[SDS-3138] Add Japan Corporate Number checksum

sds/src/scanner/regex_rule/config.rs

@@ -248,6 +248,7 @@ pub enum SecondaryValidator {
     IbanChecker,
     IrishPpsChecksum,
     ItalianNationalIdChecksum,
+    JapanCorporateNumberChecksum,
     JapanDrivingLicenseChecksum,
     JwtClaimsValidator {
         config: JwtClaimsValidatorConfig,

sds/src/secondary_validation/japan_corporate_number_checksum.rs

@@ -0,0 +1,92 @@
+use crate::secondary_validation::{Validator, get_next_digit};
+
+pub struct JapanCorporateNumberChecksum;
+
+impl Validator for JapanCorporateNumberChecksum {
+    fn is_valid_match(&self, regex_match: &str) -> bool {
+        if regex_match.len() != 13 || !regex_match.bytes().all(|byte| byte.is_ascii_digit()) {
+            return false;
+        }
+
+        let mut chars = regex_match.chars();
+        let Some(check_digit) = get_next_digit(&mut chars) else {
+            return false;
+        };
+        let mut sum = 0;
+        for weight in [2, 1].into_iter().cycle().take(12) {
+            let Some(digit) = get_next_digit(&mut chars) else {
+                return false;
+            };
+            sum += digit * weight;
+        }
+
+        // NTA: https://www.houjin-bangou.nta.go.jp/documents/houjinbangounokousei.pdf
+        check_digit == 9 - sum % 9
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn accepts_official_examples() {
+        // NTA worked example: https://www.houjin-bangou.nta.go.jp/documents/checkdigit.pdf
+        assert!(JapanCorporateNumberChecksum.is_valid_match("8700110005901"));
+        // NTA API sample: https://www.houjin-bangou.nta.go.jp/pc/download/images/k-sample-explain-r0404.pdf
+        assert!(JapanCorporateNumberChecksum.is_valid_match("8040001999013"));
+    }
+
+    #[test]
+    fn handles_every_remainder() {
+        // Synthetic bases: only the units digit varies, giving sums 0 through 8.
+        for (units, check_digit) in (0..=8).zip((1..=9).rev()) {
+            let number = format!("{check_digit}00000000000{units}");
+            assert!(
+                JapanCorporateNumberChecksum.is_valid_match(&number),
+                "{number}"
+            );
+        }
+        // Synthetic base with a tens digit of 1 gives sum 2.
+        assert!(JapanCorporateNumberChecksum.is_valid_match("7000000000010"));
+    }
+
+    #[test]
+    fn rejects_wrong_check_digits() {
+        for check_digit in b'0'..=b'9' {
+            let mut number = *b"8700110005901";
+            number[0] = check_digit;
+            let number = std::str::from_utf8(&number).unwrap();
+            assert_eq!(
+                JapanCorporateNumberChecksum.is_valid_match(number),
+                check_digit == b'8',
+                "{number}"
+            );
+        }
+        assert!(!JapanCorporateNumberChecksum.is_valid_match("8700110005902"));
+    }
+
+    #[test]
+    fn rejects_malformed_input() {
+        for number in [
+            "",
+            "870011000590",
+            "87001100059010",
+            "870011000590A",
+            "8 700110005901",
+            "8-700110005901",
+            " 8700110005901",
+            "8700110005901 ",
+            "８700110005901",
+            "870011000590１",
+            "87001100059é",
+            "1234567890123",
+            "9876543210987",
+        ] {
+            assert!(
+                !JapanCorporateNumberChecksum.is_valid_match(number),
+                "{number}"
+            );
+        }
+    }
+}

sds/src/secondary_validation/mod.rs

@@ -29,6 +29,7 @@ mod iban_checker;
 mod irish_pps_checksum;
 mod iso_7064_checksum;
 mod italian_national_id_checksum;
+mod japan_corporate_number_checksum;
 mod japan_driving_license_checksum;
 pub mod jwt_claims_validator;
 mod jwt_expiration_checker;
@@ -101,6 +102,7 @@ pub use crate::secondary_validation::iso_7064_checksum::{
     Mod97_10checksum, Mod661_26checksum, Mod1271_36Checksum,
 };
 pub use crate::secondary_validation::italian_national_id_checksum::ItalianNationalIdChecksum;
+pub use crate::secondary_validation::japan_corporate_number_checksum::JapanCorporateNumberChecksum;
 pub use crate::secondary_validation::japan_driving_license_checksum::JapanDrivingLicenseChecksum;
 pub use crate::secondary_validation::jwt_claims_validator::JwtClaimsValidator;
 pub use crate::secondary_validation::jwt_expiration_checker::JwtExpirationChecker;
@@ -249,6 +251,9 @@ impl SecondaryValidator {
             SecondaryValidator::IbanChecker => Arc::new(IbanChecker),
             SecondaryValidator::IrishPpsChecksum => Arc::new(IrishPpsChecksum),
             SecondaryValidator::ItalianNationalIdChecksum => Arc::new(ItalianNationalIdChecksum),
+            SecondaryValidator::JapanCorporateNumberChecksum => {
+                Arc::new(JapanCorporateNumberChecksum)
+            }
             SecondaryValidator::JapanDrivingLicenseChecksum => {
                 Arc::new(JapanDrivingLicenseChecksum)
             }