Make regex more specific

Author: steven10a

src/__tests__/unit/checks/pii.test.ts

@@ -43,30 +43,67 @@ describe('pii guardrail', () => {
     await expect(pii({}, '', config)).rejects.toThrow('Text cannot be empty or null');
   });
 
-  it('detects Korean Resident Registration Number (KR_RRN)', async () => {
+  it('detects valid Korean Resident Registration Number (KR_RRN)', async () => {
     const config = PIIConfig.parse({
       entities: [PIIEntity.KR_RRN],
       block: false,
     });
-    const text = 'Korean RRN: 123456-1234567';
+    // Valid format: YYMMDD-GNNNNNN (900101 = Jan 1, 1990, gender digit 1)
+    const text = 'Korean RRN: 900101-1234567';
 
     const result = await pii({}, text, config);
 
     expect(result.tripwireTriggered).toBe(false);
-    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toEqual(['123456-1234567']);
+    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toEqual(['900101-1234567']);
     expect(result.info?.checked_text).toBe('Korean RRN: <KR_RRN>');
   });
 
+  it('detects multiple valid KR_RRN formats', async () => {
+    const config = PIIConfig.parse({
+      entities: [PIIEntity.KR_RRN],
+      block: false,
+    });
+    // Testing different valid date ranges and gender digits (1-4)
+    const text = 'RRNs: 850315-2345678, 001231-3456789, 750628-4123456';
+
+    const result = await pii({}, text, config);
+
+    expect(result.tripwireTriggered).toBe(false);
+    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toHaveLength(3);
+    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toContain('850315-2345678');
+    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toContain('001231-3456789');
+    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toContain('750628-4123456');
+  });
+
+  it('does not detect invalid KR_RRN patterns (false positives)', async () => {
+    const config = PIIConfig.parse({
+      entities: [PIIEntity.KR_RRN],
+      block: false,
+    });
+    // Invalid patterns that should NOT be detected:
+    // - Invalid month (13)
+    // - Invalid day (00, 32)
+    // - Invalid gender digit (0, 5, 9)
+    // - Random tracking numbers
+    const text = 'Invalid: 901301-1234567, 900100-1234567, 900132-1234567, 900101-0234567, 900101-5234567, 123456-7890123';
+
+    const result = await pii({}, text, config);
+
+    expect(result.tripwireTriggered).toBe(false);
+    expect(result.info?.detected_entities).toEqual({});
+    expect(result.info?.checked_text).toBe(text); // No masking should occur
+  });
+
   it('triggers tripwire for KR_RRN when block=true', async () => {
     const config = PIIConfig.parse({
       entities: [PIIEntity.KR_RRN],
       block: true,
     });
-    const text = 'Korean RRN: 123456-1234567';
+    const text = 'Korean RRN: 900101-1234567';
 
     const result = await pii({}, text, config);
 
     expect(result.tripwireTriggered).toBe(true);
-    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toEqual(['123456-1234567']);
+    expect((result.info?.detected_entities as Record<string, string[]>)?.KR_RRN).toEqual(['900101-1234567']);
   });
 });

src/checks/pii.ts

@@ -241,7 +241,8 @@ const DEFAULT_PII_PATTERNS: Record<PIIEntity, RegExp> = {
   [PIIEntity.FI_PERSONAL_IDENTITY_CODE]: /\b\d{6}[+-A]\d{3}[A-Z0-9]\b/g,
 
   // Korea
-  [PIIEntity.KR_RRN]: /\b\d{6}-\d{7}\b/g,
+  // Format: YYMMDD-GNNNNNN where YY=year, MM=month(01-12), DD=day(01-31), G=gender/century(1-4)
+  [PIIEntity.KR_RRN]: /\b\d{2}(0[1-9]|1[0-2])(0[1-9]|[12]\d|3[01])-[1-4]\d{6}\b/g,
 };
 
 /**