From 922d57631bbaabfed767ba64fb63ebb89d9811ea Mon Sep 17 00:00:00 2001
From: Calixte Denizet <calixte.denizet@gmail.com>
Date: Mon, 13 Jul 2026 11:20:15 +0200
Subject: [PATCH] Rebuild missing AcroForm fields

---
 src/core/editor/pdf_editor.js |  6 +++---
 test/unit/api_spec.js         | 26 ++++++++++++++++++++++++++
 2 files changed, 29 insertions(+), 3 deletions(-)

diff --git a/src/core/editor/pdf_editor.js b/src/core/editor/pdf_editor.js
index 1cfff4200..49f10c79c 100644
--- a/src/core/editor/pdf_editor.js
+++ b/src/core/editor/pdf_editor.js
@@ -1209,7 +1209,7 @@ class PDFEditor {
                 annotationDict.get("FT"),
                 "Sig"
               );
-              const parentRef = annotationDict.get("Parent") || null;
+              const parentRef = annotationDict.getRaw("Parent") || null;
               // We remove the parent to avoid visiting it when cloning the
               // annotation.
               // It'll be fixed later in #mergeAcroForms when merging the
@@ -2109,14 +2109,14 @@ class PDFEditor {
    * If the document has some fields but no Fields entry in the AcroForm, we
    * need to fix that by creating a Fields entry with the oldest parent field
    * for each field.
-   * @param {Map<Ref, Ref>} fieldToParent
+   * @param {RefSetCache} fieldToParent
    * @param {XRef} xref
    * @returns {Array<Ref>}
    */
   #fixFields(fieldToParent, xref) {
     const newFields = [];
     const processed = new RefSet();
-    for (const [fieldRef, parentRef] of fieldToParent) {
+    for (const [fieldRef, parentRef] of fieldToParent.items()) {
       if (!parentRef) {
         newFields.push(fieldRef);
         continue;
diff --git a/test/unit/api_spec.js b/test/unit/api_spec.js
index ce44fd7f3..03b5b9794 100644
--- a/test/unit/api_spec.js
+++ b/test/unit/api_spec.js
@@ -7501,6 +7501,32 @@ small scripts as well as for`);
         return fontIndex < 0 ? null : operatorList.argsArray[fontIndex][0];
       };
 
+      it("rebuilds a missing AcroForm Fields array", async function () {
+        const data = assemblePdf([
+          "1 0 obj\n<< /Type /Catalog /Pages 2 0 R /AcroForm 6 0 R >>\nendobj\n",
+          "2 0 obj\n<< /Type /Pages /Kids [3 0 R] /Count 1 >>\nendobj\n",
+          "3 0 obj\n<< /Type /Page /Parent 2 0 R /MediaBox [0 0 100 100] " +
+            "/Annots [4 0 R] >>\nendobj\n",
+          "4 0 obj\n<< /Type /Annot /Subtype /Widget /Rect [0 0 20 10] " +
+            "/Parent 5 0 R >>\nendobj\n",
+          "5 0 obj\n<< /FT /Tx /T (group) /Kids [4 0 R] >>\nendobj\n",
+          "6 0 obj\n<< /DA (/Helv 10 Tf) >>\nendobj\n",
+        ]);
+
+        let loadingTask = getDocument({ data });
+        let pdfDoc = await loadingTask.promise;
+        const extracted = await pdfDoc.extractPages([{ document: null }]);
+        expect(extracted).not.toBeNull();
+        await loadingTask.destroy();
+
+        loadingTask = getDocument({ data: extracted });
+        pdfDoc = await loadingTask.promise;
+        expect(Object.keys(await pdfDoc.getFieldObjects())).toEqual(["group"]);
+        const annotations = await (await pdfDoc.getPage(1)).getAnnotations();
+        expect(annotations[0].fieldName).toEqual("group");
+        await loadingTask.destroy();
+      });
+
       it("extract page 2 and check AcroForm Fields T entries", async function () {
         let loadingTask = getDocument(
           buildGetDocumentParams("form_two_pages.pdf")