feat(ocr): not well integrate with sync

apps/server/src/becca/entities/bblob.ts

@@ -10,7 +10,7 @@ class BBlob extends AbstractBeccaEntity<BBlob> {
         return "blobId";
     }
     static get hashedProperties() {
-        return ["blobId", "content", "textRepresentation"];
+        return ["blobId", "content"];
     }
 
     content!: string | Buffer;
@@ -41,6 +41,11 @@ class BBlob extends AbstractBeccaEntity<BBlob> {
             utcDateModified: this.utcDateModified
         };
     }
+
+    protected getPojoToSave() {
+        const { contentLength: _, ...pojo } = this.getPojo();
+        return pojo;
+    }
 }
 
 export default BBlob;

apps/server/src/services/blob-interface.ts

@@ -1,5 +1,6 @@
 export interface Blob {
     blobId: string;
     content: string | Buffer;
+    textRepresentation?: string | null;
     utcDateModified: string;
 }

apps/server/src/services/blob.ts

@@ -50,8 +50,8 @@ function processContent(content: Buffer | string | null, isProtected: boolean, i
     }
 }
 
-function calculateContentHash({ blobId, content }: Blob) {
-    return hash(`${blobId}|${content.toString()}`);
+function calculateContentHash({ blobId, content, textRepresentation }: Blob) {
+    return hash(`${blobId}|${content.toString()}|${textRepresentation ?? ""}`);
 }
 
 export default {

apps/server/src/services/entity_changes.ts

@@ -146,7 +146,7 @@ function fillEntityChanges(entityName: string, entityPrimaryKey: string, conditi
             };
 
             if (entityName === "blobs") {
-                const blob = sql.getRow<Blob>("SELECT blobId, content, utcDateModified FROM blobs WHERE blobId = ?", [entityId]);
+                const blob = sql.getRow<Blob>("SELECT blobId, content, textRepresentation, utcDateModified FROM blobs WHERE blobId = ?", [entityId]);
                 ec.hash = blobService.calculateContentHash(blob);
                 ec.utcDateChanged = blob.utcDateModified;
                 ec.isSynced = true; // blobs are always synced

apps/server/src/services/ocr/ocr_service.ts

@@ -2,6 +2,8 @@ import { getTesseractCode } from '@triliumnext/commons';
 import Tesseract from 'tesseract.js';
 
 import becca from '../../becca/becca.js';
+import blobService from '../blob.js';
+import entityChangesService from '../entity_changes.js';
 import log from '../log.js';
 import options from '../options.js';
 import sql from '../sql.js';
@@ -277,17 +279,14 @@ class OCRService {
         }
 
         try {
-            // Store OCR text and timestamp in blobs table
             sql.execute(`
                 UPDATE blobs SET
                     textRepresentation = ?,
                     textExtractionLastProcessed = ?
                 WHERE blobId = ?
-            `, [
-                ocrResult.text,
-                new Date().toISOString(),
-                blobId
-            ]);
+            `, [ocrResult.text, new Date().toISOString(), blobId]);
+
+            this.putBlobEntityChange(blobId);
 
             log.info(`Stored OCR result for blob ${blobId}`);
         } catch (error) {
@@ -363,10 +362,12 @@ class OCRService {
     deleteOCRResult(blobId: string): void {
         try {
             sql.execute(`
-                UPDATE blobs SET textRepresentation = NULL
+                UPDATE blobs SET textRepresentation = NULL, textExtractionLastProcessed = NULL
                 WHERE blobId = ?
             `, [blobId]);
 
+            this.putBlobEntityChange(blobId);
+
             log.info(`Deleted OCR result for blob ${blobId}`);
         } catch (error) {
             log.error(`Failed to delete OCR result for blob ${blobId}: ${error}`);
@@ -558,6 +559,29 @@ class OCRService {
     /**
      * Get processor for a given MIME type
      */
+    /**
+     * Notifies the sync system that a blob has changed, without modifying the blob's identity.
+     */
+    private putBlobEntityChange(blobId: string): void {
+        const blob = becca.getBlob({ blobId });
+        if (!blob || !blob.blobId) return;
+
+        const hash = blobService.calculateContentHash({
+            blobId: blob.blobId,
+            content: blob.content,
+            textRepresentation: blob.textRepresentation,
+            utcDateModified: blob.utcDateModified!
+        });
+        entityChangesService.putEntityChange({
+            entityName: "blobs",
+            entityId: blobId,
+            hash,
+            isErased: false,
+            utcDateChanged: blob.utcDateModified,
+            isSynced: true
+        });
+    }
+
     private getProcessorForMimeType(mimeType: string): FileProcessor | null {
         for (const processor of this.processors.values()) {
             if (processor.canProcess(mimeType)) {
@@ -641,12 +665,12 @@ class OCRService {
 
         try {
             sql.execute(`
-                UPDATE blobs SET
-                    textRepresentation = NULL,
-                    textExtractionLastProcessed = NULL
+                UPDATE blobs SET textRepresentation = NULL, textExtractionLastProcessed = NULL
                 WHERE blobId = ?
             `, [blobId]);
 
+            this.putBlobEntityChange(blobId);
+
             log.info(`Invalidated OCR result for blob ${blobId}`);
         } catch (error) {
             log.error(`Failed to invalidate OCR result for blob ${blobId}: ${error}`);