From 1b82ad8fff45f8964f5b3decf6f525d38dd4a76a Mon Sep 17 00:00:00 2001
From: Tim van der Meij <timvandermeij@gmail.com>
Date: Sat, 22 Aug 2020 15:02:29 +0200
Subject: [PATCH] Decode widget form values consistently

The helper method `_decodeFormValue` is used to ensure that it happens
in one place. Note that form values are field values, display values
and export values.
---
 src/core/annotation.js       | 43 ++++++++++++++++++++++++++++--------
 test/unit/annotation_spec.js | 19 ++++++++--------
 2 files changed, 43 insertions(+), 19 deletions(-)

diff --git a/src/core/annotation.js b/src/core/annotation.js
index 5133633c0..bca494b58 100644
--- a/src/core/annotation.js
+++ b/src/core/annotation.js
@@ -803,11 +803,14 @@ class WidgetAnnotation extends Annotation {
 
     data.annotationType = AnnotationType.WIDGET;
     data.fieldName = this._constructFieldName(dict);
-    data.fieldValue = getInheritableProperty({
+
+    const fieldValue = getInheritableProperty({
       dict,
       key: "V",
       getArray: true,
     });
+    data.fieldValue = this._decodeFormValue(fieldValue);
+
     data.alternativeText = stringToPDFString(dict.get("TU") || "");
     data.defaultAppearance =
       getInheritableProperty({ dict, key: "DA" }) ||
@@ -882,6 +885,28 @@ class WidgetAnnotation extends Annotation {
     return fieldName.join(".");
   }
 
+  /**
+   * Decode the given form value.
+   *
+   * @private
+   * @memberof WidgetAnnotation
+   * @param {Array<string>|Name|string} formValue - The (possibly encoded)
+   *   form value.
+   * @returns {Array<string>|string|null}
+   */
+  _decodeFormValue(formValue) {
+    if (Array.isArray(formValue)) {
+      return formValue
+        .filter(item => isString(item))
+        .map(item => stringToPDFString(item));
+    } else if (isName(formValue)) {
+      return stringToPDFString(formValue.name);
+    } else if (isString(formValue)) {
+      return stringToPDFString(formValue);
+    }
+    return null;
+  }
+
   /**
    * Check if a provided field flag is set.
    *
@@ -1194,7 +1219,9 @@ class TextWidgetAnnotation extends WidgetAnnotation {
     const dict = params.dict;
 
     // The field value is always a string.
-    this.data.fieldValue = stringToPDFString(this.data.fieldValue || "");
+    if (!isString(this.data.fieldValue)) {
+      this.data.fieldValue = "";
+    }
 
     // Determine the alignment of text in the field.
     let alignment = getInheritableProperty({ dict, key: "Q" });
@@ -1499,10 +1526,6 @@ class ButtonWidgetAnnotation extends WidgetAnnotation {
   }
 
   _processCheckBox(params) {
-    if (isName(this.data.fieldValue)) {
-      this.data.fieldValue = this.data.fieldValue.name;
-    }
-
     const customAppearance = params.dict.get("AP");
     if (!isDict(customAppearance)) {
       return;
@@ -1541,7 +1564,7 @@ class ButtonWidgetAnnotation extends WidgetAnnotation {
       const fieldParentValue = fieldParent.get("V");
       if (isName(fieldParentValue)) {
         this.parent = params.dict.getRaw("Parent");
-        this.data.fieldValue = fieldParentValue.name;
+        this.data.fieldValue = this._decodeFormValue(fieldParentValue);
       }
     }
 
@@ -1602,8 +1625,10 @@ class ChoiceWidgetAnnotation extends WidgetAnnotation {
         const isOptionArray = Array.isArray(option);
 
         this.data.options[i] = {
-          exportValue: isOptionArray ? xref.fetchIfRef(option[0]) : option,
-          displayValue: stringToPDFString(
+          exportValue: this._decodeFormValue(
+            isOptionArray ? xref.fetchIfRef(option[0]) : option
+          ),
+          displayValue: this._decodeFormValue(
             isOptionArray ? xref.fetchIfRef(option[1]) : option
           ),
         };
diff --git a/test/unit/annotation_spec.js b/test/unit/annotation_spec.js
index 9964cb676..46ed39974 100644
--- a/test/unit/annotation_spec.js
+++ b/test/unit/annotation_spec.js
@@ -2450,16 +2450,12 @@ describe("annotation", function () {
       }, done.fail);
     });
 
-    it("should sanitize display values in option arrays (issue 8947)", function (done) {
-      // The option value is a UTF-16BE string. The display value should be
-      // sanitized, but the export value should remain the same since that
-      // may be used as a unique identifier when exporting form values.
-      const options = ["\xFE\xFF\x00F\x00o\x00o"];
-      const expected = [
-        { exportValue: "\xFE\xFF\x00F\x00o\x00o", displayValue: "Foo" },
-      ];
+    it("should decode form values", function (done) {
+      const encodedString = "\xFE\xFF\x00F\x00o\x00o";
+      const decodedString = "Foo";
 
-      choiceWidgetDict.set("Opt", options);
+      choiceWidgetDict.set("Opt", [encodedString]);
+      choiceWidgetDict.set("V", encodedString);
 
       const choiceWidgetRef = Ref.get(984, 0);
       const xref = new XRefMock([
@@ -2473,7 +2469,10 @@ describe("annotation", function () {
         idFactoryMock
       ).then(({ data }) => {
         expect(data.annotationType).toEqual(AnnotationType.WIDGET);
-        expect(data.options).toEqual(expected);
+        expect(data.fieldValue).toEqual([decodedString]);
+        expect(data.options).toEqual([
+          { exportValue: decodedString, displayValue: decodedString },
+        ]);
         done();
       }, done.fail);
     });