Custom search attributes validation per store (#4655)

Author: rodrigozhou

common/persistence/visibility/manager/visibility_manager.go

@@ -47,6 +47,7 @@ type (
 		GetStoreNames() []string
 		HasStoreName(stName string) bool
 		GetIndexName() string
+		ValidateCustomSearchAttributes(searchAttributes map[string]any) (map[string]any, error)
 
 		// Write APIs.
 		RecordWorkflowExecutionStarted(ctx context.Context, request *RecordWorkflowExecutionStartedRequest) error

common/persistence/visibility/manager/visibility_manager_mock.go

@@ -95,6 +95,10 @@ var _ store.VisibilityStore = (*visibilityStore)(nil)
 var (
 	errUnexpectedJSONFieldType = errors.New("unexpected JSON field type")
 
+	minTime         = time.Unix(0, 0).UTC()
+	maxTime         = time.Unix(0, math.MaxInt64).UTC()
+	maxStringLength = 32766
+
 	// Default sorter uses the sorting order defined in the index template.
 	// It is indirectly built so buildPaginationQuery can have access to
 	// the fields names to build the page query from the token.
@@ -164,6 +168,44 @@ func (s *visibilityStore) GetIndexName() string {
 	return s.index
 }
 
+func (s *visibilityStore) ValidateCustomSearchAttributes(
+	searchAttributes map[string]any,
+) (map[string]any, error) {
+	validatedSearchAttributes := make(map[string]any, len(searchAttributes))
+	var invalidValueErrs []error
+	for saName, saValue := range searchAttributes {
+		var err error
+		switch value := saValue.(type) {
+		case time.Time:
+			err = validateDatetime(value)
+		case []time.Time:
+			for _, item := range value {
+				if err = validateDatetime(item); err != nil {
+					break
+				}
+			}
+		case string:
+			err = validateString(value)
+		case []string:
+			for _, item := range value {
+				if err = validateString(item); err != nil {
+					break
+				}
+			}
+		}
+		if err != nil {
+			invalidValueErrs = append(invalidValueErrs, err)
+			continue
+		}
+		validatedSearchAttributes[saName] = saValue
+	}
+	var retError error
+	if len(invalidValueErrs) > 0 {
+		retError = store.NewVisibilityStoreInvalidValuesError(invalidValueErrs)
+	}
+	return validatedSearchAttributes, retError
+}
+
 func (s *visibilityStore) RecordWorkflowExecutionStarted(
 	ctx context.Context,
 	request *store.InternalRecordWorkflowExecutionStartedRequest,
@@ -941,6 +983,14 @@ func (s *visibilityStore) generateESDoc(request *store.InternalVisibilityRequest
 		s.metricsHandler.Counter(metrics.ElasticsearchDocumentGenerateFailuresCount.GetMetricName()).Record(1)
 		return nil, serviceerror.NewInternal(fmt.Sprintf("Unable to decode search attributes: %v", err))
 	}
+	// This is to prevent existing tasks to fail indefinitely.
+	// If it's only invalid values error, then silently continue without them.
+	searchAttributes, err = s.ValidateCustomSearchAttributes(searchAttributes)
+	if err != nil {
+		if _, ok := err.(*store.VisibilityStoreInvalidValuesError); !ok {
+			return nil, err
+		}
+	}
 	for saName, saValue := range searchAttributes {
 		if saValue == nil {
 			// If search attribute value is `nil`, it means that it shouldn't be added to the document.
@@ -1321,3 +1371,25 @@ func parsePageTokenValue(
 		))
 	}
 }
+
+func validateDatetime(value time.Time) error {
+	if value.Before(minTime) || value.After(maxTime) {
+		return serviceerror.NewInvalidArgument(
+			fmt.Sprintf("Date not supported in Elasticsearch: %v", value),
+		)
+	}
+	return nil
+}
+
+func validateString(value string) error {
+	if len(value) > maxStringLength {
+		return serviceerror.NewInvalidArgument(
+			fmt.Sprintf(
+				"Strings with more than %d bytes are not supported in Elasticsearch (got %s)",
+				maxStringLength,
+				value,
+			),
+		)
+	}
+	return nil
+}

common/persistence/visibility/store/elasticsearch/visibility_store.go

@@ -25,10 +25,33 @@
 package store
 
 import (
+	"strings"
+
 	"go.temporal.io/api/serviceerror"
 )
 
+type (
+	VisibilityStoreInvalidValuesError struct {
+		errs []error
+	}
+)
+
 var (
 	// OperationNotSupportedErr is returned when visibility operation in not supported.
 	OperationNotSupportedErr = serviceerror.NewInvalidArgument("Operation not supported. Please use on Elasticsearch")
 )
+
+func (e *VisibilityStoreInvalidValuesError) Error() string {
+	var sb strings.Builder
+	sb.WriteString("Visibility store invalid values errors: ")
+	for _, err := range e.errs {
+		sb.WriteString("[")
+		sb.WriteString(err.Error())
+		sb.WriteString("]")
+	}
+	return sb.String()
+}
+
+func NewVisibilityStoreInvalidValuesError(errs []error) error {
+	return &VisibilityStoreInvalidValuesError{errs: errs}
+}

common/persistence/visibility/store/errors.go

@@ -92,6 +92,12 @@ func (s *VisibilityStore) GetIndexName() string {
 	return s.sqlStore.GetDbName()
 }
 
+func (s *VisibilityStore) ValidateCustomSearchAttributes(
+	searchAttributes map[string]any,
+) (map[string]any, error) {
+	return searchAttributes, nil
+}
+
 func (s *VisibilityStore) RecordWorkflowExecutionStarted(
 	ctx context.Context,
 	request *store.InternalRecordWorkflowExecutionStartedRequest,
@@ -514,6 +520,14 @@ func (s *VisibilityStore) prepareSearchAttributesForDb(
 	if err != nil {
 		return nil, err
 	}
+	// This is to prevent existing tasks to fail indefinitely.
+	// If it's only invalid values error, then silently continue without them.
+	searchAttributes, err = s.ValidateCustomSearchAttributes(searchAttributes)
+	if err != nil {
+		if _, ok := err.(*store.VisibilityStoreInvalidValuesError); !ok {
+			return nil, err
+		}
+	}
 
 	for name, value := range searchAttributes {
 		if value == nil {

common/persistence/visibility/store/sql/visibility_store.go

@@ -177,6 +177,12 @@ func (v *visibilityStore) GetIndexName() string {
 	return ""
 }
 
+func (v *visibilityStore) ValidateCustomSearchAttributes(
+	searchAttributes map[string]any,
+) (map[string]any, error) {
+	return searchAttributes, nil
+}
+
 // Close releases the resources held by this object
 func (v *visibilityStore) Close() {
 	v.session.Close()

common/persistence/visibility/store/standard/cassandra/visibility_store.go

@@ -86,6 +86,12 @@ func (s *visibilityStore) GetIndexName() string {
 	return ""
 }
 
+func (s *visibilityStore) ValidateCustomSearchAttributes(
+	searchAttributes map[string]any,
+) (map[string]any, error) {
+	return searchAttributes, nil
+}
+
 func (s *visibilityStore) RecordWorkflowExecutionStarted(
 	ctx context.Context,
 	request *store.InternalRecordWorkflowExecutionStartedRequest,

common/persistence/visibility/store/standard/sql/visibility_store.go

@@ -79,6 +79,12 @@ func (s *standardStore) GetIndexName() string {
 	return ""
 }
 
+func (s *standardStore) ValidateCustomSearchAttributes(
+	searchAttributes map[string]any,
+) (map[string]any, error) {
+	return searchAttributes, nil
+}
+
 func (s *standardStore) RecordWorkflowExecutionStarted(
 	ctx context.Context,
 	request *store.InternalRecordWorkflowExecutionStartedRequest,

common/persistence/visibility/store/standard/visibility_store.go

@@ -45,6 +45,11 @@ type (
 		GetName() string
 		GetIndexName() string
 
+		// Validate search attributes based on the store constraints. It returns a new map containing
+		// only search attributes with valid values. If there are invalid values, an error of type
+		// VisibilityStoreInvalidValuesError wraps all invalid values errors.
+		ValidateCustomSearchAttributes(searchAttributes map[string]any) (map[string]any, error)
+
 		// Write APIs.
 		RecordWorkflowExecutionStarted(ctx context.Context, request *InternalRecordWorkflowExecutionStartedRequest) error
 		RecordWorkflowExecutionClosed(ctx context.Context, request *InternalRecordWorkflowExecutionClosedRequest) error