fix: GSM-7 special character size calculation in UCS-2 encoding

Some GSM-7 characters (e.g. ]) requires 14bits to be encoded (so to characters units). When the SMS is encoded using UCS-2, these special characters requires 16 bits, exactly as the normal GSM-7 characters.

Author: vernig

src/libs/EncodedChar.ts

@@ -37,6 +37,10 @@ class EncodedChar {
   }
 
   sizeInBits(): number {
+    if (this.encoding === 'UCS-2' && this.isGSM7) {
+      // GSM characters are always using 16 bits in UCS-2 encoding
+      return 16;
+    }
     const bitsPerUnits = this.encoding === 'GSM-7' ? 7 : 16;
     return bitsPerUnits * this.codeUnits.length;
   }

tests/index.test.js

@@ -143,4 +143,48 @@ describe('GSM-7 Escape Characters', () => {
       expect(segmentedMessage.totalSize).toBe(1223);
     });
   });
-});
+});
+
+describe('One grapheme UCS-2 characters', () => {
+  const testCharacters = ['Á', 'Ú', 'ú', 'ç', 'í', 'Í', 'ó', 'Ó'];
+  testCharacters.forEach((character) => {
+    test(`One segment, 70 characters of "${character}"`, () => {
+      const testMessage = Array(70).fill(character).join('');
+      const segmentedMessage = new SegmentedMessage(testMessage);
+      expect(segmentedMessage.segmentsCount).toBe(1);
+      segmentedMessage.encodedChars.forEach((encodedChar) => {
+        expect(encodedChar.isGSM7).toBe(false);
+      });
+    });
+  });
+
+  testCharacters.forEach((character) => {
+    test(`Two segments, 71 characters of "${character}"`, () => {
+      const testMessage = Array(71).fill(character).join('');
+      const segmentedMessage = new SegmentedMessage(testMessage);
+      expect(segmentedMessage.segmentsCount).toBe(2);
+      segmentedMessage.encodedChars.forEach((encodedChar) => {
+        expect(encodedChar.isGSM7).toBe(false);
+      });
+    });
+  });
+});
+
+describe('Special tests', () => {
+  test('UCS2 message with special GSM characters in one segment', () => {
+    // Issue #18: wrong segmnent calculation using GSM special characters
+    const testMessage = '😀]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]'
+    const segmentedMessage = new SegmentedMessage(testMessage);
+    expect(segmentedMessage.segmentsCount).toBe(1);
+  })
+
+  test('UCS2 message with special GSM characters in two segment', () => {
+    const testMessage = '😀]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]'
+    const segmentedMessage = new SegmentedMessage(testMessage);
+    expect(segmentedMessage.segmentsCount).toBe(2);
+  }) 
+
+
+
+  ""
+})