adds allowed phrases to transform function

internal/ee/transformers/functions/functions.go

@@ -6,6 +6,7 @@ import (
 
 	mgmtv1alpha1 "github.com/nucleuscloud/neosync/backend/gen/go/protos/mgmt/v1alpha1"
 	presidioapi "github.com/nucleuscloud/neosync/internal/ee/presidio"
+	transformer_utils "github.com/nucleuscloud/neosync/worker/pkg/benthos/transformers/utils"
 )
 
 var (
@@ -40,6 +41,8 @@ func TransformPiiText(
 		return "", fmt.Errorf("received non-200 response from analyzer: %s %d %s", analyzeResp.Status(), analyzeResp.StatusCode(), string(analyzeResp.Body))
 	}
 
+	analysisResults := removeAllowedPhrases(*analyzeResp.JSON200, value, config.GetAllowedPhrases())
+
 	defaultAnon, ok, err := getDefaultAnonymizer(config.GetDefaultAnonymizer())
 	if err != nil {
 		return "", fmt.Errorf("unable to build default anonymizer: %w", err)
@@ -49,7 +52,7 @@ func TransformPiiText(
 		anonymizers["DEFAULT"] = *defaultAnon
 	}
 	anonResp, err := anonymizeClient.PostAnonymizeWithResponse(ctx, presidioapi.AnonymizeRequest{
-		AnalyzerResults: presidioapi.ToAnonymizeRecognizerResults(*analyzeResp.JSON200),
+		AnalyzerResults: presidioapi.ToAnonymizeRecognizerResults(analysisResults),
 		Text:            value,
 		Anonymizers:     &anonymizers,
 	})
@@ -63,6 +66,28 @@ func TransformPiiText(
 	return *anonResp.JSON200.Text, nil
 }
 
+func removeAllowedPhrases(
+	results []presidioapi.RecognizerResultWithAnaysisExplanation,
+	text string,
+	allowedPhrases []string,
+) []presidioapi.RecognizerResultWithAnaysisExplanation {
+	output := []presidioapi.RecognizerResultWithAnaysisExplanation{}
+	uniquePhrases := transformer_utils.ToSet(allowedPhrases)
+	textLen := len(text)
+	for _, result := range results {
+		if result.Start < 0 || result.End > textLen {
+			continue // Skip invalid ranges
+		}
+
+		phrase := text[result.Start:result.End]
+		if _, ok := uniquePhrases[phrase]; !ok {
+			output = append(output, result)
+		}
+	}
+
+	return output
+}
+
 func buildAdhocRecognizers(dtos []*mgmtv1alpha1.PiiDenyRecognizer) []presidioapi.PatternRecognizer {
 	output := []presidioapi.PatternRecognizer{}
 	for _, dto := range dtos {

internal/ee/transformers/functions/functions_test.go

@@ -43,6 +43,60 @@ func Test_TransformPiiText(t *testing.T) {
 	})
 }
 
+func Test_removeAllowedPhrases(t *testing.T) {
+	t.Run("exact", func(t *testing.T) {
+		actual := removeAllowedPhrases(
+			[]presidioapi.RecognizerResultWithAnaysisExplanation{
+				{
+					Start:               11,
+					End:                 24,
+					Score:               0.85,
+					EntityType:          "person",
+					RecognitionMetadata: &presidioapi.RecognizedMetadata{},
+				},
+			},
+			"My name is Inigo Montoya prepare to die",
+			[]string{"Inigo Montoya"},
+		)
+		require.Empty(t, actual)
+	})
+
+	t.Run("invalid_range_skip", func(t *testing.T) {
+		actual := removeAllowedPhrases(
+			[]presidioapi.RecognizerResultWithAnaysisExplanation{
+				{
+					Start:               500,
+					End:                 600,
+					Score:               0.85,
+					EntityType:          "person",
+					RecognitionMetadata: &presidioapi.RecognizedMetadata{},
+				},
+			},
+			"My name is Inigo Montoya prepare to die",
+			[]string{"Inigo Montoya"},
+		)
+		require.Empty(t, actual)
+	})
+
+	t.Run("not_found", func(t *testing.T) {
+		input := []presidioapi.RecognizerResultWithAnaysisExplanation{
+			{
+				Start:               11,
+				End:                 24,
+				Score:               0.85,
+				EntityType:          "person",
+				RecognitionMetadata: &presidioapi.RecognizedMetadata{},
+			},
+		}
+		actual := removeAllowedPhrases(
+			input,
+			"My name is Inigo Montoya prepare to die",
+			[]string{"Inigo"},
+		)
+		require.Equal(t, input, actual)
+	})
+}
+
 // func Test_TransformBloblPiiText(t *testing.T) {
 // 	env := bloblang.NewEmptyEnvironment()
 // 	mockanalyze := presidioapi.NewMockAnalyzeInterface(t)