feat(server): global embedding gql endpoint (#11809)

fix AI-30
fix AI-31
fix PD-2487

packages/backend/server/schema.prisma

@@ -114,13 +114,13 @@ model Workspace {
   name               String?  @db.VarChar
   avatarKey          String?  @map("avatar_key") @db.VarChar
 
-  features               WorkspaceFeature[]
+  features       WorkspaceFeature[]
-  docs                   WorkspaceDoc[]
+  docs           WorkspaceDoc[]
-  permissions            WorkspaceUserRole[]
+  permissions    WorkspaceUserRole[]
-  docPermissions         WorkspaceDocUserRole[]
+  docPermissions WorkspaceDocUserRole[]
-  blobs                  Blob[]
+  blobs          Blob[]
-  AiWorkspaceIgnoredDocs AiWorkspaceIgnoredDocs[]
+  ignoredDocs    AiWorkspaceIgnoredDocs[]
-  AiWorkspaceFiles       AiWorkspaceFiles[]
+  embedFiles     AiWorkspaceFiles[]
 
   @@map("workspaces")
 }

packages/backend/server/src/__tests__/models/copilot-workspace.spec.ts

@@ -104,22 +104,18 @@ test('should manage copilot workspace ignored docs', async t => {
 
 test('should insert and search embedding', async t => {
   {
-    await t.context.copilotWorkspace.addWorkspaceFile(
+    const { fileId } = await t.context.copilotWorkspace.addFile(workspace.id, {
-      workspace.id,
+      fileName: 'file1',
+      mimeType: 'text/plain',
+      size: 1,
+    });
+    await t.context.copilotWorkspace.addFileEmbeddings(workspace.id, fileId, [
       {
-        fileName: 'file1',
+        index: 0,
-        mimeType: 'text/plain',
+        content: 'content',
-
+        embedding: Array.from({ length: 1024 }, () => 1),
-        size: 1,
       },
-      [
+    ]);
-        {
-          index: 0,
-          content: 'content',
-          embedding: Array.from({ length: 1024 }, () => 1),
-        },
-      ]
-    );
 
     {
       const ret = await t.context.copilotWorkspace.matchWorkspaceFileEmbedding(

packages/backend/server/src/base/error/def.ts

@@ -711,15 +711,21 @@ export const USER_FRIENDLY_ERRORS = {
   },
   copilot_transcription_job_exists: {
     type: 'bad_request',
-    message: () => 'Transcription job already exists',
+    message: 'Transcription job already exists',
   },
   copilot_transcription_job_not_found: {
     type: 'bad_request',
-    message: () => `Transcription job not found.`,
+    message: `Transcription job not found.`,
   },
   copilot_transcription_audio_not_provided: {
     type: 'bad_request',
-    message: () => `Audio not provided.`,
+    message: `Audio not provided.`,
+  },
+  copilot_failed_to_add_workspace_file_embedding: {
+    type: 'internal_server_error',
+    args: { message: 'string' },
+    message: ({ message }) =>
+      `Failed to add workspace file embedding: ${message}`,
   },
 
   // Quota & Limit errors

packages/backend/server/src/base/error/errors.gen.ts

@@ -788,6 +788,16 @@ export class CopilotTranscriptionAudioNotProvided extends UserFriendlyError {
     super('bad_request', 'copilot_transcription_audio_not_provided', message);
   }
 }
+@ObjectType()
+class CopilotFailedToAddWorkspaceFileEmbeddingDataType {
+  @Field() message!: string
+}
+
+export class CopilotFailedToAddWorkspaceFileEmbedding extends UserFriendlyError {
+  constructor(args: CopilotFailedToAddWorkspaceFileEmbeddingDataType, message?: string | ((args: CopilotFailedToAddWorkspaceFileEmbeddingDataType) => string)) {
+    super('internal_server_error', 'copilot_failed_to_add_workspace_file_embedding', message, args);
+  }
+}
 
 export class BlobQuotaExceeded extends UserFriendlyError {
   constructor(message?: string) {
@@ -1048,6 +1058,7 @@ export enum ErrorNames {
   COPILOT_TRANSCRIPTION_JOB_EXISTS,
   COPILOT_TRANSCRIPTION_JOB_NOT_FOUND,
   COPILOT_TRANSCRIPTION_AUDIO_NOT_PROVIDED,
+  COPILOT_FAILED_TO_ADD_WORKSPACE_FILE_EMBEDDING,
   BLOB_QUOTA_EXCEEDED,
   STORAGE_QUOTA_EXCEEDED,
   MEMBER_QUOTA_EXCEEDED,
@@ -1078,5 +1089,5 @@ registerEnumType(ErrorNames, {
 export const ErrorDataUnionType = createUnionType({
   name: 'ErrorDataUnion',
   types: () =>
-    [GraphqlBadRequestDataType, HttpRequestErrorDataType, QueryTooLongDataType, ValidationErrorDataType, WrongSignInCredentialsDataType, UnknownOauthProviderDataType, InvalidOauthCallbackCodeDataType, MissingOauthQueryParameterDataType, InvalidEmailDataType, InvalidPasswordLengthDataType, WorkspacePermissionNotFoundDataType, SpaceNotFoundDataType, MemberNotFoundInSpaceDataType, NotInSpaceDataType, AlreadyInSpaceDataType, SpaceAccessDeniedDataType, SpaceOwnerNotFoundDataType, SpaceShouldHaveOnlyOneOwnerDataType, DocNotFoundDataType, DocActionDeniedDataType, DocUpdateBlockedDataType, VersionRejectedDataType, InvalidHistoryTimestampDataType, DocHistoryNotFoundDataType, BlobNotFoundDataType, ExpectToGrantDocUserRolesDataType, ExpectToRevokeDocUserRolesDataType, ExpectToUpdateDocUserRoleDataType, UnsupportedSubscriptionPlanDataType, SubscriptionAlreadyExistsDataType, SubscriptionNotExistsDataType, SameSubscriptionRecurringDataType, SubscriptionPlanNotFoundDataType, CopilotDocNotFoundDataType, CopilotMessageNotFoundDataType, CopilotPromptNotFoundDataType, CopilotProviderSideErrorDataType, CopilotInvalidContextDataType, CopilotContextFileNotSupportedDataType, CopilotFailedToModifyContextDataType, CopilotFailedToMatchContextDataType, RuntimeConfigNotFoundDataType, InvalidRuntimeConfigTypeDataType, InvalidLicenseUpdateParamsDataType, WorkspaceMembersExceedLimitToDowngradeDataType, UnsupportedClientVersionDataType, MentionUserDocAccessDeniedDataType] as const,
+    [GraphqlBadRequestDataType, HttpRequestErrorDataType, QueryTooLongDataType, ValidationErrorDataType, WrongSignInCredentialsDataType, UnknownOauthProviderDataType, InvalidOauthCallbackCodeDataType, MissingOauthQueryParameterDataType, InvalidEmailDataType, InvalidPasswordLengthDataType, WorkspacePermissionNotFoundDataType, SpaceNotFoundDataType, MemberNotFoundInSpaceDataType, NotInSpaceDataType, AlreadyInSpaceDataType, SpaceAccessDeniedDataType, SpaceOwnerNotFoundDataType, SpaceShouldHaveOnlyOneOwnerDataType, DocNotFoundDataType, DocActionDeniedDataType, DocUpdateBlockedDataType, VersionRejectedDataType, InvalidHistoryTimestampDataType, DocHistoryNotFoundDataType, BlobNotFoundDataType, ExpectToGrantDocUserRolesDataType, ExpectToRevokeDocUserRolesDataType, ExpectToUpdateDocUserRoleDataType, UnsupportedSubscriptionPlanDataType, SubscriptionAlreadyExistsDataType, SubscriptionNotExistsDataType, SameSubscriptionRecurringDataType, SubscriptionPlanNotFoundDataType, CopilotDocNotFoundDataType, CopilotMessageNotFoundDataType, CopilotPromptNotFoundDataType, CopilotProviderSideErrorDataType, CopilotInvalidContextDataType, CopilotContextFileNotSupportedDataType, CopilotFailedToModifyContextDataType, CopilotFailedToMatchContextDataType, CopilotFailedToAddWorkspaceFileEmbeddingDataType, RuntimeConfigNotFoundDataType, InvalidRuntimeConfigTypeDataType, InvalidLicenseUpdateParamsDataType, WorkspaceMembersExceedLimitToDowngradeDataType, UnsupportedClientVersionDataType, MentionUserDocAccessDeniedDataType] as const,
 });

packages/backend/server/src/models/common/copilot.ts

@@ -112,9 +112,10 @@ export const CopilotWorkspaceFileSchema = z.object({
   size: z.number(),
 });
 
-export type CopilotWorkspaceFile = z.infer<
+export type CopilotWorkspaceFileMetadata = z.infer<
   typeof CopilotWorkspaceFileSchema
-> & {
+>;
+export type CopilotWorkspaceFile = CopilotWorkspaceFileMetadata & {
   workspaceId: string;
   fileId: string;
   createdAt: Date;

packages/backend/server/src/models/copilot-workspace.ts

@@ -5,9 +5,10 @@ import { Transactional } from '@nestjs-cls/transactional';
 import { Prisma } from '@prisma/client';
 
 import { BaseModel } from './base';
-import {
+import type {
-  type CopilotWorkspaceFile,
+  CopilotWorkspaceFile,
-  type Embedding,
+  CopilotWorkspaceFileMetadata,
+  Embedding,
   FileChunkSimilarity,
 } from './common';
 
@@ -95,24 +96,40 @@ export class CopilotWorkspaceConfigModel extends BaseModel {
     return Prisma.join(groups.map(row => Prisma.sql`(${Prisma.join(row)})`));
   }
 
-  @Transactional()
+  async addFile(
-  async addWorkspaceFile(
     workspaceId: string,
-    file: Pick<CopilotWorkspaceFile, 'fileName' | 'mimeType' | 'size'>,
+    file: CopilotWorkspaceFileMetadata
-    embeddings: Embedding[]
+  ): Promise<CopilotWorkspaceFile> {
-  ): Promise<string> {
     const fileId = randomUUID();
-    await this.db.aiWorkspaceFiles.create({
+    const row = await this.db.aiWorkspaceFiles.create({
       data: { ...file, workspaceId, fileId },
     });
 
+    return row;
+  }
+
+  async getFile(workspaceId: string, fileId: string) {
+    const file = await this.db.aiWorkspaceFiles.findFirst({
+      where: {
+        workspaceId,
+        fileId,
+      },
+    });
+    return file;
+  }
+
+  @Transactional()
+  async addFileEmbeddings(
+    workspaceId: string,
+    fileId: string,
+    embeddings: Embedding[]
+  ) {
     const values = this.processEmbeddings(workspaceId, fileId, embeddings);
     await this.db.$executeRaw`
-        INSERT INTO "ai_workspace_file_embeddings"
+          INSERT INTO "ai_workspace_file_embeddings"
-        ("workspace_id", "file_id", "chunk", "content", "embedding") VALUES ${values}
+          ("workspace_id", "file_id", "chunk", "content", "embedding") VALUES ${values}
-        ON CONFLICT (workspace_id, file_id, chunk) DO NOTHING;
+          ON CONFLICT (workspace_id, file_id, chunk) DO NOTHING;
-    `;
+      `;
-    return fileId;
   }
 
   async listWorkspaceFiles(
@@ -152,5 +169,6 @@ export class CopilotWorkspaceConfigModel extends BaseModel {
         fileId,
       },
     });
+    return true;
   }
 }

packages/backend/server/src/plugins/copilot/context/job.ts

@@ -13,9 +13,9 @@ import {
 import { DocReader } from '../../../core/doc';
 import { Models } from '../../../models';
 import { CopilotStorage } from '../storage';
+import { readStream } from '../utils';
 import { OpenAIEmbeddingClient } from './embedding';
 import { EmbeddingClient } from './types';
-import { readStream } from './utils';
 
 @Injectable()
 export class CopilotContextDocJob {

packages/backend/server/src/plugins/copilot/context/resolver.ts

@@ -47,10 +47,10 @@ import {
 import { COPILOT_LOCKER, CopilotType } from '../resolver';
 import { ChatSessionService } from '../session';
 import { CopilotStorage } from '../storage';
+import { MAX_EMBEDDABLE_SIZE } from '../types';
+import { readStream } from '../utils';
 import { CopilotContextDocJob } from './job';
 import { CopilotContextService } from './service';
-import { MAX_EMBEDDABLE_SIZE } from './types';
-import { readStream } from './utils';
 
 @InputType()
 class AddContextCategoryInput {

packages/backend/server/src/plugins/copilot/context/types.ts

@@ -1,6 +1,6 @@
 import { File } from 'node:buffer';
 
-import { CopilotContextFileNotSupported, OneMB } from '../../../base';
+import { CopilotContextFileNotSupported } from '../../../base';
 import { Embedding } from '../../../models';
 import { parseDoc } from '../../../native';
 
@@ -46,8 +46,6 @@ declare global {
   }
 }
 
-export const MAX_EMBEDDABLE_SIZE = 50 * OneMB;
-
 export type Chunk = {
   index: number;
   content: string;

packages/backend/server/src/plugins/copilot/context/utils.ts

@@ -1,8 +1,3 @@
-import { Readable } from 'node:stream';
-
-import { readBufferWithLimit } from '../../../base';
-import { MAX_EMBEDDABLE_SIZE } from './types';
-
 export class GqlSignal implements AsyncDisposable {
   readonly abortController = new AbortController();
 
@@ -14,10 +9,3 @@ export class GqlSignal implements AsyncDisposable {
     this.abortController.abort();
   }
 }
-
-export function readStream(
-  readable: Readable,
-  maxSize = MAX_EMBEDDABLE_SIZE
-): Promise<Buffer> {
-  return readBufferWithLimit(readable, maxSize);
-}

packages/backend/server/src/plugins/copilot/index.ts

@@ -7,6 +7,7 @@ import { DocStorageModule } from '../../core/doc';
 import { FeatureModule } from '../../core/features';
 import { PermissionModule } from '../../core/permission';
 import { QuotaModule } from '../../core/quota';
+import { WorkspaceModule } from '../../core/workspaces';
 import {
   CopilotContextDocJob,
   CopilotContextResolver,
@@ -29,6 +30,11 @@ import {
   CopilotTranscriptionService,
 } from './transcript';
 import { CopilotWorkflowExecutors, CopilotWorkflowService } from './workflow';
+import {
+  CopilotWorkspaceEmbeddingConfigResolver,
+  CopilotWorkspaceEmbeddingResolver,
+  CopilotWorkspaceService,
+} from './workspace';
 
 @Module({
   imports: [
@@ -37,6 +43,7 @@ import { CopilotWorkflowExecutors, CopilotWorkflowService } from './workflow';
     QuotaModule,
     PermissionModule,
     ServerConfigModule,
+    WorkspaceModule,
   ],
   providers: [
     // providers
@@ -58,6 +65,10 @@ import { CopilotWorkflowExecutors, CopilotWorkflowService } from './workflow';
     // transcription
     CopilotTranscriptionService,
     CopilotTranscriptionResolver,
+    // workspace embeddings
+    CopilotWorkspaceService,
+    CopilotWorkspaceEmbeddingResolver,
+    CopilotWorkspaceEmbeddingConfigResolver,
     // gql resolvers
     UserCopilotResolver,
     PromptsManagementResolver,

packages/backend/server/src/plugins/copilot/types.ts

@@ -1,6 +1,7 @@
 import { type Tokenizer } from '@affine/server-native';
 import { z } from 'zod';
 
+import { OneMB } from '../../base';
 import { fromModelName } from '../../native';
 import type { ChatPrompt } from './prompt';
 import { PromptMessageSchema, PureMessageSchema } from './providers';
@@ -116,3 +117,5 @@ export type CopilotContextFile = {
   // embedding status
   status: 'in_progress' | 'completed' | 'failed';
 };
+
+export const MAX_EMBEDDABLE_SIZE = 50 * OneMB;

packages/backend/server/src/plugins/copilot/utils.ts

@@ -0,0 +1,11 @@
+import { Readable } from 'node:stream';
+
+import { readBufferWithLimit } from '../../base';
+import { MAX_EMBEDDABLE_SIZE } from './types';
+
+export function readStream(
+  readable: Readable,
+  maxSize = MAX_EMBEDDABLE_SIZE
+): Promise<Buffer> {
+  return readBufferWithLimit(readable, maxSize);
+}

packages/backend/server/src/plugins/copilot/workspace/index.ts

@@ -0,0 +1,5 @@
+export {
+  CopilotWorkspaceEmbeddingConfigResolver,
+  CopilotWorkspaceEmbeddingResolver,
+} from './resolver';
+export { CopilotWorkspaceService } from './service';

packages/backend/server/src/plugins/copilot/workspace/resolver.ts

@@ -0,0 +1,218 @@
+import {
+  Args,
+  Context,
+  Field,
+  Mutation,
+  ObjectType,
+  Parent,
+  ResolveField,
+  Resolver,
+} from '@nestjs/graphql';
+import type { Request } from 'express';
+import { SafeIntResolver } from 'graphql-scalars';
+import GraphQLUpload, {
+  type FileUpload,
+} from 'graphql-upload/GraphQLUpload.mjs';
+
+import {
+  BlobQuotaExceeded,
+  CopilotEmbeddingUnavailable,
+  CopilotFailedToAddWorkspaceFileEmbedding,
+  Mutex,
+  TooManyRequest,
+  UserFriendlyError,
+} from '../../../base';
+import { CurrentUser } from '../../../core/auth';
+import { AccessController } from '../../../core/permission';
+import { WorkspaceType } from '../../../core/workspaces';
+import { CopilotWorkspaceFile, Models } from '../../../models';
+import { COPILOT_LOCKER } from '../resolver';
+import { MAX_EMBEDDABLE_SIZE } from '../types';
+import { CopilotWorkspaceService } from './service';
+
+@ObjectType('CopilotWorkspaceConfig')
+export class CopilotWorkspaceConfigType {
+  @Field(() => String)
+  workspaceId!: string;
+}
+
+@ObjectType('CopilotWorkspaceFile')
+export class CopilotWorkspaceFileType implements CopilotWorkspaceFile {
+  @Field(() => String)
+  workspaceId!: string;
+
+  @Field(() => String)
+  fileId!: string;
+
+  @Field(() => String)
+  fileName!: string;
+
+  @Field(() => String)
+  mimeType!: string;
+
+  @Field(() => SafeIntResolver)
+  size!: number;
+
+  @Field(() => Date)
+  createdAt!: Date;
+}
+
+/**
+ * Workspace embedding config resolver
+ * Public apis rate limit: 10 req/m
+ * Other rate limit: 120 req/m
+ */
+@Resolver(() => WorkspaceType)
+export class CopilotWorkspaceEmbeddingResolver {
+  constructor(private readonly ac: AccessController) {}
+
+  @ResolveField(() => CopilotWorkspaceConfigType, {
+    complexity: 2,
+  })
+  async embedding(
+    @CurrentUser() user: CurrentUser,
+    @Parent() workspace: WorkspaceType
+  ): Promise<CopilotWorkspaceConfigType> {
+    await this.ac
+      .user(user.id)
+      .workspace(workspace.id)
+      .assert('Workspace.Read');
+
+    return { workspaceId: workspace.id };
+  }
+}
+
+@Resolver(() => CopilotWorkspaceConfigType)
+export class CopilotWorkspaceEmbeddingConfigResolver {
+  constructor(
+    private readonly ac: AccessController,
+    private readonly models: Models,
+    private readonly mutex: Mutex,
+    private readonly copilotWorkspace: CopilotWorkspaceService
+  ) {}
+
+  @ResolveField(() => [String], {
+    complexity: 2,
+  })
+  async ignoredDocs(
+    @Parent() config: CopilotWorkspaceConfigType
+  ): Promise<string[]> {
+    return this.models.copilotWorkspace.listIgnoredDocs(config.workspaceId);
+  }
+
+  @Mutation(() => Number, {
+    name: 'updateWorkspaceEmbeddingIgnoredDocs',
+    complexity: 2,
+    description: 'Update ignored docs',
+  })
+  async updateIgnoredDocs(
+    @CurrentUser() user: CurrentUser,
+    @Args('workspaceId', { type: () => String })
+    workspaceId: string,
+    @Args('add', { type: () => [String], nullable: true })
+    add?: string[],
+    @Args('remove', { type: () => [String], nullable: true })
+    remove?: string[]
+  ): Promise<number> {
+    await this.ac
+      .user(user.id)
+      .workspace(workspaceId)
+      .assert('Workspace.Settings.Update');
+    return await this.models.copilotWorkspace.updateIgnoredDocs(
+      workspaceId,
+      add,
+      remove
+    );
+  }
+
+  @ResolveField(() => [CopilotWorkspaceFileType], {
+    complexity: 2,
+  })
+  async files(
+    @Parent() config: CopilotWorkspaceConfigType
+  ): Promise<CopilotWorkspaceFileType[]> {
+    return this.models.copilotWorkspace.listWorkspaceFiles(config.workspaceId);
+  }
+
+  @Mutation(() => CopilotWorkspaceFileType, {
+    name: 'addWorkspaceEmbeddingFiles',
+    complexity: 2,
+    description: 'Update workspace embedding files',
+  })
+  async addFiles(
+    @Context() ctx: { req: Request },
+    @CurrentUser() user: CurrentUser,
+    @Args('workspaceId', { type: () => String })
+    workspaceId: string,
+    @Args({ name: 'blob', type: () => GraphQLUpload })
+    content: FileUpload
+  ): Promise<CopilotWorkspaceFileType> {
+    await this.ac
+      .user(user.id)
+      .workspace(workspaceId)
+      .assert('Workspace.Settings.Update');
+
+    if (!this.copilotWorkspace.canEmbedding) {
+      throw new CopilotEmbeddingUnavailable();
+    }
+
+    const lockFlag = `${COPILOT_LOCKER}:workspace:${workspaceId}`;
+    await using lock = await this.mutex.acquire(lockFlag);
+    if (!lock) {
+      throw new TooManyRequest('Server is busy');
+    }
+
+    const length = Number(ctx.req.headers['content-length']);
+    if (length && length >= MAX_EMBEDDABLE_SIZE) {
+      throw new BlobQuotaExceeded();
+    }
+
+    try {
+      const { blobId, file } = await this.copilotWorkspace.addWorkspaceFile(
+        user.id,
+        workspaceId,
+        content
+      );
+      await this.copilotWorkspace.addWorkspaceFileEmbeddingQueue({
+        userId: user.id,
+        workspaceId,
+        blobId,
+        fileId: file.fileId,
+        fileName: file.fileName,
+      });
+
+      return file;
+    } catch (e: any) {
+      // passthrough user friendly error
+      if (e instanceof UserFriendlyError) {
+        throw e;
+      }
+      throw new CopilotFailedToAddWorkspaceFileEmbedding({
+        message: e.message,
+      });
+    }
+  }
+
+  @Mutation(() => Boolean, {
+    name: 'removeWorkspaceEmbeddingFiles',
+    complexity: 2,
+    description: 'Remove workspace embedding files',
+  })
+  async removeFiles(
+    @CurrentUser() user: CurrentUser,
+    @Args('workspaceId', { type: () => String })
+    workspaceId: string,
+    @Args('fileId', { type: () => String })
+    fileId: string
+  ): Promise<boolean> {
+    await this.ac
+      .user(user.id)
+      .workspace(workspaceId)
+      .assert('Workspace.Settings.Update');
+
+    return await this.models.copilotWorkspace.removeWorkspaceFile(
+      workspaceId,
+      fileId
+    );
+  }
+}

packages/backend/server/src/plugins/copilot/workspace/service.ts

@@ -0,0 +1,87 @@
+import { createHash } from 'node:crypto';
+
+import { Injectable, OnApplicationBootstrap } from '@nestjs/common';
+
+import { FileUpload, JobQueue } from '../../../base';
+import { Models } from '../../../models';
+import { CopilotStorage } from '../storage';
+import { readStream } from '../utils';
+
+declare global {
+  interface Events {
+    'workspace.file.embedding.finished': {
+      jobId: string;
+    };
+    'workspace.file.embedding.failed': {
+      jobId: string;
+    };
+  }
+  interface Jobs {
+    'copilot.workspace.embedding.files': {
+      userId: string;
+      workspaceId: string;
+      blobId: string;
+      fileId: string;
+      fileName: string;
+    };
+  }
+}
+
+@Injectable()
+export class CopilotWorkspaceService implements OnApplicationBootstrap {
+  private supportEmbedding = false;
+
+  constructor(
+    private readonly models: Models,
+    private readonly queue: JobQueue,
+    private readonly storage: CopilotStorage
+  ) {}
+
+  async onApplicationBootstrap() {
+    const supportEmbedding =
+      await this.models.copilotContext.checkEmbeddingAvailable();
+    if (supportEmbedding) {
+      this.supportEmbedding = true;
+    }
+  }
+
+  get canEmbedding() {
+    return this.supportEmbedding;
+  }
+
+  async addWorkspaceFile(
+    userId: string,
+    workspaceId: string,
+    content: FileUpload
+  ) {
+    const fileName = content.filename;
+    const buffer = await readStream(content.createReadStream());
+    const blobId = createHash('sha256').update(buffer).digest('base64url');
+    await this.storage.put(userId, workspaceId, blobId, buffer);
+    const file = await this.models.copilotWorkspace.addFile(workspaceId, {
+      fileName,
+      mimeType: content.mimetype,
+      size: buffer.length,
+    });
+    return { blobId, file };
+  }
+
+  async getWorkspaceFile(workspaceId: string, fileId: string) {
+    return await this.models.copilotWorkspace.getFile(workspaceId, fileId);
+  }
+
+  async addWorkspaceFileEmbeddingQueue(
+    file: Jobs['copilot.workspace.embedding.files']
+  ) {
+    if (!this.supportEmbedding) return;
+
+    const { userId, workspaceId, blobId, fileId, fileName } = file;
+    await this.queue.add('copilot.workspace.embedding.files', {
+      userId,
+      workspaceId,
+      blobId,
+      fileId,
+      fileName,
+    });
+  }
+}

packages/backend/server/src/schema.gql

@@ -159,6 +159,10 @@ type CopilotDocType {
   status: ContextEmbedStatus
 }
 
+type CopilotFailedToAddWorkspaceFileEmbeddingDataType {
+  message: String!
+}
+
 type CopilotFailedToMatchContextDataType {
   content: String!
   contextId: String!
@@ -268,6 +272,21 @@ type CopilotSessionType {
   promptName: String!
 }
 
+type CopilotWorkspaceConfig {
+  files: [CopilotWorkspaceFile!]!
+  ignoredDocs: [String!]!
+  workspaceId: String!
+}
+
+type CopilotWorkspaceFile {
+  createdAt: DateTime!
+  fileId: String!
+  fileName: String!
+  mimeType: String!
+  size: SafeInt!
+  workspaceId: String!
+}
+
 input CreateChatMessageInput {
   attachments: [String!]
   blobs: [Upload!]
@@ -404,7 +423,7 @@ type EditorType {
   name: String!
 }
 
-union ErrorDataUnion = AlreadyInSpaceDataType | BlobNotFoundDataType | CopilotContextFileNotSupportedDataType | CopilotDocNotFoundDataType | CopilotFailedToMatchContextDataType | CopilotFailedToModifyContextDataType | CopilotInvalidContextDataType | CopilotMessageNotFoundDataType | CopilotPromptNotFoundDataType | CopilotProviderSideErrorDataType | DocActionDeniedDataType | DocHistoryNotFoundDataType | DocNotFoundDataType | DocUpdateBlockedDataType | ExpectToGrantDocUserRolesDataType | ExpectToRevokeDocUserRolesDataType | ExpectToUpdateDocUserRoleDataType | GraphqlBadRequestDataType | HttpRequestErrorDataType | InvalidEmailDataType | InvalidHistoryTimestampDataType | InvalidLicenseUpdateParamsDataType | InvalidOauthCallbackCodeDataType | InvalidPasswordLengthDataType | InvalidRuntimeConfigTypeDataType | MemberNotFoundInSpaceDataType | MentionUserDocAccessDeniedDataType | MissingOauthQueryParameterDataType | NotInSpaceDataType | QueryTooLongDataType | RuntimeConfigNotFoundDataType | SameSubscriptionRecurringDataType | SpaceAccessDeniedDataType | SpaceNotFoundDataType | SpaceOwnerNotFoundDataType | SpaceShouldHaveOnlyOneOwnerDataType | SubscriptionAlreadyExistsDataType | SubscriptionNotExistsDataType | SubscriptionPlanNotFoundDataType | UnknownOauthProviderDataType | UnsupportedClientVersionDataType | UnsupportedSubscriptionPlanDataType | ValidationErrorDataType | VersionRejectedDataType | WorkspaceMembersExceedLimitToDowngradeDataType | WorkspacePermissionNotFoundDataType | WrongSignInCredentialsDataType
+union ErrorDataUnion = AlreadyInSpaceDataType | BlobNotFoundDataType | CopilotContextFileNotSupportedDataType | CopilotDocNotFoundDataType | CopilotFailedToAddWorkspaceFileEmbeddingDataType | CopilotFailedToMatchContextDataType | CopilotFailedToModifyContextDataType | CopilotInvalidContextDataType | CopilotMessageNotFoundDataType | CopilotPromptNotFoundDataType | CopilotProviderSideErrorDataType | DocActionDeniedDataType | DocHistoryNotFoundDataType | DocNotFoundDataType | DocUpdateBlockedDataType | ExpectToGrantDocUserRolesDataType | ExpectToRevokeDocUserRolesDataType | ExpectToUpdateDocUserRoleDataType | GraphqlBadRequestDataType | HttpRequestErrorDataType | InvalidEmailDataType | InvalidHistoryTimestampDataType | InvalidLicenseUpdateParamsDataType | InvalidOauthCallbackCodeDataType | InvalidPasswordLengthDataType | InvalidRuntimeConfigTypeDataType | MemberNotFoundInSpaceDataType | MentionUserDocAccessDeniedDataType | MissingOauthQueryParameterDataType | NotInSpaceDataType | QueryTooLongDataType | RuntimeConfigNotFoundDataType | SameSubscriptionRecurringDataType | SpaceAccessDeniedDataType | SpaceNotFoundDataType | SpaceOwnerNotFoundDataType | SpaceShouldHaveOnlyOneOwnerDataType | SubscriptionAlreadyExistsDataType | SubscriptionNotExistsDataType | SubscriptionPlanNotFoundDataType | UnknownOauthProviderDataType | UnsupportedClientVersionDataType | UnsupportedSubscriptionPlanDataType | ValidationErrorDataType | VersionRejectedDataType | WorkspaceMembersExceedLimitToDowngradeDataType | WorkspacePermissionNotFoundDataType | WrongSignInCredentialsDataType
 
 enum ErrorNames {
   ACCESS_DENIED
@@ -427,6 +446,7 @@ enum ErrorNames {
   COPILOT_DOC_NOT_FOUND
   COPILOT_EMBEDDING_DISABLED
   COPILOT_EMBEDDING_UNAVAILABLE
+  COPILOT_FAILED_TO_ADD_WORKSPACE_FILE_EMBEDDING
   COPILOT_FAILED_TO_CREATE_MESSAGE
   COPILOT_FAILED_TO_GENERATE_TEXT
   COPILOT_FAILED_TO_MATCH_CONTEXT
@@ -907,6 +927,9 @@ type Mutation {
 
   """add a file to context"""
   addContextFile(content: Upload!, options: AddContextFileInput!): CopilotContextFile!
+
+  """Update workspace embedding files"""
+  addWorkspaceEmbeddingFiles(blob: Upload!, workspaceId: String!): CopilotWorkspaceFile!
   addWorkspaceFeature(feature: FeatureType!, workspaceId: String!): Boolean!
   approveMember(userId: String!, workspaceId: String!): Boolean!
 
@@ -995,6 +1018,9 @@ type Mutation {
 
   """remove a file from context"""
   removeContextFile(options: RemoveContextFileInput!): Boolean!
+
+  """Remove workspace embedding files"""
+  removeWorkspaceEmbeddingFiles(fileId: String!, workspaceId: String!): Boolean!
   removeWorkspaceFeature(feature: FeatureType!, workspaceId: String!): Boolean!
   resumeSubscription(idempotencyKey: String @deprecated(reason: "use header `Idempotency-Key`"), plan: SubscriptionPlan = Pro, workspaceId: String): SubscriptionType!
   retryAudioTranscription(jobId: String!, workspaceId: String!): TranscriptionResultType
@@ -1037,6 +1063,9 @@ type Mutation {
   """Update workspace"""
   updateWorkspace(input: UpdateWorkspaceInput!): WorkspaceType!
 
+  """Update ignored docs"""
+  updateWorkspaceEmbeddingIgnoredDocs(add: [String!], remove: [String!], workspaceId: String!): Int!
+
   """Upload user avatar"""
   uploadAvatar(avatar: Upload!): UserType!
   verifyEmail(token: String!): Boolean!
@@ -1703,6 +1732,7 @@ type WorkspaceType {
 
   """Get get with given id"""
   doc(docId: String!): DocType!
+  embedding: CopilotWorkspaceConfig!
 
   """Enable AI"""
   enableAi: Boolean!

packages/common/graphql/src/graphql/copilot-workspace-file-add.gql

@@ -0,0 +1,9 @@
+mutation addWorkspaceEmbeddingFiles($workspaceId: String!, $blob: Upload!) {
+  addWorkspaceEmbeddingFiles(workspaceId: $workspaceId, blob: $blob) {
+    fileId
+    fileName
+    mimeType
+    size
+    createdAt
+  }
+}

packages/common/graphql/src/graphql/copilot-workspace-file-remove.gql

@@ -0,0 +1,3 @@
+mutation removeWorkspaceEmbeddingFiles($workspaceId: String!, $fileId: String!) {
+  removeWorkspaceEmbeddingFiles(workspaceId: $workspaceId, fileId: $fileId)
+}

packages/common/graphql/src/graphql/copilot-workspace-get.gql

@@ -0,0 +1,14 @@
+query getWorkspaceEmbeddingConfig($workspaceId: String!) {
+  workspace(id: $workspaceId) {
+    embedding {
+      files {
+        fileId
+        fileName
+        mimeType
+        size
+        createdAt
+      }
+      ignoredDocs
+    }
+  }
+}

packages/common/graphql/src/graphql/copilot-workspace-ignored-docs.gql

@@ -0,0 +1,3 @@
+mutation updateWorkspaceEmbeddingIgnoredDocs($workspaceId: String!, $add: [String!], $remove: [String!]) {
+  updateWorkspaceEmbeddingIgnoredDocs(workspaceId: $workspaceId, add: $add, remove: $remove)
+}

packages/common/graphql/src/graphql/index.ts

@@ -741,6 +741,60 @@ export const getCopilotSessionsQuery = {
 }`,
 };
 
+export const addWorkspaceEmbeddingFilesMutation = {
+  id: 'addWorkspaceEmbeddingFilesMutation' as const,
+  op: 'addWorkspaceEmbeddingFiles',
+  query: `mutation addWorkspaceEmbeddingFiles($workspaceId: String!, $blob: Upload!) {
+  addWorkspaceEmbeddingFiles(workspaceId: $workspaceId, blob: $blob) {
+    fileId
+    fileName
+    mimeType
+    size
+    createdAt
+  }
+}`,
+  file: true,
+};
+
+export const removeWorkspaceEmbeddingFilesMutation = {
+  id: 'removeWorkspaceEmbeddingFilesMutation' as const,
+  op: 'removeWorkspaceEmbeddingFiles',
+  query: `mutation removeWorkspaceEmbeddingFiles($workspaceId: String!, $fileId: String!) {
+  removeWorkspaceEmbeddingFiles(workspaceId: $workspaceId, fileId: $fileId)
+}`,
+};
+
+export const getWorkspaceEmbeddingConfigQuery = {
+  id: 'getWorkspaceEmbeddingConfigQuery' as const,
+  op: 'getWorkspaceEmbeddingConfig',
+  query: `query getWorkspaceEmbeddingConfig($workspaceId: String!) {
+  workspace(id: $workspaceId) {
+    embedding {
+      files {
+        fileId
+        fileName
+        mimeType
+        size
+        createdAt
+      }
+      ignoredDocs
+    }
+  }
+}`,
+};
+
+export const updateWorkspaceEmbeddingIgnoredDocsMutation = {
+  id: 'updateWorkspaceEmbeddingIgnoredDocsMutation' as const,
+  op: 'updateWorkspaceEmbeddingIgnoredDocs',
+  query: `mutation updateWorkspaceEmbeddingIgnoredDocs($workspaceId: String!, $add: [String!], $remove: [String!]) {
+  updateWorkspaceEmbeddingIgnoredDocs(
+    workspaceId: $workspaceId
+    add: $add
+    remove: $remove
+  )
+}`,
+};
+
 export const createCheckoutSessionMutation = {
   id: 'createCheckoutSessionMutation' as const,
   op: 'createCheckoutSession',

packages/common/graphql/src/schema.ts

@@ -239,6 +239,11 @@ export interface CopilotDocType {
   status: Maybe<ContextEmbedStatus>;
 }
 
+export interface CopilotFailedToAddWorkspaceFileEmbeddingDataType {
+  __typename?: 'CopilotFailedToAddWorkspaceFileEmbeddingDataType';
+  message: Scalars['String']['output'];
+}
+
 export interface CopilotFailedToMatchContextDataType {
   __typename?: 'CopilotFailedToMatchContextDataType';
   content: Scalars['String']['output'];
@@ -359,6 +364,23 @@ export interface CopilotSessionType {
   promptName: Scalars['String']['output'];
 }
 
+export interface CopilotWorkspaceConfig {
+  __typename?: 'CopilotWorkspaceConfig';
+  files: Array<CopilotWorkspaceFile>;
+  ignoredDocs: Array<Scalars['String']['output']>;
+  workspaceId: Scalars['String']['output'];
+}
+
+export interface CopilotWorkspaceFile {
+  __typename?: 'CopilotWorkspaceFile';
+  createdAt: Scalars['DateTime']['output'];
+  fileId: Scalars['String']['output'];
+  fileName: Scalars['String']['output'];
+  mimeType: Scalars['String']['output'];
+  size: Scalars['SafeInt']['output'];
+  workspaceId: Scalars['String']['output'];
+}
+
 export interface CreateChatMessageInput {
   attachments?: InputMaybe<Array<Scalars['String']['input']>>;
   blobs?: InputMaybe<Array<Scalars['Upload']['input']>>;
@@ -507,6 +529,7 @@ export type ErrorDataUnion =
   | BlobNotFoundDataType
   | CopilotContextFileNotSupportedDataType
   | CopilotDocNotFoundDataType
+  | CopilotFailedToAddWorkspaceFileEmbeddingDataType
   | CopilotFailedToMatchContextDataType
   | CopilotFailedToModifyContextDataType
   | CopilotInvalidContextDataType
@@ -572,6 +595,7 @@ export enum ErrorNames {
   COPILOT_DOC_NOT_FOUND = 'COPILOT_DOC_NOT_FOUND',
   COPILOT_EMBEDDING_DISABLED = 'COPILOT_EMBEDDING_DISABLED',
   COPILOT_EMBEDDING_UNAVAILABLE = 'COPILOT_EMBEDDING_UNAVAILABLE',
+  COPILOT_FAILED_TO_ADD_WORKSPACE_FILE_EMBEDDING = 'COPILOT_FAILED_TO_ADD_WORKSPACE_FILE_EMBEDDING',
   COPILOT_FAILED_TO_CREATE_MESSAGE = 'COPILOT_FAILED_TO_CREATE_MESSAGE',
   COPILOT_FAILED_TO_GENERATE_TEXT = 'COPILOT_FAILED_TO_GENERATE_TEXT',
   COPILOT_FAILED_TO_MATCH_CONTEXT = 'COPILOT_FAILED_TO_MATCH_CONTEXT',
@@ -1040,6 +1064,8 @@ export interface Mutation {
   addContextDoc: CopilotContextDoc;
   /** add a file to context */
   addContextFile: CopilotContextFile;
+  /** Update workspace embedding files */
+  addWorkspaceEmbeddingFiles: CopilotWorkspaceFile;
   addWorkspaceFeature: Scalars['Boolean']['output'];
   approveMember: Scalars['Boolean']['output'];
   /** Ban an user */
@@ -1107,6 +1133,8 @@ export interface Mutation {
   removeContextDoc: Scalars['Boolean']['output'];
   /** remove a file from context */
   removeContextFile: Scalars['Boolean']['output'];
+  /** Remove workspace embedding files */
+  removeWorkspaceEmbeddingFiles: Scalars['Boolean']['output'];
   removeWorkspaceFeature: Scalars['Boolean']['output'];
   resumeSubscription: SubscriptionType;
   retryAudioTranscription: Maybe<TranscriptionResultType>;
@@ -1142,6 +1170,8 @@ export interface Mutation {
   updateUserFeatures: Array<FeatureType>;
   /** Update workspace */
   updateWorkspace: WorkspaceType;
+  /** Update ignored docs */
+  updateWorkspaceEmbeddingIgnoredDocs: Scalars['Int']['output'];
   /** Upload user avatar */
   uploadAvatar: UserType;
   verifyEmail: Scalars['Boolean']['output'];
@@ -1171,6 +1201,11 @@ export interface MutationAddContextFileArgs {
   options: AddContextFileInput;
 }
 
+export interface MutationAddWorkspaceEmbeddingFilesArgs {
+  blob: Scalars['Upload']['input'];
+  workspaceId: Scalars['String']['input'];
+}
+
 export interface MutationAddWorkspaceFeatureArgs {
   feature: FeatureType;
   workspaceId: Scalars['String']['input'];
@@ -1364,6 +1399,11 @@ export interface MutationRemoveContextFileArgs {
   options: RemoveContextFileInput;
 }
 
+export interface MutationRemoveWorkspaceEmbeddingFilesArgs {
+  fileId: Scalars['String']['input'];
+  workspaceId: Scalars['String']['input'];
+}
+
 export interface MutationRemoveWorkspaceFeatureArgs {
   feature: FeatureType;
   workspaceId: Scalars['String']['input'];
@@ -1494,6 +1534,12 @@ export interface MutationUpdateWorkspaceArgs {
   input: UpdateWorkspaceInput;
 }
 
+export interface MutationUpdateWorkspaceEmbeddingIgnoredDocsArgs {
+  add?: InputMaybe<Array<Scalars['String']['input']>>;
+  remove?: InputMaybe<Array<Scalars['String']['input']>>;
+  workspaceId: Scalars['String']['input'];
+}
+
 export interface MutationUploadAvatarArgs {
   avatar: Scalars['Upload']['input'];
 }
@@ -2232,6 +2278,7 @@ export interface WorkspaceType {
   createdAt: Scalars['DateTime']['output'];
   /** Get get with given id */
   doc: DocType;
+  embedding: CopilotWorkspaceConfig;
   /** Enable AI */
   enableAi: Scalars['Boolean']['output'];
   /** Enable doc embedding */
@@ -3169,6 +3216,71 @@ export type GetCopilotSessionsQuery = {
   } | null;
 };
 
+export type AddWorkspaceEmbeddingFilesMutationVariables = Exact<{
+  workspaceId: Scalars['String']['input'];
+  blob: Scalars['Upload']['input'];
+}>;
+
+export type AddWorkspaceEmbeddingFilesMutation = {
+  __typename?: 'Mutation';
+  addWorkspaceEmbeddingFiles: {
+    __typename?: 'CopilotWorkspaceFile';
+    fileId: string;
+    fileName: string;
+    mimeType: string;
+    size: number;
+    createdAt: string;
+  };
+};
+
+export type RemoveWorkspaceEmbeddingFilesMutationVariables = Exact<{
+  workspaceId: Scalars['String']['input'];
+  fileId: Scalars['String']['input'];
+}>;
+
+export type RemoveWorkspaceEmbeddingFilesMutation = {
+  __typename?: 'Mutation';
+  removeWorkspaceEmbeddingFiles: boolean;
+};
+
+export type GetWorkspaceEmbeddingConfigQueryVariables = Exact<{
+  workspaceId: Scalars['String']['input'];
+}>;
+
+export type GetWorkspaceEmbeddingConfigQuery = {
+  __typename?: 'Query';
+  workspace: {
+    __typename?: 'WorkspaceType';
+    embedding: {
+      __typename?: 'CopilotWorkspaceConfig';
+      ignoredDocs: Array<string>;
+      files: Array<{
+        __typename?: 'CopilotWorkspaceFile';
+        fileId: string;
+        fileName: string;
+        mimeType: string;
+        size: number;
+        createdAt: string;
+      }>;
+    };
+  };
+};
+
+export type UpdateWorkspaceEmbeddingIgnoredDocsMutationVariables = Exact<{
+  workspaceId: Scalars['String']['input'];
+  add?: InputMaybe<
+    Array<Scalars['String']['input']> | Scalars['String']['input']
+  >;
+  remove?: InputMaybe<
+    Array<Scalars['String']['input']> | Scalars['String']['input']
+  >;
+}>;
+
+export type UpdateWorkspaceEmbeddingIgnoredDocsMutation = {
+  __typename?: 'Mutation';
+  updateWorkspaceEmbeddingIgnoredDocs: number;
+};
+
 export type CreateCheckoutSessionMutationVariables = Exact<{
   input: CreateCheckoutSessionInput;
 }>;
@@ -4427,6 +4539,11 @@ export type Queries =
       variables: GetCopilotSessionsQueryVariables;
       response: GetCopilotSessionsQuery;
     }
+  | {
+      name: 'getWorkspaceEmbeddingConfigQuery';
+      variables: GetWorkspaceEmbeddingConfigQueryVariables;
+      response: GetWorkspaceEmbeddingConfigQuery;
+    }
   | {
       name: 'getDocRolePermissionsQuery';
       variables: GetDocRolePermissionsQueryVariables;
@@ -4784,6 +4901,21 @@ export type Mutations =
       variables: UpdateCopilotSessionMutationVariables;
       response: UpdateCopilotSessionMutation;
     }
+  | {
+      name: 'addWorkspaceEmbeddingFilesMutation';
+      variables: AddWorkspaceEmbeddingFilesMutationVariables;
+      response: AddWorkspaceEmbeddingFilesMutation;
+    }
+  | {
+      name: 'removeWorkspaceEmbeddingFilesMutation';
+      variables: RemoveWorkspaceEmbeddingFilesMutationVariables;
+      response: RemoveWorkspaceEmbeddingFilesMutation;
+    }
+  | {
+      name: 'updateWorkspaceEmbeddingIgnoredDocsMutation';
+      variables: UpdateWorkspaceEmbeddingIgnoredDocsMutationVariables;
+      response: UpdateWorkspaceEmbeddingIgnoredDocsMutation;
+    }
   | {
       name: 'createCheckoutSessionMutation';
       variables: CreateCheckoutSessionMutationVariables;

packages/frontend/i18n/src/i18n.gen.ts

@@ -8193,6 +8193,12 @@ export function useAFFiNEI18N(): {
       * `Audio not provided.`
       */
     ["error.COPILOT_TRANSCRIPTION_AUDIO_NOT_PROVIDED"](): string;
+    /**
+      * `Failed to add workspace file embedding: {{message}}`
+      */
+    ["error.COPILOT_FAILED_TO_ADD_WORKSPACE_FILE_EMBEDDING"](options: {
+        readonly message: string;
+    }): string;
     /**
       * `You have exceeded your blob size quota.`
       */

packages/frontend/i18n/src/resources/en.json

@@ -2027,6 +2027,7 @@
   "error.COPILOT_TRANSCRIPTION_JOB_EXISTS": "Transcription job already exists",
   "error.COPILOT_TRANSCRIPTION_JOB_NOT_FOUND": "Transcription job not found.",
   "error.COPILOT_TRANSCRIPTION_AUDIO_NOT_PROVIDED": "Audio not provided.",
+  "error.COPILOT_FAILED_TO_ADD_WORKSPACE_FILE_EMBEDDING": "Failed to add workspace file embedding: {{message}}",
   "error.BLOB_QUOTA_EXCEEDED": "You have exceeded your blob size quota.",
   "error.STORAGE_QUOTA_EXCEEDED": "You have exceeded your storage quota.",
   "error.MEMBER_QUOTA_EXCEEDED": "You have exceeded your workspace member quota.",