feat(server): search docs by keywork from indexer (#12863)

#### PR Dependency Tree


* **PR #12867**
  * **PR #12863** 👈
    * **PR #12837**
    * **PR #12866**

This tree was auto-generated by
[Charcoal](https://github.com/danerwilliams/charcoal)

packages/backend/server/src/models/user.ts

@@ -45,6 +45,10 @@ interface UserFilter {
   withDisabled?: boolean;
 }
 
+export interface ItemWithUserId {
+  userId: string;
+}
+
 export type PublicUser = Pick<User, keyof typeof publicUserSelect>;
 export type WorkspaceUser = Pick<User, keyof typeof workspaceUserSelect>;
 export type { ConnectedAccount, User };
@@ -78,6 +82,19 @@ export class UserModel extends BaseModel {
     });
   }
 
+  async getPublicUsersMap<T extends ItemWithUserId>(
+    items: T[]
+  ): Promise<Map<string, PublicUser>> {
+    const userIds: string[] = [];
+    for (const item of items) {
+      if (item.userId) {
+        userIds.push(item.userId);
+      }
+    }
+    const users = await this.getPublicUsers(userIds);
+    return new Map(users.map(user => [user.id, user]));
+  }
+
   async getWorkspaceUser(id: string): Promise<WorkspaceUser | null> {
     return this.db.user.findUnique({
       select: workspaceUserSelect,

packages/backend/server/src/plugins/indexer/__tests__/__snapshots__/service.spec.ts.md

@@ -521,3 +521,38 @@ Generated by [AVA](https://avajs.dev).
         'blob3 name.docx',
       ],
     ]
+
+## should search docs by keyword work
+
+> Snapshot 1
+
+    [
+      {
+        blockId: 'block1',
+        createdAt: Date 2025-06-20 00:00:00 UTC {},
+        highlight: '<b>hello</b> world',
+        title: 'hello world',
+        updatedAt: Date 2025-06-20 00:00:00 UTC {},
+      },
+      {
+        blockId: 'block2',
+        createdAt: Date 2025-06-20 00:00:01 UTC {},
+        highlight: '<b>hello</b> world 2',
+        title: 'hello world 2',
+        updatedAt: Date 2025-06-20 00:00:01 UTC {},
+      },
+      {
+        blockId: 'block3',
+        createdAt: Date 2025-06-20 00:00:02 UTC {},
+        highlight: '<b>hello</b> world 3',
+        title: 'hello world 3',
+        updatedAt: Date 2025-06-20 00:00:02 UTC {},
+      },
+      {
+        blockId: 'block4',
+        createdAt: Date 2025-06-20 00:00:03 UTC {},
+        highlight: '<b>hello</b> world 4',
+        title: '',
+        updatedAt: Date 2025-06-20 00:00:03 UTC {},
+      },
+    ]

packages/backend/server/src/plugins/indexer/__tests__/service.spec.ts

@@ -2213,3 +2213,101 @@ test('should search blob names work', async t => {
 });
 
 // #endregion
+
+// #region searchDocsByKeyword()
+
+test('should search docs by keyword work', async t => {
+  const workspaceId = workspace.id;
+  const docId1 = randomUUID();
+  const docId2 = randomUUID();
+  const docId3 = randomUUID();
+  const docId4 = randomUUID();
+
+  await module.create(Mockers.DocMeta, {
+    workspaceId,
+    docId: docId1,
+    title: 'hello world 1',
+  });
+  await module.create(Mockers.DocMeta, {
+    workspaceId,
+    docId: docId2,
+    title: 'hello world 2',
+  });
+  await module.create(Mockers.DocMeta, {
+    workspaceId,
+    docId: docId3,
+    title: 'hello world 3',
+  });
+
+  await indexerService.write(
+    SearchTable.block,
+    [
+      {
+        workspaceId,
+        docId: docId1,
+        blockId: 'block1',
+        content: 'hello world',
+        flavour: 'affine:page',
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date('2025-06-20T00:00:00.000Z'),
+        updatedAt: new Date('2025-06-20T00:00:00.000Z'),
+      },
+      {
+        workspaceId,
+        docId: docId2,
+        blockId: 'block2',
+        content: 'hello world 2',
+        flavour: 'affine:text',
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date('2025-06-20T00:00:01.000Z'),
+        updatedAt: new Date('2025-06-20T00:00:01.000Z'),
+      },
+      {
+        workspaceId,
+        docId: docId3,
+        blockId: 'block3',
+        content: 'hello world 3',
+        flavour: 'affine:text',
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date('2025-06-20T00:00:02.000Z'),
+        updatedAt: new Date('2025-06-20T00:00:02.000Z'),
+      },
+      {
+        workspaceId,
+        docId: docId4,
+        blockId: 'block4',
+        content: 'hello world 4',
+        flavour: 'affine:text',
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date('2025-06-20T00:00:03.000Z'),
+        updatedAt: new Date('2025-06-20T00:00:03.000Z'),
+      },
+    ],
+    {
+      refresh: true,
+    }
+  );
+
+  const rows = await indexerService.searchDocsByKeyword(workspaceId, 'hello');
+
+  t.is(rows.length, 4);
+  t.snapshot(
+    rows
+      .map(row =>
+        omit(row, [
+          'docId',
+          'createdByUserId',
+          'updatedByUserId',
+          'createdByUser',
+          'updatedByUser',
+        ])
+      )
+      .sort((a, b) => a.blockId.localeCompare(b.blockId))
+  );
+});
+
+// #endregion

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

@@ -26,6 +26,7 @@ import { IndexerService } from './service';
 export class IndexerModule {}
 
 export { IndexerService };
+export type { SearchDoc } from './types';
 
 declare global {
   interface Events {

packages/backend/server/src/plugins/indexer/service.ts

@@ -33,6 +33,7 @@ import {
 } from './tables';
 import {
   AggregateInput,
+  SearchDoc,
   SearchHighlight,
   SearchInput,
   SearchQuery,
@@ -433,6 +434,155 @@ export class IndexerService {
     return blobNameMap;
   }
 
+  async searchDocsByKeyword(
+    workspaceId: string,
+    keyword: string,
+    options?: {
+      limit?: number;
+    }
+  ): Promise<SearchDoc[]> {
+    const limit = options?.limit ?? 20;
+    const result = await this.aggregate({
+      table: SearchTable.block,
+      field: 'docId',
+      query: {
+        type: SearchQueryType.boolean,
+        occur: SearchQueryOccur.must,
+        queries: [
+          {
+            type: SearchQueryType.match,
+            field: 'workspaceId',
+            match: workspaceId,
+          },
+          {
+            type: SearchQueryType.boolean,
+            occur: SearchQueryOccur.must,
+            queries: [
+              {
+                type: SearchQueryType.match,
+                field: 'content',
+                match: keyword,
+              },
+              {
+                type: SearchQueryType.boolean,
+                occur: SearchQueryOccur.should,
+                queries: [
+                  {
+                    type: SearchQueryType.match,
+                    field: 'content',
+                    match: keyword,
+                  },
+                  {
+                    type: SearchQueryType.boost,
+                    boost: 1.5,
+                    query: {
+                      type: SearchQueryType.match,
+                      field: 'flavour',
+                      match: 'affine:page',
+                    },
+                  },
+                ],
+              },
+            ],
+          },
+        ],
+      },
+      options: {
+        hits: {
+          fields: [
+            'blockId',
+            'flavour',
+            'content',
+            'createdAt',
+            'updatedAt',
+            'createdByUserId',
+            'updatedByUserId',
+          ],
+          highlights: [
+            {
+              field: 'content',
+              before: '<b>',
+              end: '</b>',
+            },
+          ],
+          pagination: {
+            limit: 2,
+          },
+        },
+        pagination: {
+          limit,
+        },
+      },
+    });
+
+    const docs: SearchDoc[] = [];
+    const missingTitles: { workspaceId: string; docId: string }[] = [];
+    const userIds: { userId: string }[] = [];
+
+    for (const bucket of result.buckets) {
+      const docId = bucket.key;
+      const blockId = bucket.hits.nodes[0].fields.blockId[0] as string;
+      const flavour = bucket.hits.nodes[0].fields.flavour[0] as string;
+      const content = bucket.hits.nodes[0].fields.content[0] as string;
+      const createdAt = bucket.hits.nodes[0].fields.createdAt[0] as Date;
+      const updatedAt = bucket.hits.nodes[0].fields.updatedAt[0] as Date;
+      const createdByUserId = bucket.hits.nodes[0].fields
+        .createdByUserId[0] as string;
+      const updatedByUserId = bucket.hits.nodes[0].fields
+        .updatedByUserId[0] as string;
+      const highlight = bucket.hits.nodes[0].highlights?.content?.[0] as string;
+      let title = '';
+
+      // hit title block
+      if (flavour === 'affine:page') {
+        title = content;
+      } else {
+        // hit content block, missing title
+        missingTitles.push({ workspaceId, docId });
+      }
+
+      docs.push({
+        docId,
+        blockId,
+        title,
+        highlight,
+        createdAt,
+        updatedAt,
+        createdByUserId,
+        updatedByUserId,
+      });
+      userIds.push({ userId: createdByUserId }, { userId: updatedByUserId });
+    }
+
+    if (missingTitles.length > 0) {
+      const metas = await this.models.doc.findMetas(missingTitles, {
+        select: {
+          title: true,
+        },
+      });
+      const titleMap = new Map<string, string>();
+      for (const meta of metas) {
+        if (meta?.title) {
+          titleMap.set(meta.docId, meta.title);
+        }
+      }
+      for (const doc of docs) {
+        if (!doc.title) {
+          doc.title = titleMap.get(doc.docId) ?? '';
+        }
+      }
+    }
+
+    const userMap = await this.models.user.getPublicUsersMap(userIds);
+
+    for (const doc of docs) {
+      doc.createdByUser = userMap.get(doc.createdByUserId);
+      doc.updatedByUser = userMap.get(doc.updatedByUserId);
+    }
+
+    return docs;
+  }
+
   #formatSearchNodes(nodes: SearchNode[]) {
     return nodes.map(node => ({
       ...node,

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

@@ -9,6 +9,7 @@ import {
 } from '@nestjs/graphql';
 import { GraphQLJSONObject } from 'graphql-scalars';
 
+import { PublicUser } from '../../models';
 import { SearchTable } from './tables';
 
 export enum SearchQueryType {
@@ -40,6 +41,19 @@ registerEnumType(SearchQueryOccur, {
   description: 'Search query occur',
 });
 
+export interface SearchDoc {
+  docId: string;
+  blockId: string;
+  title: string;
+  highlight: string;
+  createdAt: Date;
+  updatedAt: Date;
+  createdByUserId: string;
+  updatedByUserId: string;
+  createdByUser?: PublicUser;
+  updatedByUser?: PublicUser;
+}
+
 @InputType()
 export class SearchQuery {
   @Field(() => SearchQueryType)