Fix: Remove unused attachment data from ipynb files when attachment links are deleted

extensions/ipynb/src/serializers.ts

@@ -120,9 +120,23 @@ function createRawCellFromNotebookCell(cell: NotebookCellData): nbformat.IRawCel
 		source: splitMultilineString(cell.value.replace(/\r\n/g, '\n')),
 		metadata: cellMetadata?.metadata || {} // This cannot be empty.
 	};
+
+	// Only include attachments that are actually referenced in the cell content
 	if (cellMetadata?.attachments) {
-		rawCell.attachments = cellMetadata.attachments;
+		const referencedAttachments = getReferencedAttachmentNames(cell.value);
+		const usedAttachments: { [key: string]: any } = {};
+
+		for (const [filename, attachmentData] of Object.entries(cellMetadata.attachments)) {
+			if (referencedAttachments.has(filename)) {
+				usedAttachments[filename] = attachmentData;
+			}
+		}
+
+		if (Object.keys(usedAttachments).length > 0) {
+			rawCell.attachments = usedAttachments;
+		}
 	}
+
 	if (cellMetadata?.id) {
 		rawCell.id = cellMetadata.id;
 	}
@@ -149,6 +163,21 @@ function splitMultilineString(source: nbformat.MultilineString): string[] {
 	return [];
 }
 
+function getReferencedAttachmentNames(source: string): Set<string> {
+	const filenames = new Set<string>();
+	// Match: ![...](attachment:filename) or ![...](<attachment:filename>)
+	// Uses non-greedy match to stop at first ) or > that ends the link
+	const re = /!\[.*?\]\(<?attachment:(.*?)>?\)/gm;
+
+	let match;
+	while ((match = re.exec(source))) {
+		if (match[1]) {
+			filenames.add(match[1]);
+		}
+	}
+	return filenames;
+}
+
 function translateCellDisplayOutput(output: NotebookCellOutput): JupyterOutput {
 	const customMetadata = output.metadata as CellOutputMetadata | undefined;
 	let result: JupyterOutput;
@@ -371,9 +400,23 @@ export function createMarkdownCellFromNotebookCell(cell: NotebookCellData): nbfo
 		source: splitMultilineString(cell.value.replace(/\r\n/g, '\n')),
 		metadata: cellMetadata?.metadata || {} // This cannot be empty.
 	};
+
+	// Only include attachments that are actually referenced in the cell content
 	if (cellMetadata?.attachments) {
-		markdownCell.attachments = cellMetadata.attachments;
+		const referencedAttachments = getReferencedAttachmentNames(cell.value);
+		const usedAttachments: { [key: string]: any } = {};
+
+		for (const [filename, attachmentData] of Object.entries(cellMetadata.attachments)) {
+			if (referencedAttachments.has(filename)) {
+				usedAttachments[filename] = attachmentData;
+			}
+		}
+
+		if (Object.keys(usedAttachments).length > 0) {
+			markdownCell.attachments = usedAttachments;
+		}
 	}
+
 	if (cellMetadata?.id) {
 		markdownCell.id = cellMetadata.id;
 	}

extensions/ipynb/src/test/serializers.test.ts

@@ -77,7 +77,7 @@ suite(`ipynb serializer`, () => {
 
 
 	test('Serialize', async () => {
-		const markdownCell = new vscode.NotebookCellData(vscode.NotebookCellKind.Markup, '# header1', 'markdown');
+		const markdownCell = new vscode.NotebookCellData(vscode.NotebookCellKind.Markup, '# header1\n![img](attachment:image.png)', 'markdown');
 		markdownCell.metadata = {
 			attachments: {
 				'image.png': {
@@ -103,7 +103,7 @@ suite(`ipynb serializer`, () => {
 			}
 		});
 
-		const markdownCell2 = new vscode.NotebookCellData(vscode.NotebookCellKind.Markup, '# header1', 'markdown');
+		const markdownCell2 = new vscode.NotebookCellData(vscode.NotebookCellKind.Markup, '# header1\n![img](attachment:image.png)', 'markdown');
 		markdownCell2.metadata = {
 			id: '123',
 			metadata: {
@@ -122,7 +122,7 @@ suite(`ipynb serializer`, () => {
 
 		assert.deepStrictEqual(nbMarkdownCell, {
 			cell_type: 'markdown',
-			source: ['# header1'],
+			source: ['# header1\n', '![img](attachment:image.png)'],
 			metadata: {
 				foo: 'bar',
 			},
@@ -135,6 +135,71 @@ suite(`ipynb serializer`, () => {
 		});
 	});
 
+	test('Serialize with unused attachments should not include them', async () => {
+		// Test case: markdown cell has attachment metadata but no attachment reference in content
+		const markdownCell = new vscode.NotebookCellData(vscode.NotebookCellKind.Markup, '# header1', 'markdown');
+		markdownCell.metadata = {
+			attachments: {
+				'unused-image.png': {
+					'image/png': 'unusedImageData'
+				}
+			},
+			id: '123',
+			metadata: {
+				foo: 'bar'
+			}
+		};
+
+		const nbMarkdownCell = createMarkdownCellFromNotebookCell(markdownCell);
+
+		// Should not include attachments since they're not referenced in the content
+		assert.deepStrictEqual(nbMarkdownCell, {
+			cell_type: 'markdown',
+			source: ['# header1'],
+			metadata: {
+				foo: 'bar',
+			},
+			id: '123'
+		});
+	});
+
+	test('Serialize with used attachments should include them', async () => {
+		// Test case: markdown cell has attachment metadata and attachment reference in content
+		const markdownCell = new vscode.NotebookCellData(vscode.NotebookCellKind.Markup, 
+			'# header1\n![alt text](attachment:used-image.png)', 'markdown');
+		markdownCell.metadata = {
+			attachments: {
+				'used-image.png': {
+					'image/png': 'usedImageData'
+				},
+				'unused-image.png': {
+					'image/png': 'unusedImageData'
+				}
+			},
+			id: '123',
+			metadata: {
+				foo: 'bar'
+			}
+		};
+
+		const nbMarkdownCell = createMarkdownCellFromNotebookCell(markdownCell);
+
+		// Should only include attachments that are referenced in the content
+		assert.deepStrictEqual(nbMarkdownCell, {
+			cell_type: 'markdown',
+			source: ['# header1\n', '![alt text](attachment:used-image.png)'],
+			metadata: {
+				foo: 'bar',
+			},
+			attachments: {
+				'used-image.png': {
+					'image/png': 'usedImageData'
+				}
+			},
+			id: '123'
+		});
+	});
+
 	suite('Outputs', () => {
 		function validateCellOutputTranslation(
 			outputs: nbformat.IOutput[],