fix(node): add PdfAnnotation type and annotations mapping to TypeScript wrapper

The Node binding's TypeScript layer was missing the PdfAnnotation interface
and annotations field on ExtractionResult, causing pdf_annotations e2e test
to fail with "expected undefined to be defined".

Author: Goldziher

crates/kreuzberg-node/typescript/core/type-converters.ts

@@ -330,6 +330,12 @@ function convertResult(rawResult: unknown): ExtractionResult {
 		returnObj.processingWarnings = processingWarningsData as Array<{ source: string; message: string }>;
 	}
 
+	// biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+	const annotationsData = result["annotations"];
+	if (Array.isArray(annotationsData)) {
+		returnObj.annotations = annotationsData as import("../types.js").PdfAnnotation[];
+	}
+
 	return returnObj;
 }
 

crates/kreuzberg-node/typescript/types.ts

@@ -1124,6 +1124,20 @@ export interface BoundingBox {
 	y1: number;
 }
 
+/**
+ * A PDF annotation extracted from a document page.
+ */
+export interface PdfAnnotation {
+	/** Type of annotation (e.g., "text", "highlight", "link", "underline") */
+	annotationType: string;
+	/** Text content of the annotation, if available */
+	content?: string | null;
+	/** Page number (1-indexed) where the annotation appears */
+	pageNumber: number;
+	/** Bounding box of the annotation on the page */
+	boundingBox?: BoundingBox | null;
+}
+
 /**
  * Metadata for a semantic element.
  *
@@ -1207,6 +1221,9 @@ export interface ExtractionResult {
 
 	/** Structured document tree when include_document_structure is enabled, null otherwise */
 	document?: Record<string, unknown> | null;
+
+	/** PDF annotations when extract_annotations is enabled, null otherwise */
+	annotations?: PdfAnnotation[] | null;
 }
 
 /** Post-processor execution stage in the extraction pipeline. */