Codex SG tax id validation

CHANGELOG.md

@@ -8,6 +8,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/) and this p
 
 ### Added
 
+- `sg`: UEN check-character validation now covers business, local-company, and
+  other-entity UENs, for both UEN identities and tax identity codes.
 - `net`: added `SandboxAuthorities` (defaulting to `lookup.sandbox.gobl.org`)
   and `WithSandbox`. Sandbox and live trust lists remain separate.
 
@@ -30,6 +32,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/) and this p
 
 ### Fixed
 
+- `sg`: tax identity codes accept IRAS-assigned GST registration numbers ending
+  in a digit (for example, `M201189853`).
 - `head`: `SignedPayload` and `Header.Verify` now return an error for `null`
   signature entries instead of panicking.
 

examples/sg/invoice-gst-number.json

@@ -9,7 +9,7 @@
       "name": "EXAMPLE SUPPLIER",
       "tax_id": {
         "country": "SG",
-        "code": "201312345A"
+        "code": "201312345N"
       },
       "identities": [
         {
@@ -51,6 +51,4 @@
         ]
       }
     ]
-
   }
-

examples/sg/invoice-receipt.json

@@ -9,7 +9,7 @@
       "name": "EXAMPLE SUPPLIER",
       "tax_id": {
         "country": "SG",
-        "code": "201312345A"
+        "code": "201312345N"
       }
     },
     "lines": [
@@ -28,5 +28,4 @@
         ]
       }
     ]
-
-  }
+  }

examples/sg/invoice-simple.json

@@ -12,7 +12,7 @@
       "name": "EXAMPLE SUPPLIER",
       "tax_id": {
         "country": "SG",
-        "code": "201312345A"
+        "code": "201312345N"
       },
       "addresses": [
         {
@@ -38,5 +38,4 @@
         ]
       }
     ]
-
-  }
+  }

examples/sg/invoice.json

@@ -9,7 +9,7 @@
       "name": "EXAMPLE SUPPLIER",
       "tax_id": {
         "country": "SG",
-        "code": "201312345A"
+        "code": "201312345N"
       },
       "addresses": [
         {
@@ -45,6 +45,4 @@
         ]
       }
     ]
-
   }
-

examples/sg/out/invoice-gst-number.json

@@ -4,7 +4,7 @@
 		"uuid": "8a51fd30-2a27-11ee-be56-0242ac120002",
 		"dig": {
 			"alg": "sha256",
-			"val": "74947258d585ebd9c7aa85faabddd006d2f7de405d5f282c1216b67af40f128e"
+			"val": "e29c5f4e3e6edc881569e43ec910d2e83af2efafe1da9ef5c63785e290bd1284"
 		}
 	},
 	"doc": {
@@ -19,7 +19,7 @@
 			"name": "EXAMPLE SUPPLIER",
 			"tax_id": {
 				"country": "SG",
-				"code": "201312345A"
+				"code": "201312345N"
 			},
 			"identities": [
 				{

examples/sg/out/invoice-receipt.json

@@ -4,7 +4,7 @@
 		"uuid": "8a51fd30-2a27-11ee-be56-0242ac120002",
 		"dig": {
 			"alg": "sha256",
-			"val": "6137fe1dff83bf55a48085f255bbc8059b2f9e3d4f1082483fa1a89a582cd00b"
+			"val": "6e84b526f70c2a8ca4b4d36b3974bb5899a4f8b8c6d1a3d127e331f8d22ad0b3"
 		}
 	},
 	"doc": {
@@ -19,7 +19,7 @@
 			"name": "EXAMPLE SUPPLIER",
 			"tax_id": {
 				"country": "SG",
-				"code": "201312345A"
+				"code": "201312345N"
 			}
 		},
 		"lines": [

examples/sg/out/invoice-simple.json

@@ -4,7 +4,7 @@
 		"uuid": "8a51fd30-2a27-11ee-be56-0242ac120002",
 		"dig": {
 			"alg": "sha256",
-			"val": "e2240ad7bef4e9eac7ac3bee35a3119e120c8d8f6b2c40017cd491771a021030"
+			"val": "4eb9444cb74ab5f2b7063083e20365a954eeae776ee76b2b7943a3db051b092d"
 		}
 	},
 	"doc": {
@@ -23,7 +23,7 @@
 			"name": "EXAMPLE SUPPLIER",
 			"tax_id": {
 				"country": "SG",
-				"code": "201312345A"
+				"code": "201312345N"
 			},
 			"addresses": [
 				{

examples/sg/out/invoice.json

@@ -4,7 +4,7 @@
 		"uuid": "8a51fd30-2a27-11ee-be56-0242ac120002",
 		"dig": {
 			"alg": "sha256",
-			"val": "370f9f130b445765ea0d4d9eec22c551bde0760220878a2fa0b11816bd2068cd"
+			"val": "1cbbba2a8f2fd6c9e4466fe772ad3cefe6ea2ae1fceb109fa222ad5e7d3aa079"
 		}
 	},
 	"doc": {
@@ -19,7 +19,7 @@
 			"name": "EXAMPLE SUPPLIER",
 			"tax_id": {
 				"country": "SG",
-				"code": "201312345A"
+				"code": "201312345N"
 			},
 			"addresses": [
 				{

regimes/sg/bill_invoices_test.go

@@ -72,7 +72,7 @@ func TestValidInvoiceWithUEN(t *testing.T) {
 	inv.Supplier.Identities = []*org.Identity{
 		{
 			Type: sg.IdentityTypeUEN,
-			Code: "199912345A",
+			Code: "197401143C",
 		},
 	}
 	require.NoError(t, inv.Calculate())

regimes/sg/org_identities.go

@@ -22,13 +22,33 @@ const (
 )
 
 var (
-	regexpsUENIdentities = []*regexp.Regexp{
-		regexp.MustCompile(`^(19[0-9]{2}|20[0-9]{2})\d{5}[A-Z]$`), // UEN (ROC)
-		regexp.MustCompile(`^\d{8}[A-Z]$`),                        // UEN (ROB)
-		regexp.MustCompile(`^[TS]\d{2}[A-Z]{2}\d{4}[A-Z]$`),       // UEN (Others)
+	regexpUENBusiness     = regexp.MustCompile(`^\d{8}[A-Z]$`)
+	regexpUENLocalCompany = regexp.MustCompile(`^(19\d{2}|20\d{2})\d{5}[A-Z]$`)
+	regexpUENOther        = regexp.MustCompile(`^[RST]\d{2}[A-Z]{2}\d{4}[A-Z]$`)
+
+	// Entity types assigned to UENs for bodies other than businesses and
+	// local companies. See https://www.uen.gov.sg/ueninternet/faces/pages/admin/aboutUEN.jspx.
+	uenOtherEntityTypes = map[string]struct{}{
+		"CC": {}, "CD": {}, "CH": {}, "CL": {}, "CM": {}, "CP": {}, "CS": {}, "CX": {},
+		"DP": {}, "FB": {}, "FC": {}, "FM": {}, "FN": {}, "GA": {}, "GB": {}, "GS": {},
+		"HS": {}, "LL": {}, "LP": {}, "MB": {}, "MC": {}, "MD": {}, "MH": {}, "MM": {},
+		"MQ": {}, "NB": {}, "NR": {}, "PA": {}, "PB": {}, "PF": {}, "RF": {}, "RP": {},
+		"SM": {}, "SS": {}, "TC": {}, "TU": {}, "VH": {}, "XL": {},
 	}
 )
 
+const (
+	uenBusinessCheckAlphabet     = "XMKECAWLJDB"
+	uenLocalCompanyCheckAlphabet = "ZKCMDNERGWH"
+	uenOtherCheckAlphabet        = "ABCDEFGHJKLMNPQRSTUVWX0123456789"
+)
+
+var (
+	uenBusinessWeights     = []int{10, 4, 9, 3, 8, 2, 7, 1}
+	uenLocalCompanyWeights = []int{10, 8, 6, 4, 9, 7, 5, 3, 1}
+	uenOtherWeights        = []int{4, 3, 5, 3, 10, 2, 2, 5, 7}
+)
+
 var identityDefinitions = []*cbc.Definition{
 	{
 		Code: IdentityTypeUEN,
@@ -76,13 +96,38 @@ func orgIdentityCheckUEN(value any) bool {
 	if !ok || code == "" {
 		return false
 	}
-	val := code.String()
-	match := false
-	for _, re := range regexpsUENIdentities {
-		if re.MatchString(val) {
-			match = true
-			break
-		}
+	return validateUENCode(code.String())
+}
+
+// validateUENCode verifies a UEN's shape and its check character. The check
+// character tables are the community reverse-engineered versions used by
+// python-stdnum; ACRA does not publish the algorithm.
+func validateUENCode(code string) bool {
+	switch {
+	case regexpUENBusiness.MatchString(code):
+		return code[8] == uenDigitCheck(code, uenBusinessWeights, uenBusinessCheckAlphabet)
+	case regexpUENLocalCompany.MatchString(code):
+		return code[9] == uenDigitCheck(code, uenLocalCompanyWeights, uenLocalCompanyCheckAlphabet)
+	case regexpUENOther.MatchString(code):
+		_, knownType := uenOtherEntityTypes[code[3:5]]
+		return knownType && code[9] == uenOtherCheck(code)
+	default:
+		return false
+	}
+}
+
+func uenDigitCheck(code string, weights []int, alphabet string) byte {
+	sum := 0
+	for i, weight := range weights {
+		sum += int(code[i]-'0') * weight
+	}
+	return alphabet[sum%len(alphabet)]
+}
+
+func uenOtherCheck(code string) byte {
+	sum := 0
+	for i, weight := range uenOtherWeights {
+		sum += strings.IndexByte(uenOtherCheckAlphabet, code[i]) * weight
 	}
-	return match
+	return uenOtherCheckAlphabet[(sum-5)%11]
 }

regimes/sg/org_identities_test.go

@@ -35,14 +35,19 @@ func TestValidateUENIdentity(t *testing.T) {
 		code string
 		err  bool
 	}{
-		{name: "UEN (ROC)", code: "199912345A", err: false},
-		{name: "UEN (ROB)", code: "12345678A", err: false},
-		{name: "UEN (Others)", code: "T12AB1234A", err: false},
+		{name: "UEN (ROC)", code: "197401143C", err: false},
+		{name: "UEN (ROB)", code: "00192200M", err: false},
+		{name: "UEN (Others)", code: "T08GB0020K", err: false},
+		{name: "UEN (Others, 1800s)", code: "R00LL0001F", err: false},
 		{name: "NIRC/FIN", code: "S1234567A", err: true},
 		{name: "Invalid short", code: "1234567A", err: true},
 		{name: "Invalid UEN (ROC)", code: "2199123456", err: true},
 		{name: "Invalid UEN (ROB)", code: "1234567A", err: true},
 		{name: "Invalid UEN (Others)", code: "T12A1234A", err: true},
+		{name: "Unknown UEN entity type", code: "T12AB1234D", err: true},
+		{name: "Invalid UEN (ROB) check character", code: "00192200A", err: true},
+		{name: "Invalid UEN (ROC) check character", code: "197401143A", err: true},
+		{name: "Invalid UEN (Others) check character", code: "T08GB0020A", err: true},
 		{name: "Empty code", code: "", err: true},
 	}
 

regimes/sg/tax_identity.go

@@ -1,7 +1,6 @@
 package sg
 
 import (
-	"errors"
 	"regexp"
 
 	"github.com/invopop/gobl/cbc"
@@ -13,15 +12,9 @@ import (
 // Reference: https://mytax.iras.gov.sg/ESVWeb/default.aspx?target=GSTListingSearch
 // Reference: https://www.oecd.org/content/dam/oecd/en/topics/policy-issue-focus/aeoi/singapore-tin.pdf
 // Reference:https://www.mof.gov.sg/docs/default-source/default-document-library/news-and-publications/press-releases/annexe060808.pdf?sfvrsn=4ee26b50_2
-// Singapore’s tax authority does not publish a public checksum algorithm for UEN or GST numbers.
-// Indeed, IRAS directs users to verify UENs via the official portal
-
-// regexpsGSTCode uses the UEN identities as a base and adds the GST format used
-// for international companies.
-var regexpsGSTCode = append(
-	regexpsUENIdentities,
-	regexp.MustCompile(`^M[A-Z0-9]\d{7}[A-Z]$`),
-)
+// IRAS-assigned GST registration numbers have no public checksum algorithm.
+// The final character may be either a letter or a digit.
+var regexpGSTNumber = regexp.MustCompile(`^M[A-Z0-9]\d{7}[A-Z0-9]$`)
 
 func taxIdentityRules() *rules.Set {
 	return rules.For(new(tax.Identity),
@@ -40,23 +33,6 @@ func isValidTaxIdentityCode(value any) bool {
 	if !ok || code == "" {
 		return false
 	}
-	return validateTaxCode(code) == nil
-}
-
-func validateTaxCode(code cbc.Code) error {
-	if code == "" {
-		return nil
-	}
-	val := code.String()
-	match := false
-	for _, re := range regexpsGSTCode {
-		if re.MatchString(val) {
-			match = true
-			break
-		}
-	}
-	if !match {
-		return errors.New("invalid format")
-	}
-	return nil
+	text := code.String()
+	return regexpGSTNumber.MatchString(text) || validateUENCode(text)
 }

regimes/sg/tax_identity_test.go

@@ -20,19 +20,21 @@ func TestTaxIdentityRules(t *testing.T) {
 		{name: "sole proprietorship GST", code: "MR2345678A"},
 		{name: "overseas vendor GST", code: "MB2345678A"},
 		{name: "overseas vendor GST 2", code: "MX2345678A"},
+		{name: "IRAS-assigned GST with digit suffix", code: "M201189853"},
 		{name: "invalid GST short", code: "M91234567", err: "IDENTITY-01"},
 		{name: "invalid GST long", code: "M91234567XA", err: "IDENTITY-01"},
 		{name: "invalid GST no M", code: "912345678X", err: "IDENTITY-01"},
-		{name: "invalid GST no end letter", code: "M912345678", err: "IDENTITY-01"},
 		// UEN identities also
-		{name: "UEN (ROC)", code: "199912345A"},
-		{name: "UEN (ROB)", code: "12345678A"},
-		{name: "UEN (Others)", code: "T12AB1234A"},
+		{name: "UEN (ROC)", code: "197401143C"},
+		{name: "UEN (ROB)", code: "00192200M"},
+		{name: "UEN (Others)", code: "T08GB0020K"},
 		{name: "NIRC/FIN", code: "S1234567A", err: "IDENTITY-01"},
 		{name: "Invalid short", code: "1234567A", err: "IDENTITY-01"},
 		{name: "Invalid UEN (ROC)", code: "2199123456", err: "IDENTITY-01"},
 		{name: "Invalid UEN (ROB)", code: "1234567A", err: "IDENTITY-01"},
 		{name: "Invalid UEN (Others)", code: "T12A1234A", err: "IDENTITY-01"},
+		{name: "Unknown UEN entity type", code: "T12AB1234D", err: "IDENTITY-01"},
+		{name: "Invalid UEN check character", code: "197401143A", err: "IDENTITY-01"},
 	}
 
 	for _, tt := range tests {