Merge pull request #18493 from calixteman/fix_utf16_alt_text

[Editor] Correctly save a non-ascii alt text

src/core/annotation.js

@@ -42,12 +42,12 @@ import {
   escapeString,
   getInheritableProperty,
   getRotationMatrix,
-  isAscii,
   isNumberArray,
   lookupMatrix,
   lookupNormalRect,
   lookupRect,
   numberToString,
+  stringToAsciiOrUTF16BE,
   stringToUTF16String,
 } from "./core_utils.js";
 import {
@@ -2133,9 +2133,12 @@ class WidgetAnnotation extends Annotation {
       value,
     };
 
-    const encoder = val =>
-      isAscii(val) ? val : stringToUTF16String(val, /* bigEndian = */ true);
-    dict.set("V", Array.isArray(value) ? value.map(encoder) : encoder(value));
+    dict.set(
+      "V",
+      Array.isArray(value)
+        ? value.map(stringToAsciiOrUTF16BE)
+        : stringToAsciiOrUTF16BE(value)
+    );
     this.amendSavedDict(annotationStorage, dict);
 
     const maybeMK = this._getMKDict(rotation);
@@ -3852,21 +3855,13 @@ class FreeTextAnnotation extends MarkupAnnotation {
     freetext.set("Rect", rect);
     const da = `/Helv ${fontSize} Tf ${getPdfColor(color, /* isFill */ true)}`;
     freetext.set("DA", da);
-    freetext.set(
-      "Contents",
-      isAscii(value)
-        ? value
-        : stringToUTF16String(value, /* bigEndian = */ true)
-    );
+    freetext.set("Contents", stringToAsciiOrUTF16BE(value));
     freetext.set("F", 4);
     freetext.set("Border", [0, 0, 0]);
     freetext.set("Rotate", rotation);
 
     if (user) {
-      freetext.set(
-        "T",
-        isAscii(user) ? user : stringToUTF16String(user, /* bigEndian = */ true)
-      );
+      freetext.set("T", stringToAsciiOrUTF16BE(user));
     }
 
     if (apRef || ap) {
@@ -4600,10 +4595,7 @@ class HighlightAnnotation extends MarkupAnnotation {
     highlight.set("CA", opacity);
 
     if (user) {
-      highlight.set(
-        "T",
-        isAscii(user) ? user : stringToUTF16String(user, /* bigEndian = */ true)
-      );
+      highlight.set("T", stringToAsciiOrUTF16BE(user));
     }
 
     if (apRef || ap) {
@@ -4885,10 +4877,7 @@ class StampAnnotation extends MarkupAnnotation {
     stamp.set("Rotate", rotation);
 
     if (user) {
-      stamp.set(
-        "T",
-        isAscii(user) ? user : stringToUTF16String(user, /* bigEndian = */ true)
-      );
+      stamp.set("T", stringToAsciiOrUTF16BE(user));
     }
 
     if (apRef || ap) {

src/core/core_utils.js

@@ -613,6 +613,10 @@ function getNewAnnotationsMap(annotationStorage) {
   return newAnnotationsByPage.size > 0 ? newAnnotationsByPage : null;
 }
 
+function stringToAsciiOrUTF16BE(str) {
+  return isAscii(str) ? str : stringToUTF16String(str, /* bigEndian = */ true);
+}
+
 function isAscii(str) {
   return /^[\x00-\x7F]*$/.test(str);
 }
@@ -699,6 +703,7 @@ export {
   readUint16,
   readUint32,
   recoverJsURL,
+  stringToAsciiOrUTF16BE,
   stringToUTF16HexString,
   stringToUTF16String,
   toRomanNumerals,

src/core/struct_tree.js

@@ -16,6 +16,7 @@
 import { AnnotationPrefix, stringToPDFString, warn } from "../shared/util.js";
 import { Dict, isName, Name, Ref, RefSetCache } from "./primitives.js";
 import { NumberTree } from "./name_number_tree.js";
+import { stringToAsciiOrUTF16BE } from "./core_utils.js";
 import { writeObject } from "./writer.js";
 
 const MAX_DEPTH = 40;
@@ -316,19 +317,19 @@ class StructTreeRoot {
         tagDict.set("S", Name.get(type));
 
         if (title) {
-          tagDict.set("T", title);
+          tagDict.set("T", stringToAsciiOrUTF16BE(title));
         }
         if (lang) {
           tagDict.set("Lang", lang);
         }
         if (alt) {
-          tagDict.set("Alt", alt);
+          tagDict.set("Alt", stringToAsciiOrUTF16BE(alt));
         }
         if (expanded) {
-          tagDict.set("E", expanded);
+          tagDict.set("E", stringToAsciiOrUTF16BE(expanded));
         }
         if (actualText) {
-          tagDict.set("ActualText", actualText);
+          tagDict.set("ActualText", stringToAsciiOrUTF16BE(actualText));
         }
 
         await this.#updateParentTag({

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();
     });