From 92263c6a580b58d3102a93e38530617a05b656af Mon Sep 17 00:00:00 2001
From: Calixte Denizet <calixte.denizet@gmail.com>
Date: Wed, 24 Jul 2024 18:55:31 +0200
Subject: [PATCH] [Editor] Correctly save a non-ascii alt text

---
 src/core/struct_tree.js | 15 +++++++++----
 test/unit/api_spec.js   | 47 ++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 57 insertions(+), 5 deletions(-)

diff --git a/src/core/struct_tree.js b/src/core/struct_tree.js
index 9f4f552b0f3e64..1461a9bd8e6fc2 100644
--- a/src/core/struct_tree.js
+++ b/src/core/struct_tree.js
@@ -15,6 +15,7 @@
 
 import { AnnotationPrefix, stringToPDFString, warn } from "../shared/util.js";
 import { Dict, isName, Name, Ref, RefSetCache } from "./primitives.js";
+import { isAscii, stringToUTF16String } from "./core_utils.js";
 import { NumberTree } from "./name_number_tree.js";
 import { writeObject } from "./writer.js";
 
@@ -281,6 +282,12 @@ class StructTreeRoot {
     }
   }
 
+  static convertString(str) {
+    return isAscii(str)
+      ? str
+      : stringToUTF16String(str, /* bigEndian = */ true);
+  }
+
   static async #writeKids({
     newAnnotationsByPage,
     structTreeRootRef,
@@ -316,19 +323,19 @@ class StructTreeRoot {
         tagDict.set("S", Name.get(type));
 
         if (title) {
-          tagDict.set("T", title);
+          tagDict.set("T", this.convertString(title));
         }
         if (lang) {
           tagDict.set("Lang", lang);
         }
         if (alt) {
-          tagDict.set("Alt", alt);
+          tagDict.set("Alt", this.convertString(alt));
         }
         if (expanded) {
-          tagDict.set("E", expanded);
+          tagDict.set("E", this.convertString(expanded));
         }
         if (actualText) {
-          tagDict.set("ActualText", actualText);
+          tagDict.set("ActualText", this.convertString(actualText));
         }
 
         await this.#updateParentTag({
diff --git a/test/unit/api_spec.js b/test/unit/api_spec.js
index 680ec6a9eb080a..6d6a2b0bdf5bfb 100644
--- a/test/unit/api_spec.js
+++ b/test/unit/api_spec.js
@@ -2524,6 +2524,21 @@ describe("api", function () {
           alt: "Hello World",
         },
       });
+      // Test if an alt-text using utf-16 is correctly handled.
+      // The Mahjong tile code is 0x1F000.
+      pdfDoc.annotationStorage.setValue("pdfjs_internal_editor_1", {
+        annotationType: AnnotationEditorType.STAMP,
+        rect: [128, 400, 148, 420],
+        rotation: 0,
+        bitmap: structuredClone(bitmap),
+        bitmapId: "im2",
+        pageIndex: 0,
+        structTreeParentId: "p3R_mc14",
+        accessibilityData: {
+          type: "Figure",
+          alt: "Γειά σου with a Mahjong tile 🀀",
+        },
+      });
 
       const data = await pdfDoc.saveDocument();
       await loadingTask.destroy();
@@ -2532,7 +2547,7 @@ describe("api", function () {
       pdfDoc = await loadingTask.promise;
       const page = await pdfDoc.getPage(1);
       const tree = await page.getStructTree();
-      const [predecessor, leaf] = findNode(
+      let [predecessor, leaf] = findNode(
         null,
         tree,
         0,
@@ -2560,6 +2575,36 @@ describe("api", function () {
         alt: "Hello World",
       });
 
+      let count = 0;
+      [predecessor, leaf] = findNode(null, tree, 0, node => {
+        if (node.role === "Figure") {
+          count += 1;
+          return count === 2;
+        }
+        return false;
+      });
+
+      expect(predecessor).toEqual({
+        role: "Span",
+        children: [
+          {
+            type: "content",
+            id: "p3R_mc14",
+          },
+        ],
+      });
+
+      expect(leaf).toEqual({
+        role: "Figure",
+        children: [
+          {
+            type: "annotation",
+            id: "pdfjs_internal_id_481R",
+          },
+        ],
+        alt: "Γειά σου with a Mahjong tile 🀀",
+      });
+
       await loadingTask.destroy();
     });