feat(server): get full content for doc embedding (#11138)

packages/backend/server/src/core/doc/reader.ts

@@ -42,10 +42,10 @@ export abstract class DocReader {
     protected readonly blobStorage: WorkspaceBlobStorage
   ) {}
 
-  parseDocContent(bin: Uint8Array) {
+  parseDocContent(bin: Uint8Array, maxSummaryLength = 150) {
     const doc = new YDoc();
     applyUpdate(doc, bin);
-    return parsePageDoc(doc);
+    return parsePageDoc(doc, { maxSummaryLength });
   }
 
   parseWorkspaceContent(bin: Uint8Array) {
@@ -85,6 +85,13 @@ export abstract class DocReader {
     return content;
   }
 
+  async getFullDocContent(
+    workspaceId: string,
+    docId: string
+  ): Promise<PageDocContent | null> {
+    return await this.getDocContentWithoutCache(workspaceId, docId, true);
+  }
+
   /**
    * Get workspace content, try to read from database first.
    * If not exists, read from `getWorkspaceContentWithoutCache()` and save it back to database.
@@ -131,7 +138,8 @@ export abstract class DocReader {
 
   protected abstract getDocContentWithoutCache(
     workspaceId: string,
-    guid: string
+    guid: string,
+    fullContent?: boolean
   ): Promise<PageDocContent | null>;
 
   protected abstract getWorkspaceContentWithoutCache(
@@ -180,13 +188,14 @@ export class DatabaseDocReader extends DocReader {
 
   protected override async getDocContentWithoutCache(
     workspaceId: string,
-    guid: string
+    guid: string,
+    fullContent?: boolean
   ): Promise<PageDocContent | null> {
     const docRecord = await this.workspace.getDoc(workspaceId, guid);
     if (!docRecord) {
       return null;
     }
-    return this.parseDocContent(docRecord.bin);
+    return this.parseDocContent(docRecord.bin, fullContent ? -1 : 150);
   }
 
   protected override async getWorkspaceContentWithoutCache(

packages/backend/server/src/core/utils/blocksuite.ts

@@ -115,7 +115,9 @@ export function parsePageDoc(
           continue;
         }
 
-        if (summaryLenNeeded > 0) {
+        if (summaryLenNeeded === -1) {
+          content.summary += text.toString();
+        } else if (summaryLenNeeded > 0) {
           content.summary += text.toString();
           summaryLenNeeded -= text.length;
         } else {

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

@@ -159,7 +159,7 @@ export class CopilotContextDocJob implements OnModuleInit {
     if (!this.supportEmbedding) return;
 
     try {
-      const content = await this.doc.getDocContent(workspaceId, docId);
+      const content = await this.doc.getFullDocContent(workspaceId, docId);
       if (content) {
         // no need to check if embeddings is empty, will throw internally
         const embeddings = await this.embeddingClient.getFileEmbeddings(