feat(server): search blob names from indexer (#12822)

#### PR Dependency Tree


* **PR #12822** 👈

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

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

- **New Features**
- Added the ability to search for blob names by their IDs within a
workspace.
- **Tests**
- Introduced new test cases and snapshot tests to validate searching
blob names and reading filenames from document snapshots.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

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

@@ -494,3 +494,30 @@ Generated by [AVA](https://avajs.dev).
         ],
       },
     ]
+
+## should search blob names from doc snapshot work
+
+> Snapshot 1
+
+    Map {
+      'ldZMrM4PDlsNG4Q4YvCsz623h6TKu4qI9_FpTqIypfw=' => 'test file name here.txt',
+    }
+
+## should search blob names work
+
+> Snapshot 1
+
+    [
+      [
+        'blob1',
+        'blob1 name.txt',
+      ],
+      [
+        'blob2',
+        'blob2 name.md',
+      ],
+      [
+        'blob3',
+        'blob3 name.docx',
+      ],
+    ]

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

@@ -2113,3 +2113,103 @@ test('should index doc work', async t => {
   t.is(module.event.count('doc.indexer.updated'), count + 1);
 });
 // #endregion
+
+// #region searchBlobNames()
+
+test('should search blob names from doc snapshot work', async t => {
+  const docSnapshot = await module.create(Mockers.DocSnapshot, {
+    workspaceId: workspace.id,
+    user,
+    snapshotFile: 'test-doc-with-blob.snapshot.bin',
+  });
+
+  await indexerService.indexDoc(workspace.id, docSnapshot.id, {
+    refresh: true,
+  });
+
+  const blobNameMap = await indexerService.searchBlobNames(workspace.id, [
+    'ldZMrM4PDlsNG4Q4YvCsz623h6TKu4qI9_FpTqIypfw=',
+  ]);
+
+  t.snapshot(blobNameMap);
+});
+
+test('should search blob names work', async t => {
+  const workspaceId = randomUUID();
+  const blobId1 = 'blob1';
+  const blobId2 = 'blob2';
+  const blobId3 = 'blob3';
+  const blobId4 = 'blob4';
+
+  await indexerService.write(
+    SearchTable.block,
+    [
+      {
+        workspaceId,
+        blob: blobId1,
+        content: 'blob1 name.txt',
+        flavour: 'affine:attachment',
+        docId: randomUUID(),
+        blockId: randomUUID(),
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date(),
+        updatedAt: new Date(),
+      },
+      {
+        workspaceId,
+        blob: blobId2,
+        content: 'blob2 name.md',
+        flavour: 'affine:attachment',
+        docId: randomUUID(),
+        blockId: randomUUID(),
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date(),
+        updatedAt: new Date(),
+      },
+      {
+        workspaceId,
+        blob: blobId3,
+        content: 'blob3 name.docx',
+        flavour: 'affine:attachment',
+        docId: randomUUID(),
+        blockId: randomUUID(),
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date(),
+        updatedAt: new Date(),
+      },
+      // no attachment
+      {
+        workspaceId,
+        blob: blobId3,
+        content: 'mock blob3 content',
+        flavour: 'affine:page',
+        docId: randomUUID(),
+        blockId: randomUUID(),
+        createdByUserId: user.id,
+        updatedByUserId: user.id,
+        createdAt: new Date(),
+        updatedAt: new Date(),
+      },
+    ],
+    {
+      refresh: true,
+    }
+  );
+
+  const blobNameMap = await indexerService.searchBlobNames(workspaceId, [
+    blobId1,
+    blobId2,
+    blobId3,
+    blobId4,
+  ]);
+
+  t.is(blobNameMap.size, 3);
+  t.snapshot(
+    Array.from(blobNameMap.entries()).sort((a, b) => a[0].localeCompare(b[0]))
+  );
+});
+
+// #endregion

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

@@ -387,6 +387,52 @@ export class IndexerService {
     await searchProvider.deleteByQuery(table, dsl, options);
   }
 
+  async searchBlobNames(workspaceId: string, blobIds: string[]) {
+    const result = await this.search({
+      table: SearchTable.block,
+      query: {
+        type: SearchQueryType.boolean,
+        occur: SearchQueryOccur.must,
+        queries: [
+          {
+            type: SearchQueryType.match,
+            field: 'workspaceId',
+            match: workspaceId,
+          },
+          {
+            type: SearchQueryType.match,
+            field: 'flavour',
+            match: 'affine:attachment',
+          },
+          {
+            type: SearchQueryType.boolean,
+            occur: SearchQueryOccur.should,
+            queries: blobIds.map(blobId => ({
+              type: SearchQueryType.match,
+              field: 'blob',
+              match: blobId,
+            })),
+          },
+        ],
+      },
+      options: {
+        fields: ['blob', 'content'],
+        pagination: {
+          limit: 10000,
+        },
+      },
+    });
+    const blobNameMap = new Map<string, string>();
+    for (const node of result.nodes) {
+      const blobId = node.fields.blob[0] as string;
+      const content = node.fields.content[0] as string;
+      if (blobId && content) {
+        blobNameMap.set(blobId, content);
+      }
+    }
+    return blobNameMap;
+  }
+
   #formatSearchNodes(nodes: SearchNode[]) {
     return nodes.map(node => ({
       ...node,