Harden UTF8 length test cases

bluesky-social · Dec 10, 2024 · 8f9becc · 8f9becc
1 parent 57c58f6
commit 8f9becc
Showing 1 changed file with 85 additions and 5 deletions.
diff --git a/packages/lexicon/tests/general.test.ts b/packages/lexicon/tests/general.test.ts
@@ -567,26 +567,106 @@ describe('Record validation', () => {
   })
 
   it('Applies string length constraint', () => {
+    // Shorter than two UTF8 characters
+    expect(() =>
+      lex.assertValidRecord('com.example.stringLength', {
+        $type: 'com.example.stringLength',
+        string: '',
+      }),
+    ).toThrow('Record/string must not be shorter than 2 characters')
+    expect(() =>
+      lex.assertValidRecord('com.example.stringLength', {
+        $type: 'com.example.stringLength',
+        string: 'a',
+      }),
+    ).toThrow('Record/string must not be shorter than 2 characters')
+
+    // Two to four UTF8 characters
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: 'ab',
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: '\u0301', // Combining acute accent (2 bytes)
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: 'a\u0301', // 'a' + combining acute accent (1 + 2 bytes = 3 bytes)
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: 'aé', // 'a' (1 byte) + 'é' (2 bytes) = 3 bytes
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: 'abc',
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: '一', // CJK character (3 bytes)
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: '\uD83D', // Unpaired high surrogate (3 bytes)
+    })
     lex.assertValidRecord('com.example.stringLength', {
       $type: 'com.example.stringLength',
-      string: '123',
+      string: 'abcd',
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: 'éé', // 'é' + 'é' (2 + 2 bytes = 4 bytes)
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: 'aaé', // 1 + 1 + 2 = 4 bytes
+    })
+    lex.assertValidRecord('com.example.stringLength', {
+      $type: 'com.example.stringLength',
+      string: '👋', // 4 bytes
     })
+
     expect(() =>
       lex.assertValidRecord('com.example.stringLength', {
         $type: 'com.example.stringLength',
-        string: '1',
+        string: 'abcde',
       }),
-    ).toThrow('Record/string must not be shorter than 2 characters')
+    ).toThrow('Record/string must not be longer than 4 characters')
     expect(() =>
       lex.assertValidRecord('com.example.stringLength', {
         $type: 'com.example.stringLength',
-        string: '12345',
+        string: 'a\u0301\u0301', // 1 + (2 * 2) = 5 bytes
       }),
     ).toThrow('Record/string must not be longer than 4 characters')
     expect(() =>
       lex.assertValidRecord('com.example.stringLength', {
         $type: 'com.example.stringLength',
-        string: '👨‍👩‍👧‍👧',
+        string: '\uD83D\uD83D', // Two unpaired high surrogates (3 * 2 = 6 bytes)
+      }),
+    ).toThrow('Record/string must not be longer than 4 characters')
+    expect(() =>
+      lex.assertValidRecord('com.example.stringLength', {
+        $type: 'com.example.stringLength',
+        string: 'ééé', // 2 + 2 + 2 bytes = 6 bytes
+      }),
+    ).toThrow('Record/string must not be longer than 4 characters')
+    expect(() =>
+      lex.assertValidRecord('com.example.stringLength', {
+        $type: 'com.example.stringLength',
+        string: '👋a', // 4 + 1 bytes = 5 bytes
+      }),
+    ).toThrow('Record/string must not be longer than 4 characters')
+    expect(() =>
+      lex.assertValidRecord('com.example.stringLength', {
+        $type: 'com.example.stringLength',
+        string: '👨👨', // 4 + 4 = 8 bytes
+      }),
+    ).toThrow('Record/string must not be longer than 4 characters')
+    expect(() =>
+      lex.assertValidRecord('com.example.stringLength', {
+        $type: 'com.example.stringLength',
+        string: '👨‍👩‍👧‍👧', // 4 emojis × 4 bytes + 3 ZWJs × 3 bytes = 25 bytes
       }),
     ).toThrow('Record/string must not be longer than 4 characters')
   })