fix: correct UTF-16BE string parsing and TempObject handling in binary plist parser (#65)

Author: luckyyyyy

src/lib/plist/binary-plist-creator.ts

@@ -59,7 +59,11 @@ class BinaryPlistCreator {
     const objectData: Buffer[] = [];
 
     for (const value of this._objectTable) {
-      objectOffsets.push(this._calculateObjectDataLength(objectData));
+      // Calculate offset including the header length
+      objectOffsets.push(
+        BPLIST_MAGIC_AND_VERSION.length +
+          this._calculateObjectDataLength(objectData),
+      );
       objectData.push(this._createObjectData(value));
     }
 
@@ -341,11 +345,22 @@ class BinaryPlistCreator {
     // Check if string can be ASCII
     // eslint-disable-next-line no-control-regex
     const isAscii = /^[\x00-\x7F]*$/.test(value);
-    const stringBuffer = isAscii
-      ? Buffer.from(value, 'ascii')
-      : Buffer.from(value, 'utf16le');
 
-    // Fixed the typo here - using stringBuffer.length instead of value.length for Unicode strings
+    let stringBuffer: Buffer;
+    if (isAscii) {
+      stringBuffer = Buffer.from(value, 'ascii');
+    } else {
+      // Unicode strings should be stored as UTF-16BE in binary plists
+      const utf16leBuffer = Buffer.from(value, 'utf16le');
+      stringBuffer = Buffer.alloc(utf16leBuffer.length);
+
+      // Convert UTF-16LE to UTF-16BE
+      for (let i = 0; i < utf16leBuffer.length; i += 2) {
+        stringBuffer[i] = utf16leBuffer[i + 1]; // High byte
+        stringBuffer[i + 1] = utf16leBuffer[i]; // Low byte
+      }
+    }
+
     const length = isAscii ? value.length : stringBuffer.length / 2;
     let header: Buffer;
 

src/lib/plist/binary-plist-parser.ts

@@ -270,15 +270,21 @@ class BinaryPlistParser {
    * @returns The parsed string
    */
   private _parseUnicodeString(startOffset: number, objLength: number): string {
-    // Unicode strings are stored as UTF-16BE
-    const utf16Buffer = Buffer.alloc(objLength * 2);
-    for (let j = 0; j < objLength; j++) {
-      utf16Buffer.writeUInt16BE(
-        this._buffer.readUInt16BE(startOffset + j * 2),
-        j * 2,
-      );
+    // Unicode strings are stored as UTF-16BE in binary plists
+    const bytesToRead = objLength * 2;
+    const stringBuffer = this._buffer.slice(
+      startOffset,
+      startOffset + bytesToRead,
+    );
+
+    // Convert UTF-16BE to UTF-16LE for proper decoding
+    const utf16leBuffer = Buffer.alloc(bytesToRead);
+    for (let i = 0; i < bytesToRead; i += 2) {
+      utf16leBuffer[i] = stringBuffer[i + 1]; // Low byte
+      utf16leBuffer[i + 1] = stringBuffer[i]; // High byte
     }
-    return utf16Buffer.toString('utf16le', 0, objLength * 2);
+
+    return utf16leBuffer.toString('utf16le');
   }
 
   /**
@@ -476,8 +482,10 @@ class BinaryPlistParser {
         obj.startOffset + j * this._objectRefSize,
       );
       const refValue = this._objectTable[refIdx];
-      // Ensure we're not adding a TempObject to the array
-      if (!this._isTempObject(refValue)) {
+      // Handle TempObjects correctly - they should be resolved by the time we get here
+      if (this._isTempObject(refValue)) {
+        array.push(refValue.value);
+      } else {
         array.push(refValue);
       }
     }
@@ -511,8 +519,10 @@ class BinaryPlistParser {
         );
       }
 
-      // Ensure we're not adding a TempObject to the dictionary
-      if (!this._isTempObject(value)) {
+      // Handle TempObjects correctly - they should be resolved by the time we get here
+      if (this._isTempObject(value)) {
+        dict[key] = value.value;
+      } else {
         dict[key] = value;
       }
     }

test/unit/plist/plist.spec.ts

@@ -47,6 +47,8 @@ describe('Plist Module', function () {
         nestedArray: [1, 2],
       },
       specialChars: '<Hello & World>',
+      emoji: '😀',
+      unicode: '测试',
     };
   });
 
@@ -104,7 +106,6 @@ describe('Plist Module', function () {
       const binaryPlist = createBinaryPlist(expectedPlistObject);
       expect(isBinaryPlist(binaryPlist)).to.be.true;
       expect(isBinaryPlist(Buffer.from(sampleXmlPlistContent))).to.be.false;
-
       // Create and verify
       expect(Buffer.isBuffer(binaryPlist)).to.be.true;
       expect(binaryPlist.slice(0, 6).toString()).to.equal('bplist');
@@ -201,5 +202,17 @@ describe('Plist Module', function () {
       const parsedEmptyXml = parseXmlPlist(emptyXmlResult);
       expect(parsedEmptyXml).to.deep.equal({});
     });
+
+    it('should validate that sample data contains emoji and unicode', function () {
+      // Test round-trip with the sample data
+      const binary = createBinaryPlist(expectedPlistObject);
+      const obj = parseBinaryPlist(binary) as Record<string, any>;
+      const xmlResult = createXmlPlist(expectedPlistObject);
+      expect(obj).to.have.property('emoji', '😀');
+      expect(obj).to.have.property('unicode', '测试');
+      // Verify the XML contains the encoded characters
+      expect(xmlResult).to.include('😀');
+      expect(xmlResult).to.include('测试');
+    });
   });
 });