optimize string migration: only return new value if needed

migrations/string_normalization/migration.go

@@ -19,6 +19,8 @@
 package string_normalization
 
 import (
+	"golang.org/x/text/unicode/norm"
+
 	"github.com/onflow/cadence/migrations"
 	"github.com/onflow/cadence/runtime/interpreter"
 	"github.com/onflow/cadence/runtime/sema"
@@ -41,13 +43,30 @@ func (StringNormalizingMigration) Migrate(
 	_ interpreter.StorageMapKey,
 	value interpreter.Value,
 	_ *interpreter.Interpreter,
-) (interpreter.Value, error) {
+) (
+	interpreter.Value,
+	error,
+) {
+
+	// Normalize strings and characters to NFC.
+	// If the value is already in NFC, skip the migration.
+
 	switch value := value.(type) {
 	case *interpreter.StringValue:
-		return interpreter.NewUnmeteredStringValue(value.Str), nil
+		unnormalizedStr := value.UnnormalizedStr
+		normalizedStr := norm.NFC.String(unnormalizedStr)
+		if normalizedStr == unnormalizedStr {
+			return nil, nil
+		}
+		return interpreter.NewStringValue_Unsafe(normalizedStr, unnormalizedStr), nil //nolint:staticcheck
 
 	case interpreter.CharacterValue:
-		return interpreter.NewUnmeteredCharacterValue(value.Str), nil
+		unnormalizedStr := value.UnnormalizedStr
+		normalizedStr := norm.NFC.String(unnormalizedStr)
+		if normalizedStr == unnormalizedStr {
+			return nil, nil
+		}
+		return interpreter.NewCharacterValue_Unsafe(normalizedStr, unnormalizedStr), nil //nolint:staticcheck
 	}
 
 	return nil, nil

runtime/interpreter/value.go

@@ -817,10 +817,21 @@ type CharacterValue struct {
 	UnnormalizedStr string
 }
 
-func NewUnmeteredCharacterValue(r string) CharacterValue {
+func NewUnmeteredCharacterValue(str string) CharacterValue {
 	return CharacterValue{
-		Str:             norm.NFC.String(r),
-		UnnormalizedStr: r,
+		Str:             norm.NFC.String(str),
+		UnnormalizedStr: str,
+	}
+}
+
+// Deprecated: NewStringValue_UnsafeNewCharacterValue_Unsafe creates a new character value
+// from the given normalized and unnormalized string.
+// NOTE: this function is unsafe, as it does not normalize the string.
+// It should only be used for e.g. migration purposes.
+func NewCharacterValue_Unsafe(normalizedStr, unnormalizedStr string) CharacterValue {
+	return CharacterValue{
+		Str:             normalizedStr,
+		UnnormalizedStr: unnormalizedStr,
 	}
 }
 
@@ -1045,6 +1056,19 @@ func NewUnmeteredStringValue(str string) *StringValue {
 	}
 }
 
+// Deprecated: NewStringValue_Unsafe creates a new string value
+// from the given normalized and unnormalized string.
+// NOTE: this function is unsafe, as it does not normalize the string.
+// It should only be used for e.g. migration purposes.
+func NewStringValue_Unsafe(normalizedStr, unnormalizedStr string) *StringValue {
+	return &StringValue{
+		Str:             normalizedStr,
+		UnnormalizedStr: unnormalizedStr,
+		// a negative value indicates the length has not been initialized, see Length()
+		length: -1,
+	}
+}
+
 func NewStringValue(
 	memoryGauge common.MemoryGauge,
 	memoryUsage common.MemoryUsage,