search: strip "-" from GovernmentIDs for similarity

moov-io · Jan 16, 2025 · 22b54ed · 22b54ed
1 parent 8b061bd
commit 22b54ed
Show file tree

Hide file tree

Showing 3 changed files with 64 additions and 3 deletions.
diff --git a/pkg/ofac/mapper_business_test.go b/pkg/ofac/mapper_business_test.go
@@ -98,6 +98,20 @@ func TestMapperBusiness__FromSource(t *testing.T) {
 		require.Equal(t, expectedContact, found.Contact)
 	})
 
+	t.Run("44525", func(t *testing.T) {
+		found := ofactest.FindEntity(t, "44525")
+		require.Equal(t, "BEL-KAP-STEEL LLC", found.Name)
+
+		t.Logf("%#v", found.Business)
+
+		expectedGovernmentIDs := []search.GovernmentID{
+			{Type: "tax-id", Country: "United States", Identifier: "52-2083095"},
+			{Type: "business-registration", Country: "Connecticut", Identifier: "0582030"},
+			{Type: "business-registration", Country: "Florida", Identifier: "M99000000961"},
+		}
+		require.ElementsMatch(t, expectedGovernmentIDs, found.Business.GovernmentIDs)
+	})
+
 	t.Run("50544", func(t *testing.T) {
 		found := ofactest.FindEntity(t, "50544")
 

diff --git a/pkg/search/similarity_exact.go b/pkg/search/similarity_exact.go
@@ -38,6 +38,10 @@ func compareExactIdentifiers[Q any, I any](w io.Writer, query Entity[Q], index E
 	}
 }
 
+func normalizeIdentifier(id string) string {
+	return strings.ReplaceAll(id, "-", "")
+}
+
 // comparePersonExactIDs checks exact matches for Person-specific identifiers
 func comparePersonExactIDs(w io.Writer, query *Person, index *Person, weight float64) scorePiece {
 	if query == nil || index == nil {
@@ -57,7 +61,7 @@ func comparePersonExactIDs(w io.Writer, query *Person, index *Person, weight flo
 			for _, iID := range index.GovernmentIDs {
 				if strings.EqualFold(string(qID.Type), string(iID.Type)) &&
 					strings.EqualFold(qID.Country, iID.Country) &&
-					strings.EqualFold(qID.Identifier, iID.Identifier) {
+					strings.EqualFold(normalizeIdentifier(qID.Identifier), normalizeIdentifier(iID.Identifier)) {
 					score += 15.0
 					hasMatch = true
 					goto GovIDDone // Break both loops on first match
@@ -103,7 +107,7 @@ func compareBusinessExactIDs(w io.Writer, query *Business, index *Business, weig
 				// Exact match on all identifier fields
 				if strings.EqualFold(string(qID.Type), string(iID.Type)) &&
 					strings.EqualFold(qID.Country, iID.Country) &&
-					strings.EqualFold(qID.Identifier, iID.Identifier) {
+					strings.EqualFold(normalizeIdentifier(qID.Identifier), normalizeIdentifier(iID.Identifier)) {
 					score += 15.0
 					hasMatch = true
 					goto IdentifierDone
@@ -149,7 +153,7 @@ func compareOrgExactIDs(w io.Writer, query *Organization, index *Organization, w
 				// Exact match on all identifier fields
 				if strings.EqualFold(string(qID.Type), string(iID.Type)) &&
 					strings.EqualFold(qID.Country, iID.Country) &&
-					strings.EqualFold(qID.Identifier, iID.Identifier) {
+					strings.EqualFold(normalizeIdentifier(qID.Identifier), normalizeIdentifier(iID.Identifier)) {
 					score += 15.0
 					hasMatch = true
 					goto IdentifierDone

diff --git a/pkg/search/similarity_exact_test.go b/pkg/search/similarity_exact_test.go
@@ -0,0 +1,43 @@
+package search
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestCompareBusinessExactIDs(t *testing.T) {
+	cases := []struct {
+		query, index Business
+		expected     scorePiece
+	}{
+		{
+			query: Business{
+				GovernmentIDs: []GovernmentID{
+					{Type: GovernmentIDTax, Country: "United States", Identifier: "522083095"},
+				},
+			},
+			index: Business{
+				GovernmentIDs: []GovernmentID{
+					{Type: GovernmentIDTax, Country: "United States", Identifier: "52-2083095"},
+				},
+			},
+			expected: scorePiece{
+				score:          1.0,
+				weight:         50,
+				matched:        true,
+				required:       true,
+				exact:          true,
+				fieldsCompared: 1,
+				pieceType:      "identifiers",
+			},
+		},
+	}
+	for _, tc := range cases {
+		t.Run(fmt.Sprintf("%#v", tc.query.GovernmentIDs), func(t *testing.T) {
+			got := compareBusinessExactIDs(nil, &tc.query, &tc.index, criticalIdWeight)
+			require.Equal(t, tc.expected, got)
+		})
+	}
+}