openai.ts 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375
  1. import assert from 'node:assert';
  2. import { Readable, Transform } from 'stream';
  3. import { pipeline } from 'stream/promises';
  4. import type { IPagePopulatedToShowRevision } from '@growi/core';
  5. import { PageGrant, isPopulated } from '@growi/core';
  6. import type { HydratedDocument, Types } from 'mongoose';
  7. import mongoose from 'mongoose';
  8. import type OpenAI from 'openai';
  9. import { toFile } from 'openai';
  10. import ThreadRelationModel from '~/features/openai/server/models/thread-relation';
  11. import VectorStoreModel, { VectorStoreScopeType, type VectorStoreDocument } from '~/features/openai/server/models/vector-store';
  12. import VectorStoreFileRelationModel, {
  13. type VectorStoreFileRelation,
  14. prepareVectorStoreFileRelations,
  15. } from '~/features/openai/server/models/vector-store-file-relation';
  16. import type { PageDocument, PageModel } from '~/server/models/page';
  17. import { configManager } from '~/server/service/config-manager';
  18. import { createBatchStream } from '~/server/util/batch-stream';
  19. import loggerFactory from '~/utils/logger';
  20. import { OpenaiServiceTypes } from '../../interfaces/ai';
  21. import { convertMarkdownToHtml } from '../utils/convert-markdown-to-html';
  22. import { getClient } from './client-delegator';
  23. // import { splitMarkdownIntoChunks } from './markdown-splitter/markdown-token-splitter';
  24. import { oepnaiApiErrorHandler } from './openai-api-error-handler';
  25. const BATCH_SIZE = 100;
  26. const logger = loggerFactory('growi:service:openai');
  27. let isVectorStoreForPublicScopeExist = false;
  28. type VectorStoreFileRelationsMap = Map<string, VectorStoreFileRelation>
  29. export interface IOpenaiService {
  30. getOrCreateThread(userId: string, vectorStoreId?: string, threadId?: string): Promise<OpenAI.Beta.Threads.Thread | undefined>;
  31. getOrCreateVectorStoreForPublicScope(): Promise<VectorStoreDocument>;
  32. deleteExpiredThreads(limit: number, apiCallInterval: number): Promise<void>; // for CronJob
  33. deleteObsolatedVectorStoreRelations(): Promise<void> // for CronJob
  34. createVectorStoreFile(pages: PageDocument[]): Promise<void>;
  35. deleteVectorStoreFile(vectorStoreRelationId: Types.ObjectId, pageId: Types.ObjectId): Promise<void>;
  36. deleteObsoleteVectorStoreFile(limit: number, apiCallInterval: number): Promise<void>; // for CronJob
  37. rebuildVectorStoreAll(): Promise<void>;
  38. rebuildVectorStore(page: HydratedDocument<PageDocument>): Promise<void>;
  39. }
  40. class OpenaiService implements IOpenaiService {
  41. private get client() {
  42. const openaiServiceType = configManager.getConfig('openai:serviceType');
  43. return getClient({ openaiServiceType });
  44. }
  45. public async getOrCreateThread(userId: string, vectorStoreId?: string, threadId?: string): Promise<OpenAI.Beta.Threads.Thread> {
  46. if (vectorStoreId != null && threadId == null) {
  47. try {
  48. const thread = await this.client.createThread(vectorStoreId);
  49. await ThreadRelationModel.create({ userId, threadId: thread.id });
  50. return thread;
  51. }
  52. catch (err) {
  53. throw new Error(err);
  54. }
  55. }
  56. const threadRelation = await ThreadRelationModel.findOne({ threadId });
  57. if (threadRelation == null) {
  58. throw new Error('ThreadRelation document is not exists');
  59. }
  60. // Check if a thread entity exists
  61. // If the thread entity does not exist, the thread-relation document is deleted
  62. try {
  63. const thread = await this.client.retrieveThread(threadRelation.threadId);
  64. // Update expiration date if thread entity exists
  65. await threadRelation.updateThreadExpiration();
  66. return thread;
  67. }
  68. catch (err) {
  69. await oepnaiApiErrorHandler(err, { notFoundError: async() => { await threadRelation.remove() } });
  70. throw new Error(err);
  71. }
  72. }
  73. public async deleteExpiredThreads(limit: number, apiCallInterval: number): Promise<void> {
  74. const expiredThreadRelations = await ThreadRelationModel.getExpiredThreadRelations(limit);
  75. if (expiredThreadRelations == null) {
  76. return;
  77. }
  78. const deletedThreadIds: string[] = [];
  79. for await (const expiredThreadRelation of expiredThreadRelations) {
  80. try {
  81. const deleteThreadResponse = await this.client.deleteThread(expiredThreadRelation.threadId);
  82. logger.debug('Delete thread', deleteThreadResponse);
  83. deletedThreadIds.push(expiredThreadRelation.threadId);
  84. // sleep
  85. await new Promise(resolve => setTimeout(resolve, apiCallInterval));
  86. }
  87. catch (err) {
  88. logger.error(err);
  89. }
  90. }
  91. await ThreadRelationModel.deleteMany({ threadId: { $in: deletedThreadIds } });
  92. }
  93. public async getOrCreateVectorStoreForPublicScope(): Promise<VectorStoreDocument> {
  94. const vectorStoreDocument: VectorStoreDocument | null = await VectorStoreModel.findOne({ scopeType: VectorStoreScopeType.PUBLIC, isDeleted: false });
  95. if (vectorStoreDocument != null && isVectorStoreForPublicScopeExist) {
  96. return vectorStoreDocument;
  97. }
  98. if (vectorStoreDocument != null && !isVectorStoreForPublicScopeExist) {
  99. try {
  100. // Check if vector store entity exists
  101. // If the vector store entity does not exist, the vector store document is deleted
  102. await this.client.retrieveVectorStore(vectorStoreDocument.vectorStoreId);
  103. isVectorStoreForPublicScopeExist = true;
  104. return vectorStoreDocument;
  105. }
  106. catch (err) {
  107. await oepnaiApiErrorHandler(err, { notFoundError: vectorStoreDocument.markAsDeleted });
  108. throw new Error(err);
  109. }
  110. }
  111. const newVectorStore = await this.client.createVectorStore(VectorStoreScopeType.PUBLIC);
  112. const newVectorStoreDocument = await VectorStoreModel.create({
  113. vectorStoreId: newVectorStore.id,
  114. scopeType: VectorStoreScopeType.PUBLIC,
  115. }) as VectorStoreDocument;
  116. isVectorStoreForPublicScopeExist = true;
  117. return newVectorStoreDocument;
  118. }
  119. // TODO: https://redmine.weseek.co.jp/issues/156643
  120. // private async uploadFileByChunks(pageId: Types.ObjectId, body: string, vectorStoreFileRelationsMap: VectorStoreFileRelationsMap) {
  121. // const chunks = await splitMarkdownIntoChunks(body, 'gpt-4o');
  122. // for await (const [index, chunk] of chunks.entries()) {
  123. // try {
  124. // const file = await toFile(Readable.from(chunk), `${pageId}-chunk-${index}.md`);
  125. // const uploadedFile = await this.client.uploadFile(file);
  126. // prepareVectorStoreFileRelations(pageId, uploadedFile.id, vectorStoreFileRelationsMap);
  127. // }
  128. // catch (err) {
  129. // logger.error(err);
  130. // }
  131. // }
  132. // }
  133. private async uploadFile(pageId: Types.ObjectId, pagePath: string, revisionBody: string): Promise<OpenAI.Files.FileObject> {
  134. const convertedHtml = await convertMarkdownToHtml({ pagePath, revisionBody });
  135. const file = await toFile(Readable.from(convertedHtml), `${pageId}.html`);
  136. const uploadedFile = await this.client.uploadFile(file);
  137. return uploadedFile;
  138. }
  139. private async deleteVectorStore(vectorStoreScopeType: VectorStoreScopeType): Promise<void> {
  140. const vectorStoreDocument: VectorStoreDocument | null = await VectorStoreModel.findOne({ scopeType: vectorStoreScopeType, isDeleted: false });
  141. if (vectorStoreDocument == null) {
  142. return;
  143. }
  144. try {
  145. await this.client.deleteVectorStore(vectorStoreDocument.vectorStoreId);
  146. await vectorStoreDocument.markAsDeleted();
  147. }
  148. catch (err) {
  149. await oepnaiApiErrorHandler(err, { notFoundError: vectorStoreDocument.markAsDeleted });
  150. throw new Error(err);
  151. }
  152. }
  153. async createVectorStoreFile(pages: Array<HydratedDocument<PageDocument>>): Promise<void> {
  154. const vectorStore = await this.getOrCreateVectorStoreForPublicScope();
  155. const vectorStoreFileRelationsMap: VectorStoreFileRelationsMap = new Map();
  156. const processUploadFile = async(page: HydratedDocument<PageDocument>) => {
  157. if (page._id != null && page.grant === PageGrant.GRANT_PUBLIC && page.revision != null) {
  158. if (isPopulated(page.revision) && page.revision.body.length > 0) {
  159. const uploadedFile = await this.uploadFile(page._id, page.path, page.revision.body);
  160. prepareVectorStoreFileRelations(vectorStore._id, page._id, uploadedFile.id, vectorStoreFileRelationsMap);
  161. return;
  162. }
  163. const pagePopulatedToShowRevision = await page.populateDataToShowRevision();
  164. if (pagePopulatedToShowRevision.revision != null && pagePopulatedToShowRevision.revision.body.length > 0) {
  165. const uploadedFile = await this.uploadFile(page._id, page.path, pagePopulatedToShowRevision.revision.body);
  166. prepareVectorStoreFileRelations(vectorStore._id, page._id, uploadedFile.id, vectorStoreFileRelationsMap);
  167. }
  168. }
  169. };
  170. // Start workers to process results
  171. const workers = pages.map(processUploadFile);
  172. // Wait for all processing to complete.
  173. assert(workers.length <= BATCH_SIZE, 'workers.length must be less than or equal to BATCH_SIZE');
  174. const fileUploadResult = await Promise.allSettled(workers);
  175. fileUploadResult.forEach((result) => {
  176. if (result.status === 'rejected') {
  177. logger.error(result.reason);
  178. }
  179. });
  180. const vectorStoreFileRelations = Array.from(vectorStoreFileRelationsMap.values());
  181. const uploadedFileIds = vectorStoreFileRelations.map(data => data.fileIds).flat();
  182. if (uploadedFileIds.length === 0) {
  183. return;
  184. }
  185. const pageIds = pages.map(page => page._id);
  186. try {
  187. // Save vector store file relation
  188. await VectorStoreFileRelationModel.upsertVectorStoreFileRelations(vectorStoreFileRelations);
  189. // Create vector store file
  190. const createVectorStoreFileBatchResponse = await this.client.createVectorStoreFileBatch(vectorStore.vectorStoreId, uploadedFileIds);
  191. logger.debug('Create vector store file', createVectorStoreFileBatchResponse);
  192. // Set isAttachedToVectorStore: true when the uploaded file is attached to VectorStore
  193. await VectorStoreFileRelationModel.markAsAttachedToVectorStore(pageIds);
  194. }
  195. catch (err) {
  196. logger.error(err);
  197. // Delete all uploaded files if createVectorStoreFileBatch fails
  198. for await (const pageId of pageIds) {
  199. await this.deleteVectorStoreFile(vectorStore._id, pageId);
  200. }
  201. }
  202. }
  203. // Deletes all VectorStore documents that are marked as deleted (isDeleted: true) and have no associated VectorStoreFileRelation documents
  204. async deleteObsolatedVectorStoreRelations(): Promise<void> {
  205. const deletedVectorStoreRelations = await VectorStoreModel.find({ isDeleted: true });
  206. if (deletedVectorStoreRelations.length === 0) {
  207. return;
  208. }
  209. const currentVectorStoreRelationIds: Types.ObjectId[] = await VectorStoreFileRelationModel.aggregate([
  210. {
  211. $group: {
  212. _id: '$vectorStoreRelationId',
  213. relationCount: { $sum: 1 },
  214. },
  215. },
  216. { $match: { relationCount: { $gt: 0 } } },
  217. { $project: { _id: 1 } },
  218. ]);
  219. if (currentVectorStoreRelationIds.length === 0) {
  220. return;
  221. }
  222. await VectorStoreModel.deleteMany({ _id: { $nin: currentVectorStoreRelationIds }, isDeleted: true });
  223. }
  224. async deleteVectorStoreFile(vectorStoreRelationId: Types.ObjectId, pageId: Types.ObjectId, apiCallInterval?: number): Promise<void> {
  225. // Delete vector store file and delete vector store file relation
  226. const vectorStoreFileRelation = await VectorStoreFileRelationModel.findOne({ vectorStoreRelationId, page: pageId });
  227. if (vectorStoreFileRelation == null) {
  228. return;
  229. }
  230. const deletedFileIds: string[] = [];
  231. for await (const fileId of vectorStoreFileRelation.fileIds) {
  232. try {
  233. const deleteFileResponse = await this.client.deleteFile(fileId);
  234. logger.debug('Delete vector store file', deleteFileResponse);
  235. deletedFileIds.push(fileId);
  236. if (apiCallInterval != null) {
  237. // sleep
  238. await new Promise(resolve => setTimeout(resolve, apiCallInterval));
  239. }
  240. }
  241. catch (err) {
  242. await oepnaiApiErrorHandler(err, { notFoundError: async() => { deletedFileIds.push(fileId) } });
  243. logger.error(err);
  244. }
  245. }
  246. const undeletedFileIds = vectorStoreFileRelation.fileIds.filter(fileId => !deletedFileIds.includes(fileId));
  247. if (undeletedFileIds.length === 0) {
  248. await vectorStoreFileRelation.remove();
  249. return;
  250. }
  251. vectorStoreFileRelation.fileIds = undeletedFileIds;
  252. await vectorStoreFileRelation.save();
  253. }
  254. async deleteObsoleteVectorStoreFile(limit: number, apiCallInterval: number): Promise<void> {
  255. // Retrieves all VectorStore documents that are marked as deleted
  256. const deletedVectorStoreRelations = await VectorStoreModel.find({ isDeleted: true });
  257. if (deletedVectorStoreRelations.length === 0) {
  258. return;
  259. }
  260. // Retrieves VectorStoreFileRelation documents associated with deleted VectorStore documents
  261. const obsoleteVectorStoreFileRelations = await VectorStoreFileRelationModel.find(
  262. { vectorStoreRelationId: { $in: deletedVectorStoreRelations.map(deletedVectorStoreRelation => deletedVectorStoreRelation._id) } },
  263. ).limit(limit);
  264. if (obsoleteVectorStoreFileRelations.length === 0) {
  265. return;
  266. }
  267. // Delete obsolete VectorStoreFile
  268. for await (const vectorStoreFileRelation of obsoleteVectorStoreFileRelations) {
  269. try {
  270. await this.deleteVectorStoreFile(vectorStoreFileRelation.vectorStoreRelationId, vectorStoreFileRelation.page, apiCallInterval);
  271. }
  272. catch (err) {
  273. logger.error(err);
  274. }
  275. }
  276. }
  277. async rebuildVectorStoreAll() {
  278. await this.deleteVectorStore(VectorStoreScopeType.PUBLIC);
  279. // Create all public pages VectorStoreFile
  280. const Page = mongoose.model<HydratedDocument<PageDocument>, PageModel>('Page');
  281. const pagesStream = Page.find({ grant: PageGrant.GRANT_PUBLIC }).populate('revision').cursor({ batch_size: BATCH_SIZE });
  282. const batchStrem = createBatchStream(BATCH_SIZE);
  283. const createVectorStoreFile = this.createVectorStoreFile.bind(this);
  284. const createVectorStoreFileStream = new Transform({
  285. objectMode: true,
  286. async transform(chunk: HydratedDocument<PageDocument>[], encoding, callback) {
  287. await createVectorStoreFile(chunk);
  288. this.push(chunk);
  289. callback();
  290. },
  291. });
  292. await pipeline(pagesStream, batchStrem, createVectorStoreFileStream);
  293. }
  294. async rebuildVectorStore(page: HydratedDocument<PageDocument>) {
  295. const vectorStore = await this.getOrCreateVectorStoreForPublicScope();
  296. await this.deleteVectorStoreFile(vectorStore._id, page._id);
  297. await this.createVectorStoreFile([page]);
  298. }
  299. }
  300. let instance: OpenaiService;
  301. export const getOpenaiService = (): IOpenaiService | undefined => {
  302. if (instance != null) {
  303. return instance;
  304. }
  305. const aiEnabled = configManager.getConfig('app:aiEnabled');
  306. const openaiServiceType = configManager.getConfig('openai:serviceType');
  307. if (aiEnabled && openaiServiceType != null && OpenaiServiceTypes.includes(openaiServiceType)) {
  308. instance = new OpenaiService();
  309. return instance;
  310. }
  311. return;
  312. };