
  985. });
  986. //召回测试
  987. router.route("/searchFolder").all(async (req, res, next) => {
  988. let reqJson = req.body[0]
  989. if (reqJson) {
  990. try {
  991. let filters = {}
  992. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  993. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  994. filters = {
  995. "$or": [
  996. { "collection_ids": { "$in": folderids } },
  997. { "document_id": { "$in": documentids } }
  998. ]
  999. };
  1000. let params = {
  1001. query: reqJson.query,
  1002. searchSettings: {
  1003. useHybridSearch: reqJson.useType == 3,
  1004. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1005. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1006. hybrid_settings: {
  1007. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1008. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1009. },
  1010. filters: filters,
  1011. limit: reqJson.limit ? reqJson.limit : 100,
  1012. },
  1013. ragGenerationConfig: {
  1014. stream: false,
  1015. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1016. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1017. }
  1018. }
  1019. // console.log(params);
  1020. const response = await client.retrieval.search(params);
  1021. res.status(200).json({ result: response.results });
  1022. } catch (error) {
  1023. console.log(error);
  1024. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1025. }
  1026. }
  1027. });
  1028. //召回测试
  1029. router.route("/searchFolder2").all(async (req, res, next) => {
  1030. let reqJson = req.body
  1031. if (Object.keys(reqJson).length) {
  1032. try {
  1033. let filters = {}
  1034. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  1035. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  1036. filters = {
  1037. "$or": [
  1038. { "collection_ids": { "$in": folderids } },
  1039. { "document_id": { "$in": documentids } }
  1040. ]
  1041. };
  1042. if(!folderids.length && !documentids.length){
  1043. res.status(200).json({ result: [] });
  1044. return
  1045. }
  1046. let params = {
  1047. query: reqJson.query,
  1048. searchSettings: {
  1049. useHybridSearch: reqJson.useType == 3,
  1050. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1051. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1052. hybrid_settings: {
  1053. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1054. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1055. },
  1056. filters: filters,
  1057. limit: reqJson.limit ? reqJson.limit : 100,
  1058. },
  1059. ragGenerationConfig: {
  1060. stream: false,
  1061. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1062. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1063. }
  1064. }
  1065. // console.log(params);
  1066. const response = await client.retrieval.search(params);
  1067. res.status(200).json({ result: response.results });
  1068. } catch (error) {
  1069. console.log(error);
  1070. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1071. }
  1072. }else{
  1073. res.status(200).json({ result: "请传参" });
  1074. }
  1075. });
  1076. //召回测试
  1077. router.route("/searchFolder3").all(async (req, res, next) => {
  1078. let reqJson = req.body[0]
  1079. if (reqJson) {
  1080. reqJson.query = decodeURIComponent(reqJson.query)
  1081. try {
  1082. let filters = {}
  1083. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1084. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1085. filters = {
  1086. "$or": [
  1087. { "collection_ids": { "$in": folderids } },
  1088. { "document_id": { "$in": documentids } }
  1089. ]
  1090. };
  1091. if(!folderids.length && !documentids.length){
  1092. res.status(200).json({ result: [] });
  1093. return
  1094. }
  1095. let params = {
  1096. query: reqJson.query,
  1097. searchSettings: {
  1098. useHybridSearch: reqJson.useType == 3,
  1099. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1100. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1101. hybrid_settings: {
  1102. fullTextWeight: reqJson.fullTextWeight,
  1103. semanticWeight: reqJson.semanticWeight,
  1104. },
  1105. filters: filters,
  1106. limit: reqJson.limit ? reqJson.limit : 100,
  1107. },
  1108. ragGenerationConfig: {
  1109. stream: false,
  1110. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1111. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1112. }
  1113. }
  1114. // console.log(params);
  1115. const response = await client.retrieval.search(params);
  1116. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1117. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1118. mysql.usselect(p, async function (ret) {
  1119. res.status(200).json({ result: response.results });
  1120. });
  1121. } catch (error) {
  1122. console.log(error);
  1123. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1124. }
  1125. }
  1126. });
  1127. //召回测试
  1128. router.route("/searchFolder32").all(async (req, res, next) => {
  1129. let reqJson = req.body[0]
  1130. if (reqJson) {
  1131. reqJson.query = decodeURIComponent(reqJson.query)
  1132. try {
  1133. let filters = {}
  1134. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1135. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1136. filters = {
  1137. "$or": [
  1138. { "collection_ids": { "$in": folderids } },
  1139. { "document_id": { "$in": documentids } }
  1140. ]
  1141. };
  1142. if(!folderids.length && !documentids.length){
  1143. res.status(200).json({ result: [] });
  1144. return
  1145. }
  1146. let params = {
  1147. query: reqJson.query,
  1148. searchSettings: {
  1149. useHybridSearch: reqJson.useType == 3,
  1150. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1151. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1152. hybrid_settings: {
  1153. fullTextWeight: reqJson.fullTextWeight,
  1154. semanticWeight: reqJson.semanticWeight,
  1155. },
  1156. filters: filters,
  1157. limit: reqJson.limit ? reqJson.limit : 100,
  1158. },
  1159. ragGenerationConfig: {
  1160. stream: false,
  1161. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1162. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1163. }
  1164. }
  1165. console.log(params);
  1166. const response = await client.retrieval.search(params);
  1167. res.status(200).json({ result: response.results });
  1168. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1169. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1170. // mysql.usselect(p, async function (ret) {
  1171. // });
  1172. } catch (error) {
  1173. console.log(error);
  1174. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1175. }
  1176. }
  1177. });
  1178. //获取召回测试数据
  1179. router.route("/getRagData").all(async (req, res, next) => {
  1180. //mode的处理
  1181. if (req.body[0]) {
  1182. p = Object.values(req.body[0]);
  1183. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1184. mysql.usselect(p, async function (ret) {
  1185. res.status(200).json(ret);
  1186. });
  1187. }
  1188. });
  1189. //删除召回测试数据
  1190. router.route("/delRagData").all(async (req, res, next) => {
  1191. //mode的处理
  1192. if (req.body[0]) {
  1193. p = Object.values(req.body[0]);
  1194. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1195. mysql.usselect(p, async function (ret) {
  1196. res.status(200).json(ret);
  1197. });
  1198. }
  1199. });
  1200. function generateAPIKey(username) {
  1201. let timeStamp = new Date().getTime();
  1202. let hashStr = timeStamp + username;
  1203. let digest = crypto
  1204. .createHash("sha256")
  1205. .update(hashStr, "utf8")
  1206. .digest("hex");
  1207. return digest;
  1208. }
  1209. queryString = function (url) {
  1210. var a = url.split("?");
  1211. var json = querystring.parse(a[1]);
  1212. return json;
  1213. };
  1214. //post存儲過程處理
  1215. postmysql = function (req, res, functionname) {
  1216. //mode的處理
  1217. if (req.body[0]) {
  1218. p = Object.values(req.body[0]);
  1219. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1220. //執行存儲過程
  1221. mysql.usselect(p, function (ret) {
  1222. res.end(JSON.stringify(ret));
  1223. });
  1224. }
  1225. };
  1226. //post存儲過程處理
  1227. postmysql2 = function (req, res, functionname) {
  1228. //mode的處理
  1229. if (req.body) {
  1230. p = Object.values(req.body)[0].split(",");
  1231. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1232. //執行存儲過程
  1233. mysql.usselect(p, function (ret) {
  1234. res.end(JSON.stringify(ret));
  1235. });
  1236. }
  1237. };
  1238. //post存儲過程處理
  1239. postmysql3 = function (req, res, functionname) {
  1240. //mode的處理
  1241. if (req.body) {
  1242. console.log(req);
  1243. p = Object.values(req.body);
  1244. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1245. //執行存儲過程
  1246. mysql.usselect(p, function (ret) {
  1247. res.end(JSON.stringify(ret));
  1248. });
  1249. }
  1250. };
  1251. //get獲取數據庫信息
  1252. getmysql = function (req, res, functionname, ...p) {
  1253. p = p || [];
  1254. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1255. //執行存儲過程
  1256. mysql.usselect(p, function (ret) {
  1257. res.end(JSON.stringify(ret));
  1258. }); //
  1259. };
  1260. //get獲取數據庫信息
  1261. getmysql2 = function (req, res, functionname, ...p) {
  1262. p = p || [];
  1263. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1264. //執行存儲過程
  1265. mysql.usselect(p, function (ret) {
  1266. res.end(JSON.stringify(ret));
  1267. }); //
  1268. };
  1269. //get獲取數據庫信息
  1270. getmysql3 = function (req, res, functionname, ...p) {
  1271. p = p || [];
  1272. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1273. //執行存儲過程
  1274. mysql.usselect(p, function (ret) {
  1275. res.end(JSON.stringify(ret));
  1276. }); //
  1277. };
  1278. //post存储过程处理
  1279. userpostmysql = function (req, res, functionname) {
  1280. //mode的处理
  1281. if (req.body.mode) {
  1282. p = req.body.mode.split(",");
  1283. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1284. //执行存储过程
  1285. mysql.usselect(p, function (ret) {
  1286. res.end(JSON.stringify(ret));
  1287. });
  1288. }
  1289. };
  1290. //get获取数据库信息
  1291. usergetmysql = function (req, res, functionname, ...p) {
  1292. p = p || [];
  1293. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1294. //执行存储过程
  1295. mysql.usselect(p, function (ret) {
  1296. res.end(JSON.stringify(ret));
  1297. }); //
  1298. };
  1299. //post存儲過程處理
  1300. postmysqlLocal = function (req, res) {
  1301. //mode的處理
  1302. if (req.body[0]) {
  1303. p = Object.values(req.body[0]);
  1304. p.unshift(_localhost[0], _localhost[1]);
  1305. //執行存儲過程
  1306. mysql.usselect(p, function (ret) {
  1307. res.end(JSON.stringify(ret));
  1308. });
  1309. }
  1310. };
  1311. postmysqlOnline = function (req, res) {
  1312. //mode的處理
  1313. if (req.body[0]) {
  1314. p = Object.values(req.body[0]);
  1315. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1316. //執行存儲過程
  1317. mysql.usselect(p, function (ret) {
  1318. res.end(JSON.stringify(ret));
  1319. });
  1320. }
  1321. };
  1322. // 新增测试接口,直接返回成功
  1323. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1324. // console.log('收到的 req.body:', req.body);
  1325. // console.log('收到的 req.files:', req.files);
  1326. // console.log('收到的 headers:', req.headers);
  1327. const requestUrl = req.body.url; // 前端传入 url
  1328. if (!requestUrl) {
  1329. return res.status(400).json({error: 'Missing url'});
  1330. }
  1331. const browser = await initBrowser();
  1332. const page = await browser.newPage();
  1333. try {
  1334. console.log("开始抓取页面:", requestUrl);
  1335. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1336. await page.goto(requestUrl);
  1337. const html = await page.content();
  1338. // 新增:HTML 转 Markdown
  1339. const turndownService = new TurndownService();
  1340. // 用 cheerio 只提取正文部分
  1341. const cheerioDoc = cheerio.load(html);
  1342. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1343. // 用 cheerio 过滤无关标签
  1344. const cheerioMain = cheerio.load(mainHtml || '');
  1345. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1346. mainHtml = cheerioMain.html();
  1347. const markdownContent = turndownService.turndown(mainHtml);
  1348. // 新增:解析 <title> 标签内容
  1349. let title = '';
  1350. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1351. if (titleMatch && titleMatch[1]) {
  1352. title = titleMatch[1].trim();
  1353. }
  1354. let finalTitle;
  1355. if (title && title.trim()) {
  1356. finalTitle = title + '.md';
  1357. } else {
  1358. finalTitle = 'untitled-' + uuidv4() + '.md';
  1359. }
  1360. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1361. // 解析用户传入的参数
  1362. const {
  1363. collection_ids,
  1364. id: customId,
  1365. metadata: customMetadata,
  1366. ingestion_mode,
  1367. run_with_orchestration
  1368. } = req.body;
  1369. // 处理 metadata 字段
  1370. let finalMetadata = customMetadata;
  1371. if (typeof finalMetadata === 'string') {
  1372. try {
  1373. finalMetadata = JSON.parse(finalMetadata);
  1374. } catch (e) {
  1375. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1376. }
  1377. }
  1378. // if (!finalMetadata) {
  1379. // finalMetadata = {
  1380. // title: title || uuidv4(),
  1381. // // source_url: url,
  1382. // content_type: "markdown",
  1383. // // scraped_at: new Date().toISOString(),
  1384. // // upload_id: Date.now() + '-' + Math.random()
  1385. // };
  1386. // }
  1387. // 处理 collection_ids 字段
  1388. let finalCollectionIds = collection_ids;
  1389. if (typeof finalCollectionIds === 'string') {
  1390. try {
  1391. finalCollectionIds = JSON.parse(finalCollectionIds);
  1392. } catch (e) {
  1393. // 不是合法 JSON 字符串就当作普通字符串
  1394. }
  1395. }
  1396. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1397. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1398. finalMetadata.collection_ids = finalCollectionIds;
  1399. }
  1400. // 补全 finalMetadata.title 和 content_type
  1401. if (!finalMetadata.title) {
  1402. finalMetadata.title = finalTitle;
  1403. }
  1404. finalMetadata.content_type = 'markdown';
  1405. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1406. const FormData = require('form-data');
  1407. const fs = require('fs');
  1408. const path = require('path');
  1409. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1410. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1411. const form = new FormData();
  1412. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1413. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1414. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1415. }
  1416. if (customId) {
  1417. form.append('id', customId);
  1418. }
  1419. form.append('metadata', JSON.stringify(finalMetadata));
  1420. form.append('ingestion_mode', ingestion_mode || 'fast');
  1421. form.append('run_with_orchestration', 'false');
  1422. // 打印上传参数,方便排查问题
  1423. console.log('Request Headers:', form.getHeaders());
  1424. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1425. const response = await axios.post(url, form, {
  1426. headers: {
  1427. ...form.getHeaders(),
  1428. // 'User-Agent': 'NodeClient/1.0'
  1429. }
  1430. });
  1431. fs.unlinkSync(tmpFilePath);
  1432. result = {
  1433. results: response.data.results,
  1434. title: finalTitle
  1435. }
  1436. res.json(result);
  1437. } catch (error) {
  1438. console.error('Message:', error.message);
  1439. console.error('Stack:', error.stack);
  1440. res.status(500).json({error: error.message});
  1441. } finally {
  1442. await page.close(); // 确保页面总是被关闭
  1443. }
  1444. });
  1445. module.exports = router;