pdf-server: import saved form field values from PDF as baseline

getFieldObjects() returns the PDF's stored form values but we were only
reading field IDs/pages from it. After saving a filled form and reopening,
the panel showed nothing and there was no way to see what was filled.

- New pdfBaselineFormValues map, populated in buildFieldNameMap() from
  each field's .value (skipping empty/Off/button values). Seeds
  formFieldValues so the panel shows PDF-stored values on open.
- computeDiff takes an optional baselineFormFields param and only
  includes values that differ — opening a filled PDF doesn't mark dirty,
  editing a field does, reverting to the PDF's value marks clean again.
- importFieldValue() normalises radio-group value lookup (parent entry
  has value=undefined, children have the real export value), checkbox→
  true, listbox array→joined string.

Author: ochafik

examples/pdf-server/src/mcp-app.ts

@@ -83,6 +83,8 @@ const imageCache = new Map<string, HTMLImageElement>();
 
 /** Annotations imported from the PDF file (baseline for diff computation). */
 let pdfBaselineAnnotations: PdfAnnotationDef[] = [];
+/** Form field values stored in the PDF file itself (baseline for diff computation). */
+const pdfBaselineFormValues = new Map<string, string | boolean>();
 
 // Dirty flag — tracks unsaved local changes
 let isDirty = false;
@@ -3161,6 +3163,7 @@ function persistAnnotations(): void {
     pdfBaselineAnnotations,
     currentAnnotations,
     formFieldValues,
+    pdfBaselineFormValues,
   );
 
   // Dirty tracks whether there are unsaved changes. Undoing back to baseline
@@ -3222,6 +3225,25 @@ function restoreAnnotations(): void {
 // PDF.js Form Field Name → ID Mapping
 // =============================================================================
 
+/**
+ * Extract a meaningful value from a getFieldObjects() field array.
+ * Returns null for empty/unfilled/button fields so they don't clutter
+ * the panel or count as baseline edits.
+ */
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+function importFieldValue(fieldArr: any[]): string | boolean | null {
+  // For radio groups, getFieldObjects returns a parent entry with
+  // value=undefined plus child entries — find the first with a real value.
+  const f = fieldArr.find((x) => x.value != null) ?? fieldArr[0];
+  if (!f || f.type === "button") return null;
+  const v = f.value;
+  if (v == null || v === "" || v === "Off") return null;
+  if (f.type === "checkbox") return true;
+  if (f.type === "radiobutton") return String(v);
+  if (Array.isArray(v)) return v.join(", ");
+  return String(v);
+}
+
 /** Build mapping from field names (used by fill_form) to annotation IDs (used by annotationStorage). */
 async function buildFieldNameMap(
   doc: pdfjsLib.PDFDocumentProxy,
@@ -3231,6 +3253,7 @@ async function buildFieldNameMap(
   fieldNameToLabel.clear();
   fieldNameToOrder.clear();
   cachedFieldObjects = null;
+  pdfBaselineFormValues.clear();
   try {
     const fieldObjects = await doc.getFieldObjects();
     cachedFieldObjects = fieldObjects as Record<string, any[]> | null;
@@ -3247,6 +3270,16 @@ async function buildFieldNameMap(
         if (firstField && typeof firstField.page === "number") {
           fieldNameToPage.set(name, firstField.page + 1);
         }
+        // Import baseline value (already saved in the PDF). Skip button-type
+        // fields and empty/Off values that represent "unfilled".
+        const v = importFieldValue(fieldArr);
+        if (v !== null) {
+          pdfBaselineFormValues.set(name, v);
+          // Seed current state from baseline so the panel shows it. A
+          // restored localStorage diff (applied later in restoreAnnotations)
+          // will overwrite specific fields the user changed.
+          if (!formFieldValues.has(name)) formFieldValues.set(name, v);
+        }
       }
     }
   } catch {
@@ -4333,6 +4366,7 @@ async function reloadPdf(): Promise<void> {
   undoStack.length = 0;
   redoStack.length = 0;
   pdfBaselineAnnotations = [];
+  pdfBaselineFormValues.clear();
   pageTextCache.clear();
   pageTextItemsCache.clear();
   allMatches = [];

examples/pdf-server/src/pdf-annotations.test.ts

@@ -278,6 +278,37 @@ describe("computeDiff", () => {
     expect(diff.formFields).toEqual({ name: "Alice", agree: true });
   });
 
+  it("omits form fields matching baseline", () => {
+    const baseline = new Map<string, string | boolean>([
+      ["name", "Alice"],
+      ["agree", true],
+    ]);
+    const fields = new Map<string, string | boolean>([
+      ["name", "Alice"], // unchanged
+      ["agree", false], // changed
+      ["email", "a@b"], // new
+    ]);
+    const diff = computeDiff([], [], fields, baseline);
+    expect(diff.formFields).toEqual({ agree: false, email: "a@b" });
+  });
+
+  it("records fields cleared from baseline", () => {
+    const baseline = new Map<string, string | boolean>([["name", "Alice"]]);
+    const fields = new Map<string, string | boolean>(); // cleared
+    const diff = computeDiff([], [], fields, baseline);
+    expect(diff.formFields).toEqual({ name: "" });
+  });
+
+  it("produces empty diff when all form values match baseline", () => {
+    const baseline = new Map<string, string | boolean>([
+      ["name", "Alice"],
+      ["agree", true],
+    ]);
+    const diff = computeDiff([], [], new Map(baseline), baseline);
+    expect(diff.formFields).toEqual({});
+    expect(isDiffEmpty(diff)).toBe(true);
+  });
+
   it("round-trips through mergeAnnotations", () => {
     const userStamp: PdfAnnotationDef = {
       type: "stamp",

examples/pdf-server/src/pdf-annotations.ts

@@ -282,6 +282,7 @@ export function computeDiff(
   pdfAnnotations: PdfAnnotationDef[],
   currentAnnotations: PdfAnnotationDef[],
   formFields: Map<string, string | boolean>,
+  baselineFormFields?: Map<string, string | boolean>,
 ): AnnotationDiff {
   const pdfIds = new Set(pdfAnnotations.map((a) => a.id));
   const currentIds = new Set(currentAnnotations.map((a) => a.id));
@@ -294,11 +295,21 @@ export function computeDiff(
     .filter((a) => !currentIds.has(a.id))
     .map((a) => a.id);
 
-  // Form fields
+  // Form fields: only values that differ from what's already in the PDF.
+  // Without a baseline, every filled field is a user edit (back-compat).
   const formFieldsObj: Record<string, string | boolean> = {};
   for (const [k, v] of formFields) {
+    if (baselineFormFields?.get(k) === v) continue;
     formFieldsObj[k] = v;
   }
+  // Fields present in baseline but cleared in current are also a change
+  if (baselineFormFields) {
+    for (const [k, v] of baselineFormFields) {
+      if (!formFields.has(k) && v !== "" && v !== false) {
+        formFieldsObj[k] = formFields.get(k) ?? "";
+      }
+    }
+  }
 
   return { added, removed, formFields: formFieldsObj };
 }