refactor: modularize codebase and add 404 unit tests (#13)

compliance-agent/src/pipeline/patterns.rs

@@ -256,3 +256,159 @@ fn walkdir(path: &Path) -> Result<Vec<walkdir::DirEntry>, CoreError> {
 
     Ok(entries)
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    // --- compile_regex tests ---
+
+    #[test]
+    fn compile_regex_valid_pattern() {
+        let re = compile_regex(r"\bfoo\b");
+        assert!(re.is_match("hello foo bar"));
+        assert!(!re.is_match("foobar"));
+    }
+
+    #[test]
+    fn compile_regex_invalid_pattern_returns_fallback() {
+        // An invalid regex should return the fallback "^$" that only matches empty strings
+        let re = compile_regex(r"[invalid");
+        assert!(re.is_match(""));
+        assert!(!re.is_match("anything"));
+    }
+
+    // --- GDPR pattern tests ---
+
+    #[test]
+    fn gdpr_pii_logging_matches() {
+        let scanner = GdprPatternScanner::new();
+        let pattern = &scanner.patterns[0]; // gdpr-pii-logging
+                                            // Regex: (log|print|console\.|logger\.|tracing::)\s*[\.(].*\b(pii_keyword)\b
+        assert!(pattern.pattern.is_match("console.log(email)"));
+        assert!(pattern.pattern.is_match("console.log(user.ssn)"));
+        assert!(pattern.pattern.is_match("print(phone_number)"));
+        assert!(pattern.pattern.is_match("tracing::(ip_addr)"));
+        assert!(pattern.pattern.is_match("log.debug(credit_card)"));
+    }
+
+    #[test]
+    fn gdpr_pii_logging_no_false_positive() {
+        let scanner = GdprPatternScanner::new();
+        let pattern = &scanner.patterns[0];
+        // Regular logging without PII fields should not match
+        assert!(!pattern
+            .pattern
+            .is_match("logger.info(\"request completed\")"));
+        assert!(!pattern.pattern.is_match("let email = user.email;"));
+    }
+
+    #[test]
+    fn gdpr_no_consent_matches() {
+        let scanner = GdprPatternScanner::new();
+        let pattern = &scanner.patterns[1]; // gdpr-no-consent
+        assert!(pattern.pattern.is_match("collect personal data"));
+        assert!(pattern.pattern.is_match("store user_data in db"));
+        assert!(pattern.pattern.is_match("save pii to disk"));
+    }
+
+    #[test]
+    fn gdpr_user_model_matches() {
+        let scanner = GdprPatternScanner::new();
+        let pattern = &scanner.patterns[2]; // gdpr-no-delete-endpoint
+        assert!(pattern.pattern.is_match("struct User {"));
+        assert!(pattern.pattern.is_match("class User(Model):"));
+    }
+
+    #[test]
+    fn gdpr_hardcoded_retention_matches() {
+        let scanner = GdprPatternScanner::new();
+        let pattern = &scanner.patterns[3]; // gdpr-hardcoded-retention
+        assert!(pattern.pattern.is_match("retention = 30"));
+        assert!(pattern.pattern.is_match("ttl: 3600"));
+        assert!(pattern.pattern.is_match("expire = 86400"));
+    }
+
+    // --- OAuth pattern tests ---
+
+    #[test]
+    fn oauth_implicit_grant_matches() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[0]; // oauth-implicit-grant
+        assert!(pattern.pattern.is_match("response_type = \"token\""));
+        assert!(pattern.pattern.is_match("grant_type: implicit"));
+        assert!(pattern.pattern.is_match("response_type='token'"));
+    }
+
+    #[test]
+    fn oauth_implicit_grant_no_false_positive() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[0];
+        assert!(!pattern.pattern.is_match("response_type = \"code\""));
+        assert!(!pattern.pattern.is_match("grant_type: authorization_code"));
+    }
+
+    #[test]
+    fn oauth_authorization_code_matches() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[1]; // oauth-missing-pkce
+        assert!(pattern.pattern.is_match("uses authorization_code flow"));
+        assert!(pattern.pattern.is_match("authorization code grant"));
+    }
+
+    #[test]
+    fn oauth_token_localstorage_matches() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[2]; // oauth-token-localstorage
+        assert!(pattern
+            .pattern
+            .is_match("localStorage.setItem('access_token', tok)"));
+        assert!(pattern
+            .pattern
+            .is_match("localStorage.getItem(\"refresh_token\")"));
+    }
+
+    #[test]
+    fn oauth_token_localstorage_no_false_positive() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[2];
+        assert!(!pattern
+            .pattern
+            .is_match("localStorage.setItem('theme', 'dark')"));
+        assert!(!pattern
+            .pattern
+            .is_match("sessionStorage.setItem('token', t)"));
+    }
+
+    #[test]
+    fn oauth_token_url_matches() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[3]; // oauth-token-url
+        assert!(pattern.pattern.is_match("access_token = build_url(query)"));
+        assert!(pattern.pattern.is_match("bearer = url.param"));
+    }
+
+    // --- Pattern rule file extension filtering ---
+
+    #[test]
+    fn gdpr_patterns_cover_common_languages() {
+        let scanner = GdprPatternScanner::new();
+        for pattern in &scanner.patterns {
+            assert!(
+                pattern.file_extensions.contains(&"rs".to_string()),
+                "Pattern {} should cover .rs files",
+                pattern.id
+            );
+        }
+    }
+
+    #[test]
+    fn oauth_localstorage_only_js_ts() {
+        let scanner = OAuthPatternScanner::new();
+        let pattern = &scanner.patterns[2]; // oauth-token-localstorage
+        assert!(pattern.file_extensions.contains(&"js".to_string()));
+        assert!(pattern.file_extensions.contains(&"ts".to_string()));
+        assert!(!pattern.file_extensions.contains(&"rs".to_string()));
+        assert!(!pattern.file_extensions.contains(&"py".to_string()));
+    }
+}