pbl.js 61 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680
  1. // ROUTES FOR OUR APIselectClassStudent
  2. // =============================================================================
  3. const express = require("express");
  4. const request = require("request");
  5. const bcrypt = require("bcryptjs");
  6. const router = express.Router(); // get an instance of the express Router
  7. const querystring = require("querystring");
  8. var mysql = require("./mysql");
  9. // const _mysqlLabor = ["183.36.25.93", "pbl"]; //edu數據庫信息
  10. // const _mysqluser = ["183.36.25.93", "cocorobouser"]; //用户数据库信息
  11. // const _getmysqlLabor = ["123.58.32.151", "pbl"]; //edu數據庫信息f
  12. // const _getmysqluser = ["123.58.32.151", "cocorobouser"]; //edu數據庫信息
  13. const _mysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  14. const _mysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  15. // const _getmysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  16. // const _getmysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  17. const _getmysqlLabor2 = ["172.16.12.7", "pbl"]; //edu數據庫信息
  18. const _getmysqlLabor = ["172.16.12.7", "pbl"]; //edu數據庫信息
  19. const _getmysqluser = ["172.16.12.7", "cocorobouser"]; //edu數據庫信息
  20. const _localhost = ["127.0.0.1", "pbl"]; //edu數據庫信息
  21. const crypto = require("crypto");
  22. const https = require("https");
  23. const { Pool } = require('pg');
  24. const axios = require('axios');
  25. const multer = require("multer");
  26. const {chromium} = require('playwright');
  27. const TurndownService = require('turndown');
  28. const cheerio = require('cheerio');
  29. const { v4: uuidv4 } = require('uuid');
  30. const loginHandler = require('./login'); // 确保路径正确
  31. const { r2rClient } = require("r2r-js");
  32. const client = new r2rClient("https://r2rserver.cocorobo.cn/");
  33. // 配置 multer 存储选项
  34. const storage = multer.diskStorage({
  35. destination: function (req, file, cb) {
  36. req.setEncoding('utf-8');
  37. cb(null, 'uploads/'); // 文件上传目录
  38. },
  39. filename: function (req, file, cb) {
  40. req.setEncoding('utf-8');
  41. // 使用 Buffer 处理文件名,确保编码正确
  42. const fileExtension = file.originalname.split('.').pop(); // 获取文件后缀
  43. const fileNameWithoutExt = Buffer.from(file.originalname.slice(0, -(fileExtension.length + 1)), 'latin1').toString('utf8'); // 转换编码
  44. cb(null, fileNameWithoutExt + '-' + Date.now() + '.' + fileExtension); // 使用时间戳和原始文件名
  45. }
  46. });
  47. const upload = multer({
  48. storage: storage, limits: {
  49. fileSize: 100 * 1024 * 1024 // 100MB
  50. }
  51. });
  52. // 全局浏览器实例
  53. let globalBrowser = null;
  54. async function initBrowser() {
  55. if (!globalBrowser) {
  56. globalBrowser = await chromium.launch({headless: true});
  57. console.log('Browser initialized');
  58. }
  59. return globalBrowser;
  60. }
  61. router.route("/login").all(async (req, res, next) => {
  62. loginHandler(req, res)
  63. });
  64. router.route("/logout").all(async (req, res, next) => {
  65. try {
  66. const response = await client.users.logout();
  67. res.status(200).json({ result: response.results });
  68. } catch (error) {
  69. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  70. }
  71. });
  72. //获取文档列表
  73. router.route("/getDocument").all(async (req, res, next) => {
  74. try {
  75. const response = await client.documents.list({
  76. limit: 10,
  77. offset: 0,
  78. });
  79. console.log(response);
  80. res.status(200).json({ result: response.results });
  81. } catch (error) {
  82. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  83. }
  84. });
  85. //列出文档块
  86. router.route("/listChunks").all(async (req, res, next) => {
  87. try {
  88. const response = await client.documents.listChunks({
  89. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  90. });
  91. console.log(response);
  92. res.status(200).json({ result: response.results });
  93. } catch (error) {
  94. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  95. }
  96. });
  97. //提取实体和关系
  98. router.route("/extract").all(async (req, res, next) => {
  99. try {
  100. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  101. console.log(response);
  102. res.status(200).json({ result: response.results });
  103. } catch (error) {
  104. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  105. }
  106. });
  107. router.route("/selectUser").all((req, res, next) => {
  108. var json = queryString(req.url);
  109. getmysql(req, res, "selectUser", json["userid"]);
  110. });
  111. // 文件上传接口
  112. router.post('/upload', upload.single('file'), async (req, res) => {
  113. req.setEncoding('utf-8')
  114. try {
  115. console.log(req.file); // 打印上传的文件信息
  116. console.log(`File path: uploads/${req.file.filename}`);
  117. const path = require('path');
  118. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  119. console.log(filePath);
  120. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  121. const fs = require('fs');
  122. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  123. const response = await client.documents.create({
  124. file: file,
  125. metadata: { title: filename }
  126. });
  127. console.log(response);
  128. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  129. } catch (error) {
  130. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  131. }
  132. });
  133. //获取文档列表
  134. router.route("/getFile").all(async (req, res, next) => {
  135. //mode的处理
  136. if (req.body[0]) {
  137. p = Object.values(req.body[0]);
  138. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  139. mysql.usselect(p, async function (ret) {
  140. let data = ret[0];
  141. // const array = data.map(item => item.documentid);
  142. const array = data.map(item => item.documentid);
  143. try {
  144. const response = await client.documents.list({ ids: array });
  145. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  146. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  147. // console.log(response);
  148. let data2 = response.results.map(doc => {
  149. doc.documentid = doc.id;
  150. delete doc.id;
  151. return doc;
  152. });
  153. // console.log(data2);
  154. // 合并data和data2,根据documentid
  155. const mergedData = data.map(item => {
  156. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  157. return { ...item, ...correspondingDoc };
  158. });
  159. res.status(200).json({ result: mergedData });
  160. } catch (error) {
  161. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  162. }
  163. });
  164. }
  165. });
  166. //获取文档列表
  167. router.route("/getFileR2r").all(async (req, res, next) => {
  168. //mode的处理
  169. if (req.body[0]) {
  170. let fileid = decodeURIComponent(req.body[0].fileids).split(",");
  171. try {
  172. const response = await client.documents.list({ ids: fileid });
  173. res.status(200).json({ result: response.results });
  174. } catch (error) {
  175. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  176. }
  177. }
  178. });
  179. //获取文档列表
  180. router.route("/getFileD").all(async (req, res, next) => {
  181. //mode的处理
  182. if (req.body[0]) {
  183. p = Object.values(req.body[0]);
  184. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  185. mysql.usselect(p, async function (ret) {
  186. let data = ret[0];
  187. // const array = data.map(item => item.documentid);
  188. const array = data.map(item => item.documentid);
  189. try {
  190. const response = await client.documents.list({ ids: array });
  191. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  192. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  193. // console.log(response);
  194. let data2 = response.results.map(doc => {
  195. doc.documentid = doc.id;
  196. delete doc.id;
  197. return doc;
  198. });
  199. // console.log(data2);
  200. // 合并data和data2,根据documentid
  201. const mergedData = data.map(item => {
  202. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  203. return { ...item, ...correspondingDoc };
  204. });
  205. res.status(200).json({ result: mergedData });
  206. } catch (error) {
  207. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  208. }
  209. });
  210. }
  211. });
  212. //获取文档列表
  213. router.route("/getFileArray").all(async (req, res, next) => {
  214. //mode的处理
  215. if (req.body[0]) {
  216. // const array = data.map(item => item.documentid);
  217. try {
  218. let data = decodeURIComponent(req.body[0].array).split(",");
  219. const response = await client.documents.list({ ids: data });
  220. res.status(200).json({ result: response.results });
  221. } catch (error) {
  222. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  223. }
  224. }
  225. });
  226. //获取文档列表
  227. router.route("/getFileD2").all(async (req, res, next) => {
  228. //mode的处理
  229. if (req.body) {
  230. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  231. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  232. mysql.usselect(p, async function (ret) {
  233. let data = ret[0];
  234. // const array = data.map(item => item.documentid);
  235. const array = data.map(item => item.documentid);
  236. try {
  237. const response = await client.documents.list({ ids: array });
  238. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  239. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  240. // console.log(response);
  241. let data2 = response.results.map(doc => {
  242. doc.documentid = doc.id;
  243. delete doc.id;
  244. return doc;
  245. });
  246. // console.log(data2);
  247. // 合并data和data2,根据documentid
  248. const mergedData = data.map(item => {
  249. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  250. return { ...item, ...correspondingDoc };
  251. });
  252. res.status(200).json({ result: mergedData });
  253. } catch (error) {
  254. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  255. }
  256. });
  257. }
  258. });
  259. //获取文档列表
  260. router.route("/getFileD2").all(async (req, res, next) => {
  261. //mode的处理
  262. if (req.body) {
  263. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  264. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  265. mysql.usselect(p, async function (ret) {
  266. let data = ret[0];
  267. // const array = data.map(item => item.documentid);
  268. const array = data.map(item => item.documentid);
  269. try {
  270. const response = await client.documents.list({ ids: array });
  271. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  272. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  273. // console.log(response);
  274. let data2 = response.results.map(doc => {
  275. doc.documentid = doc.id;
  276. delete doc.id;
  277. return doc;
  278. });
  279. // console.log(data2);
  280. // 合并data和data2,根据documentid
  281. const mergedData = data.map(item => {
  282. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  283. return { ...item, ...correspondingDoc };
  284. });
  285. res.status(200).json({ result: mergedData });
  286. } catch (error) {
  287. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  288. }
  289. });
  290. }
  291. });
  292. //获取文档列表
  293. router.route("/getFileOpen").all(async (req, res, next) => {
  294. //mode的处理
  295. if (req.body[0]) {
  296. p = Object.values(req.body[0]);
  297. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  298. mysql.usselect(p, async function (ret) {
  299. let data = ret[0];
  300. let user = ret[1];
  301. // const array = data.map(item => item.documentid);
  302. const array = data.map(item => item.documentid);
  303. try {
  304. const response = await client.documents.list({ ids: array });
  305. let data2 = response.results.map(doc => {
  306. doc.documentid = doc.id;
  307. delete doc.id;
  308. return doc;
  309. });
  310. // 合并data和data2,根据documentid
  311. const mergedData = data.map(item => {
  312. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  313. return { ...item, ...correspondingDoc };
  314. });
  315. res.status(200).json({ result: mergedData, user: user });
  316. } catch (error) {
  317. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  318. }
  319. });
  320. }
  321. });
  322. //获取文档列表2
  323. router.route("/getFileP").all(async (req, res, next) => {
  324. //mode的处理
  325. if (req.body) {
  326. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  327. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  328. mysql.usselect(p, async function (ret) {
  329. let data = ret[0];
  330. // const array = data.map(item => item.documentid);
  331. try {
  332. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  333. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  334. // console.log(response);
  335. let data2 = response.results.map(doc => {
  336. doc.documentid = doc.id;
  337. delete doc.id;
  338. return doc;
  339. });
  340. // 合并data和data2,根据documentid
  341. const mergedData = data.map(item => {
  342. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  343. return { ...item, ...correspondingDoc };
  344. });
  345. res.status(200).json({ result: mergedData });
  346. } catch (error) {
  347. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  348. }
  349. });
  350. }
  351. });
  352. //获取能关联的文件
  353. router.route("/getFile2").all(async (req, res, next) => {
  354. //mode的处理
  355. if (req.body[0]) {
  356. p = Object.values(req.body[0]);
  357. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  358. mysql.usselect(p, async function (ret) {
  359. let data = ret[0];
  360. const array = data.map(item => item.documentid);
  361. try {
  362. const response = await client.documents.list({ ids: array });
  363. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  364. // console.log(response);
  365. let data2 = response.results.map(doc => {
  366. doc.documentid = doc.id;
  367. delete doc.id;
  368. return doc;
  369. });
  370. // 合并data和data2,根据documentid
  371. const mergedData = data.map(item => {
  372. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  373. return { ...item, ...correspondingDoc };
  374. });
  375. res.status(200).json({ result: mergedData });
  376. } catch (error) {
  377. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  378. }
  379. });
  380. }
  381. });
  382. //添加文件夹
  383. router.route("/addFolder").all(async (req, res, next) => {
  384. //mode的处理
  385. if (req.body[0]) {
  386. try {
  387. const response = await client.collections.create({
  388. name: decodeURIComponent(req.body[0].n),
  389. description: decodeURIComponent(req.body[0].fd)
  390. });
  391. p = [...Object.values(req.body[0]), response.results.id];
  392. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  393. mysql.usselect(p, async function (ret) {
  394. res.end(JSON.stringify(ret));
  395. });
  396. } catch (error) {
  397. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  398. }
  399. }
  400. });
  401. //添加文件夹
  402. router.route("/addFolderP").all(async (req, res, next) => {
  403. //mode的处理
  404. if (req.body) {
  405. try {
  406. const response = await client.collections.create({
  407. name: decodeURIComponent(req.body.n),
  408. description: decodeURIComponent(req.body.fd)
  409. });
  410. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  411. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  412. mysql.usselect(p, async function (ret) {
  413. res.end(JSON.stringify(ret));
  414. });
  415. } catch (error) {
  416. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  417. }
  418. }
  419. });
  420. //删除文件夹
  421. router.route("/deleteFolder").all(async (req, res, next) => {
  422. //mode的处理
  423. if (req.body[0]) {
  424. try {
  425. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  426. const ids = req.body[0].ids;
  427. // 并行处理文件夹下的文件删除和文件夹删除
  428. await Promise.all(folderids.map(async folderid => {
  429. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  430. await Promise.all(files.results.map(file =>
  431. client.collections.removeDocument({ id: folderid, documentId: file.id })
  432. ));
  433. }));
  434. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  435. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  436. mysql.usselect(params, async function (ret) {
  437. res.end(JSON.stringify(ret));
  438. });
  439. } catch (error) {
  440. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  441. }
  442. }
  443. });
  444. //修改文件夹
  445. router.route("/updateFolder").all(async (req, res, next) => {
  446. //mode的处理
  447. if (req.body[0]) {
  448. try {
  449. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  450. const response = await client.collections.update({
  451. id: folderid,
  452. name: decodeURIComponent(req.body[0].n),
  453. description: decodeURIComponent(req.body[0].d)
  454. });
  455. p = Object.values(folderData); // 获取剩余属性的键
  456. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  457. mysql.usselect(p, async function (ret) {
  458. res.end(JSON.stringify(ret));
  459. });
  460. } catch (error) {
  461. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  462. }
  463. }
  464. });
  465. //修改文件夹
  466. router.route("/updateFolder2").all(async (req, res, next) => {
  467. //mode的处理
  468. if (req.body[0]) {
  469. try {
  470. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  471. const response = await client.collections.update({
  472. id: folderid,
  473. name: decodeURIComponent(req.body[0].n),
  474. description: decodeURIComponent(req.body[0].d)
  475. });
  476. p = Object.values(folderData); // 获取剩余属性的键
  477. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  478. mysql.usselect(p, async function (ret) {
  479. res.end(JSON.stringify(ret));
  480. });
  481. } catch (error) {
  482. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  483. }
  484. }
  485. });
  486. //移除文件
  487. router.route("/removeFile").all(async (req, res, next) => {
  488. if (req.body[0]) {
  489. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  490. let folderid = req.body[0].folderid;
  491. let ids = req.body[0].ids;
  492. // 使用 Promise.all 确保所有删除操作都被执行
  493. const removePromises = documentids.map(documentid =>
  494. client.collections.removeDocument({ id: folderid, documentId: documentid })
  495. );
  496. // 不管client.documents报错没有也要执行removeFile
  497. Promise.all(removePromises).catch(error => {
  498. console.error("移除文件时发生错误:", error);
  499. });
  500. p = [ids, folderid];
  501. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  502. mysql.usselect(p, async function (ret) {
  503. res.end(JSON.stringify(ret));
  504. });
  505. }
  506. });
  507. //移入文件
  508. router.route("/moveFile").all(async (req, res, next) => {
  509. if (req.body[0]) {
  510. try {
  511. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  512. let folderid = req.body[0].folderid
  513. let ids = decodeURIComponent(req.body[0].ids).split(",")
  514. // 使用 Promise.all 确保所有添加操作都被执行
  515. await Promise.all(documentids.map(documentid =>
  516. client.collections.addDocument({
  517. id: folderid,
  518. documentId: documentid
  519. }).catch(error => {
  520. console.error("移动文件时发生错误:", error);
  521. })
  522. ));
  523. for (const id of ids) {
  524. p = [id, folderid];
  525. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  526. await mysql.usselect(p, async function (ret) { });
  527. }
  528. res.status(200).json({ messages: "移动成功" });
  529. } catch (error) {
  530. // 这里不返回错误,确保deleteFile执行完
  531. console.error("处理过程中发生错误:", error);
  532. }
  533. }
  534. });
  535. //删除文件
  536. router.route("/deleteFile").all(async (req, res, next) => {
  537. if (req.body[0]) {
  538. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  539. let ids = req.body[0].ids;
  540. // 使用 Promise.all 确保所有删除操作都被执行
  541. Promise.all(documentids.map(documentid =>
  542. client.documents.delete({ id: documentid }).catch(error => {
  543. console.error("删除文件时发生错误:", error);
  544. })
  545. ));
  546. p = [ids];
  547. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  548. mysql.usselect(p, async function (ret) {
  549. res.end(JSON.stringify(ret));
  550. });
  551. }
  552. });
  553. //删除文件
  554. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  555. if (req.body[0]) {
  556. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  557. let ids = req.body[0].ids;
  558. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  559. console.log("文件删除成功");
  560. }).catch(error => {
  561. console.error("删除文件时发生错误:", error);
  562. })
  563. p = [ids];
  564. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  565. mysql.usselect(p, async function (ret) {
  566. res.end(JSON.stringify(ret));
  567. });
  568. }
  569. });
  570. router.route("/getFolder").all((req, res, next) => {
  571. getmysql2(req, res, "getFolder", req.body.uid);
  572. });
  573. router.route("/getFolder2").all((req, res, next) => {
  574. var json = queryString(req.url);
  575. getmysql2(req, res, "getFolder", json['uid']);
  576. });
  577. router.route("/getFolderPage").all((req, res, next) => {
  578. 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);
  579. });
  580. router.route("/getFolderPage3").all((req, res, next) => {
  581. 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);
  582. });
  583. router.route("/getFolderPage32").all((req, res, next) => {
  584. var json = queryString(req.url);
  585. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  586. });
  587. router.route("/getFolderPageOpen").all((req, res, next) => {
  588. 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);
  589. });
  590. router.route("/haveFile").all((req, res, next) => {
  591. postmysql(req, res, "haveFile");
  592. });
  593. router.route("/haveFolder").all((req, res, next) => {
  594. postmysql(req, res, "haveFolder");
  595. });
  596. //search文件
  597. router.route("/searchFile").all(async (req, res, next) => {
  598. if (req.body) {
  599. try {
  600. let filters = {}
  601. if (req.body.type == 1) {
  602. filters = { "document_id": { "$in": req.body.ids } }
  603. } else if (req.body.type == 2) {
  604. filters = { "collection_ids": { "$in": req.body.ids } }
  605. }
  606. const response = await client.retrieval.search({
  607. query: req.body.query,
  608. search_settings: {
  609. filters: filters,
  610. useSemanticSearch: true,
  611. chunkSettings: {
  612. limit: req.body.limit ? req.body.limit : 10,
  613. enabled: true
  614. },
  615. }
  616. });
  617. res.status(200).json({ result: response.results });
  618. } catch (error) {
  619. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  620. }
  621. }
  622. });
  623. //提取文件实体
  624. router.route("/extractFile").all(async (req, res, next) => {
  625. if (req.body) {
  626. try {
  627. const response = client.documents.extract({
  628. id: req.body.id
  629. })
  630. res.status(200).json({ result: response.results });
  631. } catch (error) {
  632. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  633. }
  634. }
  635. });
  636. //添加上传好的知识库文件
  637. router.route("/addFile").all((req, res, next) => {
  638. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  639. })
  640. //获取文件详情
  641. router.route("/getFileDetail").all(async (req, res, next) => {
  642. if (req.body[0]) {
  643. p = Object.values(req.body[0]);
  644. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  645. mysql.usselect(p, async function (ret) {
  646. let data = ret[0][0];
  647. // const array = data.map(item => item.documentid);
  648. try {
  649. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  650. // console.log(response);
  651. let data2 = response.results[0]
  652. // 合并data和data2,根据documentid
  653. const mergedData = { ...data, ...data2 };
  654. res.status(200).json({ result: mergedData });
  655. } catch (error) {
  656. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  657. }
  658. });
  659. }
  660. });
  661. //获取切片
  662. router.route("/getChunks").all(async (req, res, next) => {
  663. if (req.body[0]) {
  664. try {
  665. const { page = 1, limit = 10, documentId } = req.body[0];
  666. const offset = (page - 1) * limit;
  667. const response = await client.documents.listChunks({
  668. id: documentId,
  669. includeVectors: false,
  670. offset,
  671. limit,
  672. });
  673. res.status(200).json({ result: response });
  674. } catch (error) {
  675. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  676. }
  677. }
  678. });
  679. //获取实体
  680. router.route("/getEntities").all(async (req, res, next) => {
  681. if (req.body[0]) {
  682. try {
  683. const { page = 1, limit = 10, documentId } = req.body[0];
  684. const offset = (page - 1) * limit;
  685. const response = await client.documents.listEntities({
  686. id: documentId,
  687. includeVectors: false,
  688. offset,
  689. limit,
  690. });
  691. res.status(200).json({ result: response });
  692. } catch (error) {
  693. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  694. }
  695. }
  696. });
  697. //获取关系
  698. router.route("/getRelationships").all(async (req, res, next) => {
  699. if (req.body[0]) {
  700. try {
  701. const { page = 1, limit = 10, documentId } = req.body[0];
  702. const offset = (page - 1) * limit;
  703. const response = await client.documents.listRelationships({
  704. id: documentId,
  705. includeVectors: false,
  706. offset,
  707. limit,
  708. });
  709. res.status(200).json({ result: response });
  710. } catch (error) {
  711. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  712. }
  713. }
  714. });
  715. //提取整个文件夹的关系
  716. router.route("/resetGraphs").all(async (req, res, next) => {
  717. if (req.body[0]) {
  718. try {
  719. // const response = await client.graphs.pull({
  720. // collectionId: req.body[0].collectionId,
  721. // });
  722. const response = await client.graphs.reset({
  723. collectionId: req.body[0].collectionId,
  724. });
  725. res.status(200).json({ result: response });
  726. } catch (error) {
  727. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  728. }
  729. }
  730. });
  731. //获取整个文件夹的关系
  732. router.route("/getRelationshipsGraphs").all(async (req, res, next) => {
  733. if (req.body[0]) {
  734. try {
  735. await client.graphs.pull({
  736. collectionId: req.body[0].collectionId,
  737. });
  738. const response = await client.graphs.listRelationships({
  739. collectionId: req.body[0].collectionId,
  740. offset: req.body[0].page ? req.body[0].page : 0,
  741. limit: 1000
  742. });
  743. res.status(200).json({ result: response });
  744. } catch (error) {
  745. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  746. }
  747. }
  748. });
  749. //提取知识图谱
  750. router.route("/getGraphs").all(async (req, res, next) => {
  751. console.log(req.body);
  752. //mode的处理
  753. if (req.body[0]) {
  754. try {
  755. //创建临时文件夹
  756. const folder = await client.collections.create({
  757. name: "临时文件夹",
  758. description: "临时文件夹"
  759. });
  760. let folderid = folder.results.id
  761. console.log(folder.results.id);
  762. //将要提取的文件添加到临时文件夹
  763. let documentids = decodeURIComponent(req.body[0].fileids).split(",");
  764. // 使用 Promise.all 确保所有添加操作都被执行
  765. await Promise.all(documentids.map(documentid =>
  766. client.collections.addDocument({
  767. id: folderid,
  768. documentId: documentid
  769. }).catch(error => {
  770. console.error("移动文件时发生错误:", error);
  771. })
  772. ));
  773. //将临时文件夹提取知识图谱
  774. await client.graphs.pull({
  775. collectionId: folderid
  776. });
  777. let offset = 0;
  778. let results = [];
  779. let totalEntries = 0;
  780. do {
  781. const response = await client.graphs.listRelationships({
  782. collectionId: folderid,
  783. offset: offset,
  784. limit: 1000
  785. });
  786. results = results.concat(response.results);
  787. totalEntries = response.totalEntries;
  788. offset += 1000;
  789. } while (offset < totalEntries);
  790. //移除临时文件夹的文件
  791. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  792. await Promise.all(files.results.map(file =>
  793. client.collections.removeDocument({ id: folderid, documentId: file.id })
  794. ));
  795. //删除临时文件夹
  796. await client.collections.delete({ id: folderid });
  797. res.status(200).json({ result: results });
  798. } catch (error) {
  799. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  800. }
  801. }
  802. });
  803. //提取知识图谱
  804. router.route("/getGraphs2").all(async (req, res, next) => {
  805. console.log(req.body);
  806. //mode的处理
  807. if (req.body[0]) {
  808. try {
  809. let documentIds = decodeURIComponent(req.body[0].fileids).split(",");
  810. // 使用 Promise.all 并通过 map 返回新的 results 数组
  811. let results = await Promise.all(documentIds.map(async documentId => {
  812. try {
  813. const response = await client.documents.listRelationships({
  814. id: documentId,
  815. includeVectors: false,
  816. offset: 0,
  817. limit: 1000, // 修正这里的参数
  818. });
  819. return response.results;
  820. } catch (error) {
  821. console.error(`Error fetching relationships for documentId ${documentId}:`, error);
  822. return []; // 返回空数组
  823. }
  824. }));
  825. // 使用 flat() 合并所有数组
  826. results = results.flat();
  827. res.status(200).json({ result: results });
  828. } catch (error) {
  829. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  830. }
  831. }
  832. });
  833. //获取实体和关系
  834. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  835. if (req.body[0]) {
  836. try {
  837. const { documentId } = req.body[0];
  838. const offset = (page - 1) * limit;
  839. const response = await client.documents.listEntities({
  840. id: documentId,
  841. includeVectors: false,
  842. });
  843. const response2 = await client.documents.listRelationships({
  844. id: documentId,
  845. includeVectors: false,
  846. });
  847. res.status(200).json({ Entities: response, Relationships: response2 });
  848. } catch (error) {
  849. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  850. }
  851. }
  852. });
  853. //提取功能
  854. router.route("/extractFile2").all(async (req, res, next) => {
  855. try {
  856. let setting = {}
  857. if(req.body[0].prompt){
  858. setting.graph_extraction_prompt = decodeURIComponent(req.body[0].prompt)
  859. }
  860. console.log(setting);
  861. const response = await client.documents.extract({ id: req.body[0].documentid, setting });
  862. console.log(response);
  863. res.status(200).json({ result: response.results });
  864. } catch (error) {
  865. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  866. }
  867. });
  868. //复制文件夹
  869. router.route("/copyFolder").all(async (req, res, next) => {
  870. if (req.body[0]) {
  871. const response = await client.collections.create({
  872. name: decodeURIComponent(req.body[0].n),
  873. description: decodeURIComponent(req.body[0].fd)
  874. });
  875. let folderid = response.results.id;
  876. p2 = [req.body[0].folderid]
  877. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  878. mysql.usselect(p2, async function (ret2) {
  879. let file = ret2[0]
  880. if (file.length) {
  881. let documentids = file.map(item => item.documentid);
  882. // 使用 Promise.all 确保所有添加操作都被执行
  883. await Promise.all(documentids.map(documentid =>
  884. client.collections.addDocument({
  885. id: folderid,
  886. documentId: documentid
  887. }).catch(error => {
  888. console.error("移动文件时发生错误:", error);
  889. })
  890. ));
  891. }
  892. let ids = file.map(item => item.fileid);
  893. console.log(ids);
  894. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  895. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  896. mysql.usselect(p, async function (ret) {
  897. res.end(JSON.stringify(ret));
  898. });
  899. });
  900. }
  901. });
  902. // 下载文件功能
  903. router.route("/downloadFile").all(async (req, res, next) => {
  904. var json = queryString(req.url);
  905. let params = {
  906. documentid: (json && json['documentid'] || req.body[0].documentid),
  907. filename: (json && json['filename'] || req.body[0].filename),
  908. };
  909. try {
  910. const response = await client.documents.download({ id: params.documentid });
  911. console.log(response);
  912. // 根据response的类型设置Content-Type
  913. const contentType = response.type || 'application/octet-stream';
  914. res.setHeader('Content-Type', contentType);
  915. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  916. // 直接发送blob数据
  917. res.send(Buffer.from(await response.arrayBuffer()));
  918. } catch (error) {
  919. console.error("下载文件时发生错误:", error);
  920. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  921. }
  922. });
  923. // 下载文件功能
  924. router.route("/downloadFile2").all(async (req, res, next) => {
  925. var json = queryString(req.url);
  926. let params = {
  927. documentid: (json && json['documentid'] || req.body[0].documentid),
  928. filename: (json && json['filename'] || req.body[0].filename),
  929. };
  930. try {
  931. const response = await client.documents.download({ id: params.documentid });
  932. console.log(response);
  933. // 直接返回blob数据
  934. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  935. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  936. const buffer = Buffer.from(await response.arrayBuffer());
  937. res.send(buffer); // 发送blob数据
  938. } catch (error) {
  939. console.error("下载文件时发生错误:", error);
  940. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  941. }
  942. });
  943. //召回测试
  944. router.route("/ragFolder").all(async (req, res, next) => {
  945. let reqJson = req.body[0]
  946. if (reqJson) {
  947. reqJson.query = decodeURIComponent(reqJson.query)
  948. try {
  949. let filters = {}
  950. if (reqJson.type == 1) {
  951. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  952. } else if (reqJson.type == 2) {
  953. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  954. } else if (reqJson.type == 3) {
  955. filters = { "collection_ids": { "$eq": reqJson.ids } }
  956. }
  957. console.log(111);
  958. const response = await client.retrieval.search({
  959. query: reqJson.query,
  960. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  961. // searchMode: 'basic',
  962. searchSettings: {
  963. useHybridSearch: reqJson.useType == 3,
  964. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  965. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  966. hybrid_settings: {
  967. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  968. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  969. },
  970. filters: filters,
  971. limit: reqJson.limit ? reqJson.limit : 100,
  972. },
  973. ragGenerationConfig: {
  974. stream: false,
  975. temperature: reqJson.temperature ? reqJson.temperature : 0,
  976. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  977. }
  978. });
  979. // res.status(200).json({ result: response.results });
  980. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  981. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  982. mysql.usselect(p, async function (ret) {
  983. res.status(200).json({ result: response.results });
  984. });
  985. } catch (error) {
  986. console.log(error);
  987. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  988. }
  989. }
  990. });
  991. //召回测试
  992. router.route("/searchFolder").all(async (req, res, next) => {
  993. let reqJson = req.body[0]
  994. if (reqJson) {
  995. try {
  996. let filters = {}
  997. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  998. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  999. filters = {
  1000. "$or": [
  1001. { "collection_ids": { "$in": folderids } },
  1002. { "document_id": { "$in": documentids } }
  1003. ]
  1004. };
  1005. let params = {
  1006. query: reqJson.query,
  1007. searchSettings: {
  1008. useHybridSearch: reqJson.useType == 3,
  1009. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1010. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1011. hybrid_settings: {
  1012. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1013. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1014. },
  1015. filters: filters,
  1016. limit: reqJson.limit ? reqJson.limit : 100,
  1017. },
  1018. ragGenerationConfig: {
  1019. stream: false,
  1020. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1021. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1022. }
  1023. }
  1024. // console.log(params);
  1025. const response = await client.retrieval.search(params);
  1026. res.status(200).json({ result: response.results });
  1027. } catch (error) {
  1028. console.log(error);
  1029. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1030. }
  1031. }
  1032. });
  1033. //召回测试
  1034. router.route("/searchFolder2").all(async (req, res, next) => {
  1035. let reqJson = req.body
  1036. if (Object.keys(reqJson).length) {
  1037. try {
  1038. let filters = {}
  1039. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  1040. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  1041. filters = {
  1042. "$or": [
  1043. { "collection_ids": { "$in": folderids } },
  1044. { "document_id": { "$in": documentids } }
  1045. ]
  1046. };
  1047. if(!folderids.length && !documentids.length){
  1048. res.status(200).json({ result: [] });
  1049. return
  1050. }
  1051. let params = {
  1052. query: reqJson.query,
  1053. searchSettings: {
  1054. useHybridSearch: reqJson.useType == 3,
  1055. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1056. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1057. hybrid_settings: {
  1058. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  1059. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  1060. },
  1061. filters: filters,
  1062. limit: reqJson.limit ? reqJson.limit : 100,
  1063. },
  1064. ragGenerationConfig: {
  1065. stream: false,
  1066. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1067. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1068. }
  1069. }
  1070. // console.log(params);
  1071. const response = await client.retrieval.search(params);
  1072. res.status(200).json({ result: response.results });
  1073. } catch (error) {
  1074. console.log(error);
  1075. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1076. }
  1077. }else{
  1078. res.status(200).json({ result: "请传参" });
  1079. }
  1080. });
  1081. //召回测试
  1082. router.route("/searchFolder3").all(async (req, res, next) => {
  1083. let reqJson = req.body[0]
  1084. if (reqJson) {
  1085. reqJson.query = decodeURIComponent(reqJson.query)
  1086. try {
  1087. let filters = {}
  1088. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1089. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1090. filters = {
  1091. "$or": [
  1092. { "collection_ids": { "$in": folderids } },
  1093. { "document_id": { "$in": documentids } }
  1094. ]
  1095. };
  1096. if(!folderids.length && !documentids.length){
  1097. res.status(200).json({ result: [] });
  1098. return
  1099. }
  1100. let params = {
  1101. query: reqJson.query,
  1102. searchSettings: {
  1103. useHybridSearch: reqJson.useType == 3,
  1104. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1105. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1106. hybrid_settings: {
  1107. fullTextWeight: reqJson.fullTextWeight,
  1108. semanticWeight: reqJson.semanticWeight,
  1109. },
  1110. filters: filters,
  1111. limit: reqJson.limit ? reqJson.limit : 100,
  1112. },
  1113. ragGenerationConfig: {
  1114. stream: false,
  1115. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1116. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1117. }
  1118. }
  1119. // console.log(params);
  1120. const response = await client.retrieval.search(params);
  1121. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1122. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1123. mysql.usselect(p, async function (ret) {
  1124. res.status(200).json({ result: response.results });
  1125. });
  1126. } catch (error) {
  1127. console.log(error);
  1128. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1129. }
  1130. }
  1131. });
  1132. //召回测试
  1133. router.route("/searchFolder32").all(async (req, res, next) => {
  1134. let reqJson = req.body[0]
  1135. if (reqJson) {
  1136. reqJson.query = decodeURIComponent(reqJson.query)
  1137. try {
  1138. let filters = {}
  1139. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  1140. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  1141. filters = {
  1142. "$or": [
  1143. { "collection_ids": { "$in": folderids } },
  1144. { "document_id": { "$in": documentids } }
  1145. ]
  1146. };
  1147. if(!folderids.length && !documentids.length){
  1148. res.status(200).json({ result: [] });
  1149. return
  1150. }
  1151. let params = {
  1152. query: reqJson.query,
  1153. searchSettings: {
  1154. useHybridSearch: reqJson.useType == 3,
  1155. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1156. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  1157. hybrid_settings: {
  1158. fullTextWeight: reqJson.fullTextWeight,
  1159. semanticWeight: reqJson.semanticWeight,
  1160. },
  1161. filters: filters,
  1162. limit: reqJson.limit ? reqJson.limit : 100,
  1163. },
  1164. ragGenerationConfig: {
  1165. stream: false,
  1166. temperature: reqJson.temperature ? reqJson.temperature : 0,
  1167. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  1168. }
  1169. }
  1170. console.log(params);
  1171. const response = await client.retrieval.search(params);
  1172. res.status(200).json({ result: response.results });
  1173. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  1174. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  1175. // mysql.usselect(p, async function (ret) {
  1176. // });
  1177. } catch (error) {
  1178. console.log(error);
  1179. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  1180. }
  1181. }
  1182. });
  1183. //获取召回测试数据
  1184. router.route("/getRagData").all(async (req, res, next) => {
  1185. //mode的处理
  1186. if (req.body[0]) {
  1187. p = Object.values(req.body[0]);
  1188. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  1189. mysql.usselect(p, async function (ret) {
  1190. res.status(200).json(ret);
  1191. });
  1192. }
  1193. });
  1194. //删除召回测试数据
  1195. router.route("/delRagData").all(async (req, res, next) => {
  1196. //mode的处理
  1197. if (req.body[0]) {
  1198. p = Object.values(req.body[0]);
  1199. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1200. mysql.usselect(p, async function (ret) {
  1201. res.status(200).json(ret);
  1202. });
  1203. }
  1204. });
  1205. function generateAPIKey(username) {
  1206. let timeStamp = new Date().getTime();
  1207. let hashStr = timeStamp + username;
  1208. let digest = crypto
  1209. .createHash("sha256")
  1210. .update(hashStr, "utf8")
  1211. .digest("hex");
  1212. return digest;
  1213. }
  1214. queryString = function (url) {
  1215. var a = url.split("?");
  1216. var json = querystring.parse(a[1]);
  1217. return json;
  1218. };
  1219. //post存儲過程處理
  1220. postmysql = function (req, res, functionname) {
  1221. //mode的處理
  1222. if (req.body[0]) {
  1223. p = Object.values(req.body[0]);
  1224. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1225. //執行存儲過程
  1226. mysql.usselect(p, function (ret) {
  1227. res.end(JSON.stringify(ret));
  1228. });
  1229. }
  1230. };
  1231. //post存儲過程處理
  1232. postmysql2 = function (req, res, functionname) {
  1233. //mode的處理
  1234. if (req.body) {
  1235. p = Object.values(req.body)[0].split(",");
  1236. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1237. //執行存儲過程
  1238. mysql.usselect(p, function (ret) {
  1239. res.end(JSON.stringify(ret));
  1240. });
  1241. }
  1242. };
  1243. //post存儲過程處理
  1244. postmysql3 = function (req, res, functionname) {
  1245. //mode的處理
  1246. if (req.body) {
  1247. console.log(req);
  1248. p = Object.values(req.body);
  1249. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1250. //執行存儲過程
  1251. mysql.usselect(p, function (ret) {
  1252. res.end(JSON.stringify(ret));
  1253. });
  1254. }
  1255. };
  1256. //get獲取數據庫信息
  1257. getmysql = function (req, res, functionname, ...p) {
  1258. p = p || [];
  1259. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1260. //執行存儲過程
  1261. mysql.usselect(p, function (ret) {
  1262. res.end(JSON.stringify(ret));
  1263. }); //
  1264. };
  1265. //get獲取數據庫信息
  1266. getmysql2 = function (req, res, functionname, ...p) {
  1267. p = p || [];
  1268. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1269. //執行存儲過程
  1270. mysql.usselect(p, function (ret) {
  1271. res.end(JSON.stringify(ret));
  1272. }); //
  1273. };
  1274. //get獲取數據庫信息
  1275. getmysql3 = function (req, res, functionname, ...p) {
  1276. p = p || [];
  1277. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1278. //執行存儲過程
  1279. mysql.usselect(p, function (ret) {
  1280. res.end(JSON.stringify(ret));
  1281. }); //
  1282. };
  1283. //post存储过程处理
  1284. userpostmysql = function (req, res, functionname) {
  1285. //mode的处理
  1286. if (req.body.mode) {
  1287. p = req.body.mode.split(",");
  1288. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1289. //执行存储过程
  1290. mysql.usselect(p, function (ret) {
  1291. res.end(JSON.stringify(ret));
  1292. });
  1293. }
  1294. };
  1295. //get获取数据库信息
  1296. usergetmysql = function (req, res, functionname, ...p) {
  1297. p = p || [];
  1298. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1299. //执行存储过程
  1300. mysql.usselect(p, function (ret) {
  1301. res.end(JSON.stringify(ret));
  1302. }); //
  1303. };
  1304. //post存儲過程處理
  1305. postmysqlLocal = function (req, res) {
  1306. //mode的處理
  1307. if (req.body[0]) {
  1308. p = Object.values(req.body[0]);
  1309. p.unshift(_localhost[0], _localhost[1]);
  1310. //執行存儲過程
  1311. mysql.usselect(p, function (ret) {
  1312. res.end(JSON.stringify(ret));
  1313. });
  1314. }
  1315. };
  1316. postmysqlOnline = function (req, res) {
  1317. //mode的處理
  1318. if (req.body[0]) {
  1319. p = Object.values(req.body[0]);
  1320. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1321. //執行存儲過程
  1322. mysql.usselect(p, function (ret) {
  1323. res.end(JSON.stringify(ret));
  1324. });
  1325. }
  1326. };
  1327. // 新增测试接口,直接返回成功
  1328. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1329. // console.log('收到的 req.body:', req.body);
  1330. // console.log('收到的 req.files:', req.files);
  1331. // console.log('收到的 headers:', req.headers);
  1332. const requestUrl = req.body.url; // 前端传入 url
  1333. if (!requestUrl) {
  1334. return res.status(400).json({error: 'Missing url'});
  1335. }
  1336. const browser = await initBrowser();
  1337. const page = await browser.newPage();
  1338. try {
  1339. console.log("开始抓取页面:", requestUrl);
  1340. // await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1341. await page.goto(requestUrl);
  1342. const html = await page.content();
  1343. // 新增:HTML 转 Markdown
  1344. const turndownService = new TurndownService();
  1345. // 用 cheerio 只提取正文部分
  1346. const cheerioDoc = cheerio.load(html);
  1347. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1348. // 用 cheerio 过滤无关标签
  1349. const cheerioMain = cheerio.load(mainHtml || '');
  1350. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1351. mainHtml = cheerioMain.html();
  1352. const markdownContent = turndownService.turndown(mainHtml);
  1353. // 新增:解析 <title> 标签内容
  1354. let title = '';
  1355. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1356. if (titleMatch && titleMatch[1]) {
  1357. title = titleMatch[1].trim();
  1358. }
  1359. let finalTitle;
  1360. if (title && title.trim()) {
  1361. finalTitle = title + '.md';
  1362. } else {
  1363. finalTitle = 'untitled-' + uuidv4() + '.md';
  1364. }
  1365. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1366. // 解析用户传入的参数
  1367. const {
  1368. collection_ids,
  1369. id: customId,
  1370. metadata: customMetadata,
  1371. ingestion_mode,
  1372. run_with_orchestration
  1373. } = req.body;
  1374. // 处理 metadata 字段
  1375. let finalMetadata = customMetadata;
  1376. if (typeof finalMetadata === 'string') {
  1377. try {
  1378. finalMetadata = JSON.parse(finalMetadata);
  1379. } catch (e) {
  1380. return res.status(400).json({ error: 'metadata 不是合法的 JSON 字符串' });
  1381. }
  1382. }
  1383. // if (!finalMetadata) {
  1384. // finalMetadata = {
  1385. // title: title || uuidv4(),
  1386. // // source_url: url,
  1387. // content_type: "markdown",
  1388. // // scraped_at: new Date().toISOString(),
  1389. // // upload_id: Date.now() + '-' + Math.random()
  1390. // };
  1391. // }
  1392. // 处理 collection_ids 字段
  1393. let finalCollectionIds = collection_ids;
  1394. if (typeof finalCollectionIds === 'string') {
  1395. try {
  1396. finalCollectionIds = JSON.parse(finalCollectionIds);
  1397. } catch (e) {
  1398. // 不是合法 JSON 字符串就当作普通字符串
  1399. }
  1400. }
  1401. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1402. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1403. finalMetadata.collection_ids = finalCollectionIds;
  1404. }
  1405. // 补全 finalMetadata.title 和 content_type
  1406. if (!finalMetadata.title) {
  1407. finalMetadata.title = finalTitle;
  1408. }
  1409. finalMetadata.content_type = 'markdown';
  1410. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1411. const FormData = require('form-data');
  1412. const fs = require('fs');
  1413. const path = require('path');
  1414. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1415. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1416. const form = new FormData();
  1417. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1418. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1419. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1420. }
  1421. if (customId) {
  1422. form.append('id', customId);
  1423. }
  1424. form.append('metadata', JSON.stringify(finalMetadata));
  1425. form.append('ingestion_mode', ingestion_mode || 'fast');
  1426. form.append('run_with_orchestration', 'false');
  1427. // 打印上传参数,方便排查问题
  1428. console.log('Request Headers:', form.getHeaders());
  1429. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1430. const response = await axios.post(url, form, {
  1431. headers: {
  1432. ...form.getHeaders(),
  1433. // 'User-Agent': 'NodeClient/1.0'
  1434. }
  1435. });
  1436. fs.unlinkSync(tmpFilePath);
  1437. result = {
  1438. results: response.data.results,
  1439. title: finalTitle
  1440. }
  1441. res.json(result);
  1442. } catch (error) {
  1443. console.error('Message:', error.message);
  1444. console.error('Stack:', error.stack);
  1445. res.status(500).json({error: error.message});
  1446. } finally {
  1447. await page.close(); // 确保页面总是被关闭
  1448. }
  1449. });
  1450. module.exports = router;