openai.ts 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499
  1. import assert from 'node:assert';
  2. import { Readable, Transform } from 'stream';
  3. import { pipeline } from 'stream/promises';
  4. import {
  5. IUser,
  6. IUserGroup, PageGrant, Ref, getIdForRef, isPopulated,
  7. } from '@growi/core';
  8. import { isGrobPatternPath } from '@growi/core/dist/utils/page-path-utils';
  9. import escapeStringRegexp from 'escape-string-regexp';
  10. import type { HydratedDocument, Types } from 'mongoose';
  11. import mongoose from 'mongoose';
  12. import type OpenAI from 'openai';
  13. import { toFile } from 'openai';
  14. import ThreadRelationModel from '~/features/openai/server/models/thread-relation';
  15. import VectorStoreModel, { type VectorStoreDocument } from '~/features/openai/server/models/vector-store';
  16. import VectorStoreFileRelationModel, {
  17. type VectorStoreFileRelation,
  18. prepareVectorStoreFileRelations,
  19. } from '~/features/openai/server/models/vector-store-file-relation';
  20. import type { PageDocument, PageModel } from '~/server/models/page';
  21. import userGroupRelation from '~/server/models/user-group-relation';
  22. import { configManager } from '~/server/service/config-manager';
  23. import { createBatchStream } from '~/server/util/batch-stream';
  24. import loggerFactory from '~/utils/logger';
  25. import { OpenaiServiceTypes } from '../../interfaces/ai';
  26. import { type AiAssistant, AiAssistantAccessScope, AiAssistantShareScope } from '../../interfaces/ai-assistant';
  27. import AiAssistantModel, { type AiAssistantDocument } from '../models/ai-assistant';
  28. import { convertMarkdownToHtml } from '../utils/convert-markdown-to-html';
  29. import { getClient } from './client-delegator';
  30. // import { splitMarkdownIntoChunks } from './markdown-splitter/markdown-token-splitter';
  31. import { oepnaiApiErrorHandler } from './openai-api-error-handler';
  32. const BATCH_SIZE = 100;
  33. const logger = loggerFactory('growi:service:openai');
  34. // const isVectorStoreForPublicScopeExist = false;
  35. type VectorStoreFileRelationsMap = Map<string, VectorStoreFileRelation>
  36. const convertPathPatternsToRegExp = (pagePathPatterns: string[]): Array<string | RegExp> => {
  37. return pagePathPatterns.map((pagePathPattern) => {
  38. if (isGrobPatternPath(pagePathPattern)) {
  39. const trimedPagePathPattern = pagePathPattern.replace('/*', '');
  40. const escapedPagePathPattern = escapeStringRegexp(trimedPagePathPattern);
  41. return new RegExp(`^${escapedPagePathPattern}`);
  42. }
  43. return pagePathPattern;
  44. });
  45. };
  46. export interface IOpenaiService {
  47. getOrCreateThread(userId: string, vectorStoreId?: string, threadId?: string): Promise<OpenAI.Beta.Threads.Thread | undefined>;
  48. // getOrCreateVectorStoreForPublicScope(): Promise<VectorStoreDocument>;
  49. deleteExpiredThreads(limit: number, apiCallInterval: number): Promise<void>; // for CronJob
  50. deleteObsolatedVectorStoreRelations(): Promise<void> // for CronJob
  51. createVectorStoreFile(vectorStoreRelation: VectorStoreDocument, pages: PageDocument[]): Promise<void>;
  52. deleteVectorStoreFile(vectorStoreRelationId: Types.ObjectId, pageId: Types.ObjectId): Promise<void>;
  53. deleteObsoleteVectorStoreFile(limit: number, apiCallInterval: number): Promise<void>; // for CronJob
  54. // rebuildVectorStoreAll(): Promise<void>;
  55. // rebuildVectorStore(page: HydratedDocument<PageDocument>): Promise<void>;
  56. createAiAssistant(data: Omit<AiAssistant, 'vectorStore'>): Promise<AiAssistantDocument>;
  57. }
  58. class OpenaiService implements IOpenaiService {
  59. private get client() {
  60. const openaiServiceType = configManager.getConfig('crowi', 'openai:serviceType');
  61. return getClient({ openaiServiceType });
  62. }
  63. public async getOrCreateThread(userId: string, vectorStoreId?: string, threadId?: string): Promise<OpenAI.Beta.Threads.Thread> {
  64. if (vectorStoreId != null && threadId == null) {
  65. try {
  66. const thread = await this.client.createThread(vectorStoreId);
  67. await ThreadRelationModel.create({ userId, threadId: thread.id });
  68. return thread;
  69. }
  70. catch (err) {
  71. throw new Error(err);
  72. }
  73. }
  74. const threadRelation = await ThreadRelationModel.findOne({ threadId });
  75. if (threadRelation == null) {
  76. throw new Error('ThreadRelation document is not exists');
  77. }
  78. // Check if a thread entity exists
  79. // If the thread entity does not exist, the thread-relation document is deleted
  80. try {
  81. const thread = await this.client.retrieveThread(threadRelation.threadId);
  82. // Update expiration date if thread entity exists
  83. await threadRelation.updateThreadExpiration();
  84. return thread;
  85. }
  86. catch (err) {
  87. await oepnaiApiErrorHandler(err, { notFoundError: async() => { await threadRelation.remove() } });
  88. throw new Error(err);
  89. }
  90. }
  91. public async deleteExpiredThreads(limit: number, apiCallInterval: number): Promise<void> {
  92. const expiredThreadRelations = await ThreadRelationModel.getExpiredThreadRelations(limit);
  93. if (expiredThreadRelations == null) {
  94. return;
  95. }
  96. const deletedThreadIds: string[] = [];
  97. for await (const expiredThreadRelation of expiredThreadRelations) {
  98. try {
  99. const deleteThreadResponse = await this.client.deleteThread(expiredThreadRelation.threadId);
  100. logger.debug('Delete thread', deleteThreadResponse);
  101. deletedThreadIds.push(expiredThreadRelation.threadId);
  102. // sleep
  103. await new Promise(resolve => setTimeout(resolve, apiCallInterval));
  104. }
  105. catch (err) {
  106. logger.error(err);
  107. }
  108. }
  109. await ThreadRelationModel.deleteMany({ threadId: { $in: deletedThreadIds } });
  110. }
  111. // TODO: https://redmine.weseek.co.jp/issues/160332
  112. // public async getOrCreateVectorStoreForPublicScope(): Promise<VectorStoreDocument> {
  113. // const vectorStoreDocument: VectorStoreDocument | null = await VectorStoreModel.findOne({ scopeType: VectorStoreScopeType.PUBLIC, isDeleted: false });
  114. // if (vectorStoreDocument != null && isVectorStoreForPublicScopeExist) {
  115. // return vectorStoreDocument;
  116. // }
  117. // if (vectorStoreDocument != null && !isVectorStoreForPublicScopeExist) {
  118. // try {
  119. // // Check if vector store entity exists
  120. // // If the vector store entity does not exist, the vector store document is deleted
  121. // await this.client.retrieveVectorStore(vectorStoreDocument.vectorStoreId);
  122. // isVectorStoreForPublicScopeExist = true;
  123. // return vectorStoreDocument;
  124. // }
  125. // catch (err) {
  126. // await oepnaiApiErrorHandler(err, { notFoundError: vectorStoreDocument.markAsDeleted });
  127. // throw new Error(err);
  128. // }
  129. // }
  130. // const newVectorStore = await this.client.createVectorStore(VectorStoreScopeType.PUBLIC);
  131. // const newVectorStoreDocument = await VectorStoreModel.create({
  132. // vectorStoreId: newVectorStore.id,
  133. // scopeType: VectorStoreScopeType.PUBLIC,
  134. // }) as VectorStoreDocument;
  135. // isVectorStoreForPublicScopeExist = true;
  136. // return newVectorStoreDocument;
  137. // }
  138. private async createVectorStore(name: string): Promise<VectorStoreDocument> {
  139. try {
  140. const newVectorStore = await this.client.createVectorStore(name);
  141. const newVectorStoreDocument = await VectorStoreModel.create({
  142. vectorStoreId: newVectorStore.id,
  143. }) as VectorStoreDocument;
  144. return newVectorStoreDocument;
  145. }
  146. catch (err) {
  147. throw new Error(err);
  148. }
  149. }
  150. // TODO: https://redmine.weseek.co.jp/issues/160332
  151. // TODO: https://redmine.weseek.co.jp/issues/156643
  152. // private async uploadFileByChunks(pageId: Types.ObjectId, body: string, vectorStoreFileRelationsMap: VectorStoreFileRelationsMap) {
  153. // const chunks = await splitMarkdownIntoChunks(body, 'gpt-4o');
  154. // for await (const [index, chunk] of chunks.entries()) {
  155. // try {
  156. // const file = await toFile(Readable.from(chunk), `${pageId}-chunk-${index}.md`);
  157. // const uploadedFile = await this.client.uploadFile(file);
  158. // prepareVectorStoreFileRelations(pageId, uploadedFile.id, vectorStoreFileRelationsMap);
  159. // }
  160. // catch (err) {
  161. // logger.error(err);
  162. // }
  163. // }
  164. // }
  165. private async uploadFile(pageId: Types.ObjectId, pagePath: string, revisionBody: string): Promise<OpenAI.Files.FileObject> {
  166. const convertedHtml = await convertMarkdownToHtml({ pagePath, revisionBody });
  167. const file = await toFile(Readable.from(convertedHtml), `${pageId}.html`);
  168. const uploadedFile = await this.client.uploadFile(file);
  169. return uploadedFile;
  170. }
  171. // TODO: https://redmine.weseek.co.jp/issues/160333
  172. // private async deleteVectorStore(vectorStoreScopeType: VectorStoreScopeType): Promise<void> {
  173. // const vectorStoreDocument: VectorStoreDocument | null = await VectorStoreModel.findOne({ scopeType: vectorStoreScopeType, isDeleted: false });
  174. // if (vectorStoreDocument == null) {
  175. // return;
  176. // }
  177. // try {
  178. // await this.client.deleteVectorStore(vectorStoreDocument.vectorStoreId);
  179. // await vectorStoreDocument.markAsDeleted();
  180. // }
  181. // catch (err) {
  182. // await oepnaiApiErrorHandler(err, { notFoundError: vectorStoreDocument.markAsDeleted });
  183. // throw new Error(err);
  184. // }
  185. // }
  186. async createVectorStoreFile(vectorStoreRelation: VectorStoreDocument, pages: Array<HydratedDocument<PageDocument>>): Promise<void> {
  187. // const vectorStore = await this.getOrCreateVectorStoreForPublicScope();
  188. const vectorStoreFileRelationsMap: VectorStoreFileRelationsMap = new Map();
  189. const processUploadFile = async(page: HydratedDocument<PageDocument>) => {
  190. if (page._id != null && page.grant === PageGrant.GRANT_PUBLIC && page.revision != null) {
  191. if (isPopulated(page.revision) && page.revision.body.length > 0) {
  192. const uploadedFile = await this.uploadFile(page._id, page.path, page.revision.body);
  193. prepareVectorStoreFileRelations(vectorStoreRelation._id, page._id, uploadedFile.id, vectorStoreFileRelationsMap);
  194. return;
  195. }
  196. const pagePopulatedToShowRevision = await page.populateDataToShowRevision();
  197. if (pagePopulatedToShowRevision.revision != null && pagePopulatedToShowRevision.revision.body.length > 0) {
  198. const uploadedFile = await this.uploadFile(page._id, page.path, pagePopulatedToShowRevision.revision.body);
  199. prepareVectorStoreFileRelations(vectorStoreRelation._id, page._id, uploadedFile.id, vectorStoreFileRelationsMap);
  200. }
  201. }
  202. };
  203. // Start workers to process results
  204. const workers = pages.map(processUploadFile);
  205. // Wait for all processing to complete.
  206. assert(workers.length <= BATCH_SIZE, 'workers.length must be less than or equal to BATCH_SIZE');
  207. const fileUploadResult = await Promise.allSettled(workers);
  208. fileUploadResult.forEach((result) => {
  209. if (result.status === 'rejected') {
  210. logger.error(result.reason);
  211. }
  212. });
  213. const vectorStoreFileRelations = Array.from(vectorStoreFileRelationsMap.values());
  214. const uploadedFileIds = vectorStoreFileRelations.map(data => data.fileIds).flat();
  215. if (uploadedFileIds.length === 0) {
  216. return;
  217. }
  218. const pageIds = pages.map(page => page._id);
  219. try {
  220. // Save vector store file relation
  221. await VectorStoreFileRelationModel.upsertVectorStoreFileRelations(vectorStoreFileRelations);
  222. // Create vector store file
  223. const createVectorStoreFileBatchResponse = await this.client.createVectorStoreFileBatch(vectorStoreRelation.vectorStoreId, uploadedFileIds);
  224. logger.debug('Create vector store file', createVectorStoreFileBatchResponse);
  225. // Set isAttachedToVectorStore: true when the uploaded file is attached to VectorStore
  226. await VectorStoreFileRelationModel.markAsAttachedToVectorStore(pageIds);
  227. }
  228. catch (err) {
  229. logger.error(err);
  230. // Delete all uploaded files if createVectorStoreFileBatch fails
  231. for await (const pageId of pageIds) {
  232. await this.deleteVectorStoreFile(vectorStoreRelation._id, pageId);
  233. }
  234. }
  235. }
  236. // Deletes all VectorStore documents that are marked as deleted (isDeleted: true) and have no associated VectorStoreFileRelation documents
  237. async deleteObsolatedVectorStoreRelations(): Promise<void> {
  238. const deletedVectorStoreRelations = await VectorStoreModel.find({ isDeleted: true });
  239. if (deletedVectorStoreRelations.length === 0) {
  240. return;
  241. }
  242. const currentVectorStoreRelationIds: Types.ObjectId[] = await VectorStoreFileRelationModel.aggregate([
  243. {
  244. $group: {
  245. _id: '$vectorStoreRelationId',
  246. relationCount: { $sum: 1 },
  247. },
  248. },
  249. { $match: { relationCount: { $gt: 0 } } },
  250. { $project: { _id: 1 } },
  251. ]);
  252. if (currentVectorStoreRelationIds.length === 0) {
  253. return;
  254. }
  255. await VectorStoreModel.deleteMany({ _id: { $nin: currentVectorStoreRelationIds }, isDeleted: true });
  256. }
  257. async deleteVectorStoreFile(vectorStoreRelationId: Types.ObjectId, pageId: Types.ObjectId, apiCallInterval?: number): Promise<void> {
  258. // Delete vector store file and delete vector store file relation
  259. const vectorStoreFileRelation = await VectorStoreFileRelationModel.findOne({ vectorStoreRelationId, page: pageId });
  260. if (vectorStoreFileRelation == null) {
  261. return;
  262. }
  263. const deletedFileIds: string[] = [];
  264. for await (const fileId of vectorStoreFileRelation.fileIds) {
  265. try {
  266. const deleteFileResponse = await this.client.deleteFile(fileId);
  267. logger.debug('Delete vector store file', deleteFileResponse);
  268. deletedFileIds.push(fileId);
  269. if (apiCallInterval != null) {
  270. // sleep
  271. await new Promise(resolve => setTimeout(resolve, apiCallInterval));
  272. }
  273. }
  274. catch (err) {
  275. await oepnaiApiErrorHandler(err, { notFoundError: async() => { deletedFileIds.push(fileId) } });
  276. logger.error(err);
  277. }
  278. }
  279. const undeletedFileIds = vectorStoreFileRelation.fileIds.filter(fileId => !deletedFileIds.includes(fileId));
  280. if (undeletedFileIds.length === 0) {
  281. await vectorStoreFileRelation.remove();
  282. return;
  283. }
  284. vectorStoreFileRelation.fileIds = undeletedFileIds;
  285. await vectorStoreFileRelation.save();
  286. }
  287. async deleteObsoleteVectorStoreFile(limit: number, apiCallInterval: number): Promise<void> {
  288. // Retrieves all VectorStore documents that are marked as deleted
  289. const deletedVectorStoreRelations = await VectorStoreModel.find({ isDeleted: true });
  290. if (deletedVectorStoreRelations.length === 0) {
  291. return;
  292. }
  293. // Retrieves VectorStoreFileRelation documents associated with deleted VectorStore documents
  294. const obsoleteVectorStoreFileRelations = await VectorStoreFileRelationModel.find(
  295. { vectorStoreRelationId: { $in: deletedVectorStoreRelations.map(deletedVectorStoreRelation => deletedVectorStoreRelation._id) } },
  296. ).limit(limit);
  297. if (obsoleteVectorStoreFileRelations.length === 0) {
  298. return;
  299. }
  300. // Delete obsolete VectorStoreFile
  301. for await (const vectorStoreFileRelation of obsoleteVectorStoreFileRelations) {
  302. try {
  303. await this.deleteVectorStoreFile(vectorStoreFileRelation.vectorStoreRelationId, vectorStoreFileRelation.page, apiCallInterval);
  304. }
  305. catch (err) {
  306. logger.error(err);
  307. }
  308. }
  309. }
  310. // TODO: https://redmine.weseek.co.jp/issues/160332
  311. // async rebuildVectorStoreAll() {
  312. // await this.deleteVectorStore(VectorStoreScopeType.PUBLIC);
  313. // // Create all public pages VectorStoreFile
  314. // const Page = mongoose.model<HydratedDocument<PageDocument>, PageModel>('Page');
  315. // const pagesStream = Page.find({ grant: PageGrant.GRANT_PUBLIC }).populate('revision').cursor({ batch_size: BATCH_SIZE });
  316. // const batchStrem = createBatchStream(BATCH_SIZE);
  317. // const createVectorStoreFile = this.createVectorStoreFile.bind(this);
  318. // const createVectorStoreFileStream = new Transform({
  319. // objectMode: true,
  320. // async transform(chunk: HydratedDocument<PageDocument>[], encoding, callback) {
  321. // await createVectorStoreFile(chunk);
  322. // this.push(chunk);
  323. // callback();
  324. // },
  325. // });
  326. // await pipeline(pagesStream, batchStrem, createVectorStoreFileStream);
  327. // }
  328. // async rebuildVectorStore(page: HydratedDocument<PageDocument>) {
  329. // const vectorStore = await this.getOrCreateVectorStoreForPublicScope();
  330. // await this.deleteVectorStoreFile(vectorStore._id, page._id);
  331. // await this.createVectorStoreFile([page]);
  332. // }
  333. private async createVectorStoreFileWithStream(vectorStoreRelation: VectorStoreDocument, conditions: mongoose.FilterQuery<PageDocument>): Promise<void> {
  334. const Page = mongoose.model<HydratedDocument<PageDocument>, PageModel>('Page');
  335. const pagesStream = Page.find({ ...conditions })
  336. .populate('revision')
  337. .cursor({ batchSize: BATCH_SIZE });
  338. const batchStream = createBatchStream(BATCH_SIZE);
  339. const createVectorStoreFile = this.createVectorStoreFile.bind(this);
  340. const createVectorStoreFileStream = new Transform({
  341. objectMode: true,
  342. async transform(chunk: HydratedDocument<PageDocument>[], encoding, callback) {
  343. try {
  344. await createVectorStoreFile(vectorStoreRelation, chunk);
  345. this.push(chunk);
  346. callback();
  347. }
  348. catch (error) {
  349. callback(error);
  350. }
  351. },
  352. });
  353. await pipeline(pagesStream, batchStream, createVectorStoreFileStream);
  354. }
  355. private async createConditionForCreateAiAssistant(data: AiAssistant): Promise<mongoose.FilterQuery<PageDocument>> {
  356. const converterdPagePatgPatterns = convertPathPatternsToRegExp(data.pagePathPatterns);
  357. if (data.accessScope === AiAssistantAccessScope.PUBLIC_ONLY) {
  358. return {
  359. grant: PageGrant.GRANT_PUBLIC,
  360. path: { $in: converterdPagePatgPatterns },
  361. };
  362. }
  363. if (data.accessScope === AiAssistantAccessScope.GROUPS) {
  364. if (data.grantedGroups != null && data.grantedGroups.length > 0) {
  365. const ownerMemberGroups = (await userGroupRelation.findAllUserGroupIdsRelatedToUser(data.owner)).map(group => group.toString());
  366. const isValid = data.grantedGroups.every(group => ownerMemberGroups.includes(getIdForRef(group.item).toString()));
  367. if (!isValid) {
  368. throw new Error('A group to which the owner does not belong is specified.');
  369. }
  370. }
  371. return {
  372. grant: { $in: [PageGrant.GRANT_PUBLIC, PageGrant.GRANT_USER_GROUP] },
  373. path: { $in: converterdPagePatgPatterns },
  374. $or: [
  375. { 'grantedGroups.item': { $in: data.grantedGroups?.map(group => getIdForRef(group.item)) } },
  376. { grant: PageGrant.GRANT_PUBLIC },
  377. ],
  378. };
  379. }
  380. if (data.accessScope === AiAssistantAccessScope.OWNER) {
  381. return {
  382. grant: { $in: [PageGrant.GRANT_PUBLIC, PageGrant.GRANT_OWNER] },
  383. path: { $in: converterdPagePatgPatterns },
  384. $or: [
  385. { grantedUsers: { $in: [getIdForRef(data.owner)] } },
  386. { grant: PageGrant.GRANT_PUBLIC },
  387. ],
  388. };
  389. }
  390. throw new Error('Invalid accessScope value');
  391. }
  392. async createAiAssistant(data: Omit<AiAssistant, 'vectorStore'>): Promise<AiAssistantDocument> {
  393. const conditions = await this.createConditionForCreateAiAssistant(data as AiAssistant);
  394. const vectorStoreRelation = await this.createVectorStore(data.name);
  395. const aiAssistant = await AiAssistantModel.create({
  396. ...data, vectorStore: vectorStoreRelation,
  397. });
  398. // VectorStore creation process does not await
  399. this.createVectorStoreFileWithStream(vectorStoreRelation, conditions);
  400. return aiAssistant;
  401. }
  402. }
  403. let instance: OpenaiService;
  404. export const getOpenaiService = (): IOpenaiService | undefined => {
  405. if (instance != null) {
  406. return instance;
  407. }
  408. const aiEnabled = configManager.getConfig('crowi', 'app:aiEnabled');
  409. const openaiServiceType = configManager.getConfig('crowi', 'openai:serviceType');
  410. if (aiEnabled && openaiServiceType != null && OpenaiServiceTypes.includes(openaiServiceType)) {
  411. instance = new OpenaiService();
  412. return instance;
  413. }
  414. return;
  415. };