pbl.js 53 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441
  1. // ROUTES FOR OUR APIselectClassStudent
  2. // =============================================================================
  3. const express = require("express");
  4. const request = require("request");
  5. const bcrypt = require("bcryptjs");
  6. const router = express.Router(); // get an instance of the express Router
  7. const querystring = require("querystring");
  8. var mysql = require("./mysql");
  9. // const _mysqlLabor = ["183.36.25.93", "pbl"]; //edu數據庫信息
  10. // const _mysqluser = ["183.36.25.93", "cocorobouser"]; //用户数据库信息
  11. // const _getmysqlLabor = ["123.58.32.151", "pbl"]; //edu數據庫信息f
  12. // const _getmysqluser = ["123.58.32.151", "cocorobouser"]; //edu數據庫信息
  13. const _mysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  14. const _mysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  15. // const _getmysqlLabor = ["172.16.12.5", "pbl"]; //edu數據庫信息
  16. // const _getmysqluser = ["172.16.12.5", "cocorobouser"]; //edu數據庫信息
  17. const _getmysqlLabor2 = ["172.16.12.7", "pbl"]; //edu數據庫信息
  18. const _getmysqlLabor = ["172.16.12.7", "pbl"]; //edu數據庫信息
  19. const _getmysqluser = ["172.16.12.7", "cocorobouser"]; //edu數據庫信息
  20. const _localhost = ["127.0.0.1", "pbl"]; //edu數據庫信息
  21. const crypto = require("crypto");
  22. const https = require("https");
  23. const { Pool } = require('pg');
  24. const axios = require('axios');
  25. const multer = require("multer");
  26. const {chromium} = require('playwright');
  27. const TurndownService = require('turndown');
  28. const cheerio = require('cheerio');
  29. const { v4: uuidv4 } = require('uuid');
  30. const loginHandler = require('./login'); // 确保路径正确
  31. const { r2rClient } = require("r2r-js");
  32. const client = new r2rClient("https://r2rserver.cocorobo.cn/");
  33. // 配置 multer 存储选项
  34. const storage = multer.diskStorage({
  35. destination: function (req, file, cb) {
  36. req.setEncoding('utf-8');
  37. cb(null, 'uploads/'); // 文件上传目录
  38. },
  39. filename: function (req, file, cb) {
  40. req.setEncoding('utf-8');
  41. // 使用 Buffer 处理文件名,确保编码正确
  42. const fileExtension = file.originalname.split('.').pop(); // 获取文件后缀
  43. const fileNameWithoutExt = Buffer.from(file.originalname.slice(0, -(fileExtension.length + 1)), 'latin1').toString('utf8'); // 转换编码
  44. cb(null, fileNameWithoutExt + '-' + Date.now() + '.' + fileExtension); // 使用时间戳和原始文件名
  45. }
  46. });
  47. const upload = multer({
  48. storage: storage, limits: {
  49. fileSize: 100 * 1024 * 1024 // 100MB
  50. }
  51. });
  52. router.route("/login").all(async (req, res, next) => {
  53. loginHandler(req, res)
  54. });
  55. router.route("/logout").all(async (req, res, next) => {
  56. try {
  57. const response = await client.users.logout();
  58. res.status(200).json({ result: response.results });
  59. } catch (error) {
  60. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  61. }
  62. });
  63. //获取文档列表
  64. router.route("/getDocument").all(async (req, res, next) => {
  65. try {
  66. const response = await client.documents.list({
  67. limit: 10,
  68. offset: 0,
  69. });
  70. console.log(response);
  71. res.status(200).json({ result: response.results });
  72. } catch (error) {
  73. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  74. }
  75. });
  76. //列出文档块
  77. router.route("/listChunks").all(async (req, res, next) => {
  78. try {
  79. const response = await client.documents.listChunks({
  80. id: "5111165f-965c-5d61-89e3-1f74b0178032",
  81. });
  82. console.log(response);
  83. res.status(200).json({ result: response.results });
  84. } catch (error) {
  85. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  86. }
  87. });
  88. //提取实体和关系
  89. router.route("/extract").all(async (req, res, next) => {
  90. try {
  91. const response = await client.documents.extract({ id: "5111165f-965c-5d61-89e3-1f74b0178032" })
  92. console.log(response);
  93. res.status(200).json({ result: response.results });
  94. } catch (error) {
  95. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  96. }
  97. });
  98. router.route("/selectUser").all((req, res, next) => {
  99. var json = queryString(req.url);
  100. getmysql(req, res, "selectUser", json["userid"]);
  101. });
  102. // 文件上传接口
  103. router.post('/upload', upload.single('file'), async (req, res) => {
  104. req.setEncoding('utf-8')
  105. try {
  106. console.log(req.file); // 打印上传的文件信息
  107. console.log(`File path: uploads/${req.file.filename}`);
  108. const path = require('path');
  109. const filePath = path.resolve(__dirname, 'uploads', req.file.filename);
  110. console.log(filePath);
  111. let filename = Buffer.from(req.file.originalname, 'latin1').toString('utf8');
  112. const fs = require('fs');
  113. const file = new File([fs.readFileSync('./uploads/q-1739170522767.doc')], filename, { type: req.file.mimetype });
  114. const response = await client.documents.create({
  115. file: file,
  116. metadata: { title: filename }
  117. });
  118. console.log(response);
  119. res.status(200).json({ message: 'File uploaded successfully', file: req.file });
  120. } catch (error) {
  121. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  122. }
  123. });
  124. //获取文档列表
  125. router.route("/getFile").all(async (req, res, next) => {
  126. //mode的处理
  127. if (req.body[0]) {
  128. p = Object.values(req.body[0]);
  129. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  130. mysql.usselect(p, async function (ret) {
  131. let data = ret[0];
  132. // const array = data.map(item => item.documentid);
  133. const array = data.map(item => item.documentid);
  134. try {
  135. const response = await client.documents.list({ ids: array });
  136. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  137. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  138. // console.log(response);
  139. let data2 = response.results.map(doc => {
  140. doc.documentid = doc.id;
  141. delete doc.id;
  142. return doc;
  143. });
  144. // console.log(data2);
  145. // 合并data和data2,根据documentid
  146. const mergedData = data.map(item => {
  147. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  148. return { ...item, ...correspondingDoc };
  149. });
  150. res.status(200).json({ result: mergedData });
  151. } catch (error) {
  152. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  153. }
  154. });
  155. }
  156. });
  157. //获取文档列表
  158. router.route("/getFileD").all(async (req, res, next) => {
  159. //mode的处理
  160. if (req.body[0]) {
  161. p = Object.values(req.body[0]);
  162. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  163. mysql.usselect(p, async function (ret) {
  164. let data = ret[0];
  165. // const array = data.map(item => item.documentid);
  166. const array = data.map(item => item.documentid);
  167. try {
  168. const response = await client.documents.list({ ids: array });
  169. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  170. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  171. // console.log(response);
  172. let data2 = response.results.map(doc => {
  173. doc.documentid = doc.id;
  174. delete doc.id;
  175. return doc;
  176. });
  177. // console.log(data2);
  178. // 合并data和data2,根据documentid
  179. const mergedData = data.map(item => {
  180. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  181. return { ...item, ...correspondingDoc };
  182. });
  183. res.status(200).json({ result: mergedData });
  184. } catch (error) {
  185. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  186. }
  187. });
  188. }
  189. });
  190. //获取文档列表
  191. router.route("/getFileD2").all(async (req, res, next) => {
  192. //mode的处理
  193. if (req.body) {
  194. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  195. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileD");
  196. mysql.usselect(p, async function (ret) {
  197. let data = ret[0];
  198. // const array = data.map(item => item.documentid);
  199. const array = data.map(item => item.documentid);
  200. try {
  201. const response = await client.documents.list({ ids: array });
  202. // const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  203. // const response = await client.collections.listDocuments({ collection_ids: req.body[0].folderid.split(",") });
  204. // console.log(response);
  205. let data2 = response.results.map(doc => {
  206. doc.documentid = doc.id;
  207. delete doc.id;
  208. return doc;
  209. });
  210. // console.log(data2);
  211. // 合并data和data2,根据documentid
  212. const mergedData = data.map(item => {
  213. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  214. return { ...item, ...correspondingDoc };
  215. });
  216. res.status(200).json({ result: mergedData });
  217. } catch (error) {
  218. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  219. }
  220. });
  221. }
  222. });
  223. //获取文档列表
  224. router.route("/getFileOpen").all(async (req, res, next) => {
  225. //mode的处理
  226. if (req.body[0]) {
  227. p = Object.values(req.body[0]);
  228. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileOpen");
  229. mysql.usselect(p, async function (ret) {
  230. let data = ret[0];
  231. let user = ret[1];
  232. // const array = data.map(item => item.documentid);
  233. const array = data.map(item => item.documentid);
  234. try {
  235. const response = await client.documents.list({ ids: array });
  236. let data2 = response.results.map(doc => {
  237. doc.documentid = doc.id;
  238. delete doc.id;
  239. return doc;
  240. });
  241. // 合并data和data2,根据documentid
  242. const mergedData = data.map(item => {
  243. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  244. return { ...item, ...correspondingDoc };
  245. });
  246. res.status(200).json({ result: mergedData, user: user });
  247. } catch (error) {
  248. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  249. }
  250. });
  251. }
  252. });
  253. //获取文档列表2
  254. router.route("/getFileP").all(async (req, res, next) => {
  255. //mode的处理
  256. if (req.body) {
  257. p = [req.body.uid, req.body.folderid, req.body.n, req.body.page, req.body.num];
  258. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile");
  259. mysql.usselect(p, async function (ret) {
  260. let data = ret[0];
  261. // const array = data.map(item => item.documentid);
  262. try {
  263. // const response = await client.documents.list({ collection_ids: [req.body.folderid] });
  264. const response = await client.collections.listDocuments({ id: req.body[0].folderid });
  265. // console.log(response);
  266. let data2 = response.results.map(doc => {
  267. doc.documentid = doc.id;
  268. delete doc.id;
  269. return doc;
  270. });
  271. // 合并data和data2,根据documentid
  272. const mergedData = data.map(item => {
  273. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  274. return { ...item, ...correspondingDoc };
  275. });
  276. res.status(200).json({ result: mergedData });
  277. } catch (error) {
  278. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  279. }
  280. });
  281. }
  282. });
  283. //获取能关联的文件
  284. router.route("/getFile2").all(async (req, res, next) => {
  285. //mode的处理
  286. if (req.body[0]) {
  287. p = Object.values(req.body[0]);
  288. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFile2");
  289. mysql.usselect(p, async function (ret) {
  290. let data = ret[0];
  291. const array = data.map(item => item.documentid);
  292. try {
  293. const response = await client.documents.list({ ids: array });
  294. // const response = await client.documents.list({ collection_ids: [req.body[0].folderid] });
  295. // console.log(response);
  296. let data2 = response.results.map(doc => {
  297. doc.documentid = doc.id;
  298. delete doc.id;
  299. return doc;
  300. });
  301. // 合并data和data2,根据documentid
  302. const mergedData = data.map(item => {
  303. const correspondingDoc = data2.find(doc => doc.documentid === item.documentid);
  304. return { ...item, ...correspondingDoc };
  305. });
  306. res.status(200).json({ result: mergedData });
  307. } catch (error) {
  308. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  309. }
  310. });
  311. }
  312. });
  313. //添加文件夹
  314. router.route("/addFolder").all(async (req, res, next) => {
  315. //mode的处理
  316. if (req.body[0]) {
  317. try {
  318. const response = await client.collections.create({
  319. name: decodeURIComponent(req.body[0].n),
  320. description: decodeURIComponent(req.body[0].fd)
  321. });
  322. p = [...Object.values(req.body[0]), response.results.id];
  323. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  324. mysql.usselect(p, async function (ret) {
  325. res.end(JSON.stringify(ret));
  326. });
  327. } catch (error) {
  328. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  329. }
  330. }
  331. });
  332. //添加文件夹
  333. router.route("/addFolderP").all(async (req, res, next) => {
  334. //mode的处理
  335. if (req.body) {
  336. try {
  337. const response = await client.collections.create({
  338. name: decodeURIComponent(req.body.n),
  339. description: decodeURIComponent(req.body.fd)
  340. });
  341. p = [req.body.n, req.body.fd, req.body.uid, req.body.im , response.results.id];
  342. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder");
  343. mysql.usselect(p, async function (ret) {
  344. res.end(JSON.stringify(ret));
  345. });
  346. } catch (error) {
  347. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  348. }
  349. }
  350. });
  351. //删除文件夹
  352. router.route("/deleteFolder").all(async (req, res, next) => {
  353. //mode的处理
  354. if (req.body[0]) {
  355. try {
  356. const folderids = decodeURIComponent(req.body[0].folderids).split(",");
  357. const ids = req.body[0].ids;
  358. // 并行处理文件夹下的文件删除和文件夹删除
  359. await Promise.all(folderids.map(async folderid => {
  360. const files = await client.collections.listDocuments({ id: folderid, offset: 0, limit: 999 });
  361. await Promise.all(files.results.map(file =>
  362. client.collections.removeDocument({ id: folderid, documentId: file.id })
  363. ));
  364. }));
  365. await Promise.all(folderids.map(folderid => client.collections.delete({ id: folderid })));
  366. const params = [_mysqlLabor[0], _mysqlLabor[1], "deleteFolder", ids];
  367. mysql.usselect(params, async function (ret) {
  368. res.end(JSON.stringify(ret));
  369. });
  370. } catch (error) {
  371. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  372. }
  373. }
  374. });
  375. //修改文件夹
  376. router.route("/updateFolder").all(async (req, res, next) => {
  377. //mode的处理
  378. if (req.body[0]) {
  379. try {
  380. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  381. const response = await client.collections.update({
  382. id: folderid,
  383. name: decodeURIComponent(req.body[0].n),
  384. description: decodeURIComponent(req.body[0].d)
  385. });
  386. p = Object.values(folderData); // 获取剩余属性的键
  387. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder2");
  388. mysql.usselect(p, async function (ret) {
  389. res.end(JSON.stringify(ret));
  390. });
  391. } catch (error) {
  392. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  393. }
  394. }
  395. });
  396. //修改文件夹
  397. router.route("/updateFolder2").all(async (req, res, next) => {
  398. //mode的处理
  399. if (req.body[0]) {
  400. try {
  401. const { folderid, ...folderData } = req.body[0]; // 解构赋值,排除folderid
  402. const response = await client.collections.update({
  403. id: folderid,
  404. name: decodeURIComponent(req.body[0].n),
  405. description: decodeURIComponent(req.body[0].d)
  406. });
  407. p = Object.values(folderData); // 获取剩余属性的键
  408. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "updateFolder3");
  409. mysql.usselect(p, async function (ret) {
  410. res.end(JSON.stringify(ret));
  411. });
  412. } catch (error) {
  413. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  414. }
  415. }
  416. });
  417. //移除文件
  418. router.route("/removeFile").all(async (req, res, next) => {
  419. if (req.body[0]) {
  420. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  421. let folderid = req.body[0].folderid;
  422. let ids = req.body[0].ids;
  423. // 使用 Promise.all 确保所有删除操作都被执行
  424. const removePromises = documentids.map(documentid =>
  425. client.collections.removeDocument({ id: folderid, documentId: documentid })
  426. );
  427. // 不管client.documents报错没有也要执行removeFile
  428. Promise.all(removePromises).catch(error => {
  429. console.error("移除文件时发生错误:", error);
  430. });
  431. p = [ids, folderid];
  432. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "removeFile");
  433. mysql.usselect(p, async function (ret) {
  434. res.end(JSON.stringify(ret));
  435. });
  436. }
  437. });
  438. //移入文件
  439. router.route("/moveFile").all(async (req, res, next) => {
  440. if (req.body[0]) {
  441. try {
  442. let documentids = decodeURIComponent(req.body[0].documentids).split(",")
  443. let folderid = req.body[0].folderid
  444. let ids = decodeURIComponent(req.body[0].ids).split(",")
  445. // 使用 Promise.all 确保所有添加操作都被执行
  446. await Promise.all(documentids.map(documentid =>
  447. client.collections.addDocument({
  448. id: folderid,
  449. documentId: documentid
  450. }).catch(error => {
  451. console.error("移动文件时发生错误:", error);
  452. })
  453. ));
  454. for (const id of ids) {
  455. p = [id, folderid];
  456. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "moveFile");
  457. await mysql.usselect(p, async function (ret) { });
  458. }
  459. res.status(200).json({ messages: "移动成功" });
  460. } catch (error) {
  461. // 这里不返回错误,确保deleteFile执行完
  462. console.error("处理过程中发生错误:", error);
  463. }
  464. }
  465. });
  466. //删除文件
  467. router.route("/deleteFile").all(async (req, res, next) => {
  468. if (req.body[0]) {
  469. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  470. let ids = req.body[0].ids;
  471. // 使用 Promise.all 确保所有删除操作都被执行
  472. Promise.all(documentids.map(documentid =>
  473. client.documents.delete({ id: documentid }).catch(error => {
  474. console.error("删除文件时发生错误:", error);
  475. })
  476. ));
  477. p = [ids];
  478. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  479. mysql.usselect(p, async function (ret) {
  480. res.end(JSON.stringify(ret));
  481. });
  482. }
  483. });
  484. //删除文件
  485. router.route("/deleteFileByFilter").all(async (req, res, next) => {
  486. if (req.body[0]) {
  487. let documentids = decodeURIComponent(req.body[0].documentids).split(",");
  488. let ids = req.body[0].ids;
  489. client.documents.deleteByFilter({ filters: {"document_id": { "$in": documentids }} }).then(() => {
  490. console.log("文件删除成功");
  491. }).catch(error => {
  492. console.error("删除文件时发生错误:", error);
  493. })
  494. p = [ids];
  495. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "deleteFile");
  496. mysql.usselect(p, async function (ret) {
  497. res.end(JSON.stringify(ret));
  498. });
  499. }
  500. });
  501. router.route("/getFolder").all((req, res, next) => {
  502. getmysql2(req, res, "getFolder", req.body.uid);
  503. });
  504. router.route("/getFolder2").all((req, res, next) => {
  505. var json = queryString(req.url);
  506. getmysql2(req, res, "getFolder", json['uid']);
  507. });
  508. router.route("/getFolderPage").all((req, res, next) => {
  509. getmysql2(req, res, "getFolderPage2", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].page, req.body[0].num);
  510. });
  511. router.route("/getFolderPage3").all((req, res, next) => {
  512. getmysql2(req, res, "getFolderPage3", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].tag, req.body[0].page, req.body[0].num);
  513. });
  514. router.route("/getFolderPage32").all((req, res, next) => {
  515. var json = queryString(req.url);
  516. getmysql2(req, res, "getFolderPage3", json['uid'], json['n'], json['sub'], json['class'], json['tag'], json['page'], json["num"]);
  517. });
  518. router.route("/getFolderPageOpen").all((req, res, next) => {
  519. getmysql2(req, res, "getFolderPageOpen", req.body[0].uid, req.body[0].n, req.body[0].sub, req.body[0].class, req.body[0].tag, req.body[0].page, req.body[0].num);
  520. });
  521. router.route("/haveFile").all((req, res, next) => {
  522. postmysql(req, res, "haveFile");
  523. });
  524. router.route("/haveFolder").all((req, res, next) => {
  525. postmysql(req, res, "haveFolder");
  526. });
  527. //search文件
  528. router.route("/searchFile").all(async (req, res, next) => {
  529. if (req.body) {
  530. try {
  531. let filters = {}
  532. if (req.body.type == 1) {
  533. filters = { "document_id": { "$in": req.body.ids } }
  534. } else if (req.body.type == 2) {
  535. filters = { "collection_ids": { "$in": req.body.ids } }
  536. }
  537. const response = await client.retrieval.search({
  538. query: req.body.query,
  539. search_settings: {
  540. filters: filters,
  541. useSemanticSearch: true,
  542. chunkSettings: {
  543. limit: req.body.limit ? req.body.limit : 10,
  544. enabled: true
  545. },
  546. }
  547. });
  548. res.status(200).json({ result: response.results });
  549. } catch (error) {
  550. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  551. }
  552. }
  553. });
  554. //提取文件实体
  555. router.route("/extractFile").all(async (req, res, next) => {
  556. if (req.body) {
  557. try {
  558. const response = client.documents.extract({
  559. id: req.body.id
  560. })
  561. res.status(200).json({ result: response.results });
  562. } catch (error) {
  563. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  564. }
  565. }
  566. });
  567. //添加上传好的知识库文件
  568. router.route("/addFile").all((req, res, next) => {
  569. getmysql2(req, res, "add_file", req.body.n, req.body.did, req.body.uid, req.body.fid, req.body.mofid);
  570. })
  571. //获取文件详情
  572. router.route("/getFileDetail").all(async (req, res, next) => {
  573. if (req.body[0]) {
  574. p = Object.values(req.body[0]);
  575. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getFileDetail");
  576. mysql.usselect(p, async function (ret) {
  577. let data = ret[0][0];
  578. // const array = data.map(item => item.documentid);
  579. try {
  580. const response = await client.documents.list({ ids: [req.body[0].documentId] });
  581. // console.log(response);
  582. let data2 = response.results[0]
  583. // 合并data和data2,根据documentid
  584. const mergedData = { ...data, ...data2 };
  585. res.status(200).json({ result: mergedData });
  586. } catch (error) {
  587. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  588. }
  589. });
  590. }
  591. });
  592. //获取切片
  593. router.route("/getChunks").all(async (req, res, next) => {
  594. if (req.body[0]) {
  595. try {
  596. const { page = 1, limit = 10, documentId } = req.body[0];
  597. const offset = (page - 1) * limit;
  598. const response = await client.documents.listChunks({
  599. id: documentId,
  600. includeVectors: false,
  601. offset,
  602. limit,
  603. });
  604. res.status(200).json({ result: response });
  605. } catch (error) {
  606. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  607. }
  608. }
  609. });
  610. //获取实体
  611. router.route("/getEntities").all(async (req, res, next) => {
  612. if (req.body[0]) {
  613. try {
  614. const { page = 1, limit = 10, documentId } = req.body[0];
  615. const offset = (page - 1) * limit;
  616. const response = await client.documents.listEntities({
  617. id: documentId,
  618. includeVectors: false,
  619. offset,
  620. limit,
  621. });
  622. res.status(200).json({ result: response });
  623. } catch (error) {
  624. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  625. }
  626. }
  627. });
  628. //获取关系
  629. router.route("/getRelationships").all(async (req, res, next) => {
  630. if (req.body[0]) {
  631. try {
  632. const { page = 1, limit = 10, documentId } = req.body[0];
  633. const offset = (page - 1) * limit;
  634. const response = await client.documents.listRelationships({
  635. id: documentId,
  636. includeVectors: false,
  637. offset,
  638. limit,
  639. });
  640. res.status(200).json({ result: response });
  641. } catch (error) {
  642. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  643. }
  644. }
  645. });
  646. //获取实体和关系
  647. router.route("/getEntitiesAndRelationships").all(async (req, res, next) => {
  648. if (req.body[0]) {
  649. try {
  650. const { documentId } = req.body[0];
  651. const offset = (page - 1) * limit;
  652. const response = await client.documents.listEntities({
  653. id: documentId,
  654. includeVectors: false,
  655. });
  656. const response2 = await client.documents.listRelationships({
  657. id: documentId,
  658. includeVectors: false,
  659. });
  660. res.status(200).json({ Entities: response, Relationships: response2 });
  661. } catch (error) {
  662. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  663. }
  664. }
  665. });
  666. //提取功能
  667. router.route("/extractFile2").all(async (req, res, next) => {
  668. try {
  669. const response = await client.documents.extract({ id: req.body[0].documentid });
  670. console.log(response);
  671. res.status(200).json({ result: response.results });
  672. } catch (error) {
  673. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  674. }
  675. });
  676. //复制文件夹
  677. router.route("/copyFolder").all(async (req, res, next) => {
  678. if (req.body[0]) {
  679. const response = await client.collections.create({
  680. name: decodeURIComponent(req.body[0].n),
  681. description: decodeURIComponent(req.body[0].fd)
  682. });
  683. let folderid = response.results.id;
  684. p2 = [req.body[0].folderid]
  685. p2.unshift(_mysqlLabor[0], _mysqlLabor[1], "selectFolderFile");
  686. mysql.usselect(p2, async function (ret2) {
  687. let file = ret2[0]
  688. if (file.length) {
  689. let documentids = file.map(item => item.documentid);
  690. // 使用 Promise.all 确保所有添加操作都被执行
  691. await Promise.all(documentids.map(documentid =>
  692. client.collections.addDocument({
  693. id: folderid,
  694. documentId: documentid
  695. }).catch(error => {
  696. console.error("移动文件时发生错误:", error);
  697. })
  698. ));
  699. }
  700. let ids = file.map(item => item.fileid);
  701. console.log(ids);
  702. p = [req.body[0].n, req.body[0].fd, req.body[0].uid, 2, folderid, req.body[0].folderid, ids.join(",")];
  703. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addFolder2");
  704. mysql.usselect(p, async function (ret) {
  705. res.end(JSON.stringify(ret));
  706. });
  707. });
  708. }
  709. });
  710. // 下载文件功能
  711. router.route("/downloadFile").all(async (req, res, next) => {
  712. var json = queryString(req.url);
  713. let params = {
  714. documentid: (json && json['documentid'] || req.body[0].documentid),
  715. filename: (json && json['filename'] || req.body[0].filename),
  716. };
  717. try {
  718. const response = await client.documents.download({ id: params.documentid });
  719. console.log(response);
  720. // 根据response的类型设置Content-Type
  721. const contentType = response.type || 'application/octet-stream';
  722. res.setHeader('Content-Type', contentType);
  723. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  724. // 直接发送blob数据
  725. res.send(Buffer.from(await response.arrayBuffer()));
  726. } catch (error) {
  727. console.error("下载文件时发生错误:", error);
  728. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  729. }
  730. });
  731. // 下载文件功能
  732. router.route("/downloadFile2").all(async (req, res, next) => {
  733. var json = queryString(req.url);
  734. let params = {
  735. documentid: (json && json['documentid'] || req.body[0].documentid),
  736. filename: (json && json['filename'] || req.body[0].filename),
  737. };
  738. try {
  739. const response = await client.documents.download({ id: params.documentid });
  740. console.log(response);
  741. // 直接返回blob数据
  742. res.setHeader('Content-Type', response.type || 'application/octet-stream');
  743. res.setHeader('Content-Disposition', `attachment; filename="${encodeURIComponent(params.filename)}"`);
  744. const buffer = Buffer.from(await response.arrayBuffer());
  745. res.send(buffer); // 发送blob数据
  746. } catch (error) {
  747. console.error("下载文件时发生错误:", error);
  748. res.status(500).json({ error: error instanceof Error ? error.message : "发生错误" });
  749. }
  750. });
  751. //召回测试
  752. router.route("/ragFolder").all(async (req, res, next) => {
  753. let reqJson = req.body[0]
  754. if (reqJson) {
  755. reqJson.query = decodeURIComponent(reqJson.query)
  756. try {
  757. let filters = {}
  758. if (reqJson.type == 1) {
  759. filters = { "document_id": { "$in": reqJson.ids.split(",") } }
  760. } else if (reqJson.type == 2) {
  761. filters = { "collection_ids": { "$in": reqJson.ids.split(",") } }
  762. } else if (reqJson.type == 3) {
  763. filters = { "collection_ids": { "$eq": reqJson.ids } }
  764. }
  765. console.log(111);
  766. const response = await client.retrieval.search({
  767. query: reqJson.query,
  768. // searchMode: reqJson.useType == 1 ? 'basic' : reqJson.useType == 2 ? 'custom' : 'advanced',
  769. // searchMode: 'basic',
  770. searchSettings: {
  771. useHybridSearch: reqJson.useType == 3,
  772. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  773. useFullTextSearch: reqJson.useType == 2 || reqJson.useType == 3,
  774. hybrid_settings: {
  775. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  776. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  777. },
  778. filters: filters,
  779. limit: reqJson.limit ? reqJson.limit : 100,
  780. },
  781. ragGenerationConfig: {
  782. stream: false,
  783. temperature: reqJson.temperature ? reqJson.temperature : 0,
  784. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  785. }
  786. });
  787. // res.status(200).json({ result: response.results });
  788. p = [reqJson.userid, reqJson.ids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  789. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  790. mysql.usselect(p, async function (ret) {
  791. res.status(200).json({ result: response.results });
  792. });
  793. } catch (error) {
  794. console.log(error);
  795. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  796. }
  797. }
  798. });
  799. //召回测试
  800. router.route("/searchFolder").all(async (req, res, next) => {
  801. let reqJson = req.body[0]
  802. if (reqJson) {
  803. try {
  804. let filters = {}
  805. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  806. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  807. filters = {
  808. "$or": [
  809. { "collection_ids": { "$in": folderids } },
  810. { "document_id": { "$in": documentids } }
  811. ]
  812. };
  813. let params = {
  814. query: reqJson.query,
  815. searchSettings: {
  816. useHybridSearch: reqJson.useType == 3,
  817. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  818. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  819. hybrid_settings: {
  820. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  821. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  822. },
  823. filters: filters,
  824. limit: reqJson.limit ? reqJson.limit : 100,
  825. },
  826. ragGenerationConfig: {
  827. stream: false,
  828. temperature: reqJson.temperature ? reqJson.temperature : 0,
  829. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  830. }
  831. }
  832. // console.log(params);
  833. const response = await client.retrieval.search(params);
  834. res.status(200).json({ result: response.results });
  835. } catch (error) {
  836. console.log(error);
  837. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  838. }
  839. }
  840. });
  841. //召回测试
  842. router.route("/searchFolder2").all(async (req, res, next) => {
  843. let reqJson = req.body
  844. if (Object.keys(reqJson).length) {
  845. try {
  846. let filters = {}
  847. let documentids = reqJson.documentids ? reqJson.documentids.split(",").filter(id => id) : [];
  848. let folderids = reqJson.folderids ? reqJson.folderids.split(",").filter(id => id) : [];
  849. filters = {
  850. "$or": [
  851. { "collection_ids": { "$in": folderids } },
  852. { "document_id": { "$in": documentids } }
  853. ]
  854. };
  855. if(!folderids.length && !documentids.length){
  856. res.status(200).json({ result: [] });
  857. return
  858. }
  859. let params = {
  860. query: reqJson.query,
  861. searchSettings: {
  862. useHybridSearch: reqJson.useType == 3,
  863. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  864. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  865. hybrid_settings: {
  866. fullTextWeight: reqJson.fullTextWeight ? reqJson.fullTextWeight : 1,
  867. semanticWeight: reqJson.semanticWeight ? reqJson.semanticWeight : 5,
  868. },
  869. filters: filters,
  870. limit: reqJson.limit ? reqJson.limit : 100,
  871. },
  872. ragGenerationConfig: {
  873. stream: false,
  874. temperature: reqJson.temperature ? reqJson.temperature : 0,
  875. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  876. }
  877. }
  878. // console.log(params);
  879. const response = await client.retrieval.search(params);
  880. res.status(200).json({ result: response.results });
  881. } catch (error) {
  882. console.log(error);
  883. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  884. }
  885. }else{
  886. res.status(200).json({ result: "请传参" });
  887. }
  888. });
  889. //召回测试
  890. router.route("/searchFolder3").all(async (req, res, next) => {
  891. let reqJson = req.body[0]
  892. if (reqJson) {
  893. reqJson.query = decodeURIComponent(reqJson.query)
  894. try {
  895. let filters = {}
  896. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  897. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  898. filters = {
  899. "$or": [
  900. { "collection_ids": { "$in": folderids } },
  901. { "document_id": { "$in": documentids } }
  902. ]
  903. };
  904. if(!folderids.length && !documentids.length){
  905. res.status(200).json({ result: [] });
  906. return
  907. }
  908. let params = {
  909. query: reqJson.query,
  910. searchSettings: {
  911. useHybridSearch: reqJson.useType == 3,
  912. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  913. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  914. hybrid_settings: {
  915. fullTextWeight: reqJson.fullTextWeight,
  916. semanticWeight: reqJson.semanticWeight,
  917. },
  918. filters: filters,
  919. limit: reqJson.limit ? reqJson.limit : 100,
  920. },
  921. ragGenerationConfig: {
  922. stream: false,
  923. temperature: reqJson.temperature ? reqJson.temperature : 0,
  924. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  925. }
  926. }
  927. // console.log(params);
  928. const response = await client.retrieval.search(params);
  929. p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  930. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  931. mysql.usselect(p, async function (ret) {
  932. res.status(200).json({ result: response.results });
  933. });
  934. } catch (error) {
  935. console.log(error);
  936. res.status(500).json({ error: error instanceof Error ? error.message : "An error occurred" });
  937. }
  938. }
  939. });
  940. //召回测试
  941. router.route("/searchFolder32").all(async (req, res, next) => {
  942. let reqJson = req.body[0]
  943. if (reqJson) {
  944. reqJson.query = decodeURIComponent(reqJson.query)
  945. try {
  946. let filters = {}
  947. let documentids = reqJson.documentids ? decodeURIComponent(reqJson.documentids).split(",").filter(id => id) : [];
  948. let folderids = reqJson.folderids ? decodeURIComponent(reqJson.folderids).split(",").filter(id => id) : [];
  949. filters = {
  950. "$or": [
  951. { "collection_ids": { "$in": folderids } },
  952. { "document_id": { "$in": documentids } }
  953. ]
  954. };
  955. if(!folderids.length && !documentids.length){
  956. res.status(200).json({ result: [] });
  957. return
  958. }
  959. let params = {
  960. query: reqJson.query,
  961. searchSettings: {
  962. useHybridSearch: reqJson.useType == 3,
  963. useSemanticSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  964. useFullTextSearch: reqJson.useType == 1 || reqJson.useType == 2 || reqJson.useType == 3,
  965. hybrid_settings: {
  966. fullTextWeight: reqJson.fullTextWeight,
  967. semanticWeight: reqJson.semanticWeight,
  968. },
  969. filters: filters,
  970. limit: reqJson.limit ? reqJson.limit : 100,
  971. },
  972. ragGenerationConfig: {
  973. stream: false,
  974. temperature: reqJson.temperature ? reqJson.temperature : 0,
  975. maxTokens: reqJson.tokens ? reqJson.tokens : 5000
  976. }
  977. }
  978. console.log(params);
  979. const response = await client.retrieval.search(params);
  980. res.status(200).json({ result: response.results });
  981. // p = [reqJson.userid, reqJson.folderids, encodeURIComponent(JSON.stringify(reqJson)), encodeURIComponent(JSON.stringify(response.results))];
  982. // p.unshift(_mysqlLabor[0], _mysqlLabor[1], "addRagData");
  983. // mysql.usselect(p, async function (ret) {
  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. //获取召回测试数据
  992. router.route("/getRagData").all(async (req, res, next) => {
  993. //mode的处理
  994. if (req.body[0]) {
  995. p = Object.values(req.body[0]);
  996. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "getRagData");
  997. mysql.usselect(p, async function (ret) {
  998. res.status(200).json(ret);
  999. });
  1000. }
  1001. });
  1002. //删除召回测试数据
  1003. router.route("/delRagData").all(async (req, res, next) => {
  1004. //mode的处理
  1005. if (req.body[0]) {
  1006. p = Object.values(req.body[0]);
  1007. p.unshift(_mysqlLabor[0], _mysqlLabor[1], "delRagData");
  1008. mysql.usselect(p, async function (ret) {
  1009. res.status(200).json(ret);
  1010. });
  1011. }
  1012. });
  1013. function generateAPIKey(username) {
  1014. let timeStamp = new Date().getTime();
  1015. let hashStr = timeStamp + username;
  1016. let digest = crypto
  1017. .createHash("sha256")
  1018. .update(hashStr, "utf8")
  1019. .digest("hex");
  1020. return digest;
  1021. }
  1022. queryString = function (url) {
  1023. var a = url.split("?");
  1024. var json = querystring.parse(a[1]);
  1025. return json;
  1026. };
  1027. //post存儲過程處理
  1028. postmysql = function (req, res, functionname) {
  1029. //mode的處理
  1030. if (req.body[0]) {
  1031. p = Object.values(req.body[0]);
  1032. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1033. //執行存儲過程
  1034. mysql.usselect(p, function (ret) {
  1035. res.end(JSON.stringify(ret));
  1036. });
  1037. }
  1038. };
  1039. //post存儲過程處理
  1040. postmysql2 = function (req, res, functionname) {
  1041. //mode的處理
  1042. if (req.body) {
  1043. p = Object.values(req.body)[0].split(",");
  1044. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1045. //執行存儲過程
  1046. mysql.usselect(p, function (ret) {
  1047. res.end(JSON.stringify(ret));
  1048. });
  1049. }
  1050. };
  1051. //post存儲過程處理
  1052. postmysql3 = function (req, res, functionname) {
  1053. //mode的處理
  1054. if (req.body) {
  1055. console.log(req);
  1056. p = Object.values(req.body);
  1057. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1058. //執行存儲過程
  1059. mysql.usselect(p, function (ret) {
  1060. res.end(JSON.stringify(ret));
  1061. });
  1062. }
  1063. };
  1064. //get獲取數據庫信息
  1065. getmysql = function (req, res, functionname, ...p) {
  1066. p = p || [];
  1067. p.unshift(_getmysqlLabor[0], _getmysqlLabor[1], functionname);
  1068. //執行存儲過程
  1069. mysql.usselect(p, function (ret) {
  1070. res.end(JSON.stringify(ret));
  1071. }); //
  1072. };
  1073. //get獲取數據庫信息
  1074. getmysql2 = function (req, res, functionname, ...p) {
  1075. p = p || [];
  1076. p.unshift(_mysqlLabor[0], _mysqlLabor[1], functionname);
  1077. //執行存儲過程
  1078. mysql.usselect(p, function (ret) {
  1079. res.end(JSON.stringify(ret));
  1080. }); //
  1081. };
  1082. //get獲取數據庫信息
  1083. getmysql3 = function (req, res, functionname, ...p) {
  1084. p = p || [];
  1085. p.unshift(_getmysqlLabor2[0], _getmysqlLabor2[1], functionname);
  1086. //執行存儲過程
  1087. mysql.usselect(p, function (ret) {
  1088. res.end(JSON.stringify(ret));
  1089. }); //
  1090. };
  1091. //post存储过程处理
  1092. userpostmysql = function (req, res, functionname) {
  1093. //mode的处理
  1094. if (req.body.mode) {
  1095. p = req.body.mode.split(",");
  1096. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1097. //执行存储过程
  1098. mysql.usselect(p, function (ret) {
  1099. res.end(JSON.stringify(ret));
  1100. });
  1101. }
  1102. };
  1103. //get获取数据库信息
  1104. usergetmysql = function (req, res, functionname, ...p) {
  1105. p = p || [];
  1106. p.unshift(_mysqluser[0], _mysqluser[1], functionname);
  1107. //执行存储过程
  1108. mysql.usselect(p, function (ret) {
  1109. res.end(JSON.stringify(ret));
  1110. }); //
  1111. };
  1112. //post存儲過程處理
  1113. postmysqlLocal = function (req, res) {
  1114. //mode的處理
  1115. if (req.body[0]) {
  1116. p = Object.values(req.body[0]);
  1117. p.unshift(_localhost[0], _localhost[1]);
  1118. //執行存儲過程
  1119. mysql.usselect(p, function (ret) {
  1120. res.end(JSON.stringify(ret));
  1121. });
  1122. }
  1123. };
  1124. postmysqlOnline = function (req, res) {
  1125. //mode的處理
  1126. if (req.body[0]) {
  1127. p = Object.values(req.body[0]);
  1128. p.unshift(_mysqlLabor[0], _mysqlLabor[1]);
  1129. //執行存儲過程
  1130. mysql.usselect(p, function (ret) {
  1131. res.end(JSON.stringify(ret));
  1132. });
  1133. }
  1134. };
  1135. // 新增测试接口,直接返回成功
  1136. router.post('/knowledge-base/html-to-markdown', upload.none(), async (req, res) => {
  1137. // console.log('收到的 req.body:', req.body);
  1138. // console.log('收到的 req.files:', req.files);
  1139. // console.log('收到的 headers:', req.headers);
  1140. const requestUrl = req.body.url; // 前端传入 url
  1141. if (!requestUrl) {
  1142. return res.status(400).json({error: 'Missing url'});
  1143. }
  1144. try {
  1145. const browser = await chromium.launch({headless: true});
  1146. const page = await browser.newPage();
  1147. await page.goto(requestUrl, {waitUntil: 'networkidle'});
  1148. const html = await page.content();
  1149. await browser.close();
  1150. // 新增:HTML 转 Markdown
  1151. const turndownService = new TurndownService();
  1152. // 用 cheerio 只提取正文部分
  1153. const cheerioDoc = cheerio.load(html);
  1154. let mainHtml = cheerioDoc('main').html() || cheerioDoc('article').html() || cheerioDoc('#content').html() || cheerioDoc('.content').html() || cheerioDoc('.post').html() || cheerioDoc('body').html();
  1155. // 用 cheerio 过滤无关标签
  1156. const cheerioMain = cheerio.load(mainHtml || '');
  1157. cheerioMain('style, script, noscript, iframe, link, header, footer, nav, form, .ads, .advertisement').remove();
  1158. mainHtml = cheerioMain.html();
  1159. const markdownContent = turndownService.turndown(mainHtml);
  1160. // 新增:解析 <title> 标签内容
  1161. let title = '';
  1162. const titleMatch = html.match(/<title>([\s\S]*?)<\/title>/i);
  1163. if (titleMatch && titleMatch[1]) {
  1164. title = titleMatch[1].trim();
  1165. }
  1166. let finalTitle;
  1167. if (title && title.trim()) {
  1168. finalTitle = title + '.md';
  1169. } else {
  1170. finalTitle = 'untitled-' + uuidv4() + '.md';
  1171. }
  1172. console.log("抓取到的内容长度:", html.length, "markdown的内容长度:", markdownContent.length);
  1173. // 解析用户传入的参数
  1174. const {
  1175. collection_ids,
  1176. id: customId,
  1177. metadata: customMetadata,
  1178. ingestion_mode,
  1179. run_with_orchestration
  1180. } = req.body;
  1181. // 处理 metadata 字段
  1182. let finalMetadata = customMetadata;
  1183. if (typeof finalMetadata === 'string') {
  1184. try {
  1185. finalMetadata = JSON.parse(finalMetadata);
  1186. } catch (e) {
  1187. return res.status(400).json({error: 'metadata 不是合法的 JSON 字符串'});
  1188. }
  1189. }
  1190. // if (!finalMetadata) {
  1191. // finalMetadata = {
  1192. // title: title || uuidv4(),
  1193. // // source_url: url,
  1194. // content_type: "markdown",
  1195. // // scraped_at: new Date().toISOString(),
  1196. // // upload_id: Date.now() + '-' + Math.random()
  1197. // };
  1198. // }
  1199. // 处理 collection_ids 字段
  1200. let finalCollectionIds = collection_ids;
  1201. if (typeof finalCollectionIds === 'string') {
  1202. try {
  1203. finalCollectionIds = JSON.parse(finalCollectionIds);
  1204. } catch (e) {
  1205. // 不是合法 JSON 字符串就当作普通字符串
  1206. }
  1207. }
  1208. // 如果 metadata 里没有 collection_ids,但用户传了 collection_ids,则加进去
  1209. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1210. finalMetadata.collection_ids = finalCollectionIds;
  1211. }
  1212. // 补全 finalMetadata.title 和 content_type
  1213. if (!finalMetadata.title) {
  1214. finalMetadata.title = finalTitle;
  1215. }
  1216. finalMetadata.content_type = 'markdown';
  1217. // 用 axios 发送 multipart/form-data 方式上传到 r2r
  1218. const FormData = require('form-data');
  1219. const fs = require('fs');
  1220. const path = require('path');
  1221. const tmpFilePath = path.join(__dirname, `${title || 'untitled'}-${uuidv4()}.md`);
  1222. fs.writeFileSync(tmpFilePath, markdownContent, 'utf8');
  1223. const form = new FormData();
  1224. form.append('file', fs.createReadStream(tmpFilePath), finalTitle);
  1225. if (finalCollectionIds && Array.isArray(finalCollectionIds)) {
  1226. form.append('collection_ids', JSON.stringify(finalCollectionIds));
  1227. }
  1228. if (customId) {
  1229. form.append('id', customId);
  1230. }
  1231. form.append('metadata', JSON.stringify(finalMetadata));
  1232. form.append('ingestion_mode', ingestion_mode || 'fast');
  1233. form.append('run_with_orchestration', 'false');
  1234. // 打印上传参数,方便排查问题
  1235. console.log('Request Headers:', form.getHeaders());
  1236. const url = 'https://r2rserver.cocorobo.cn/v3/documents';
  1237. const response = await axios.post(url, form, {
  1238. headers: {
  1239. ...form.getHeaders(),
  1240. // 'User-Agent': 'NodeClient/1.0'
  1241. }
  1242. });
  1243. fs.unlinkSync(tmpFilePath);
  1244. result = {
  1245. results: response.data.results,
  1246. title: finalTitle
  1247. }
  1248. res.json(result);
  1249. } catch (error) {
  1250. console.error('Error occurred:', error);
  1251. res.status(500).json({error: error.message});
  1252. }
  1253. });
  1254. module.exports = router;