Further optimize RegionMap.MarshalJSON().

- Stop calling the encoder for values as well.
- Use tests to ensure that regions always contain valid utf8.

Allocations are now down 80% in FormatHandler.

Author: bojanz

address.go

@@ -6,7 +6,6 @@ package address
 
 import (
 	"bytes"
-	"encoding/json"
 	"fmt"
 	"regexp"
 	"sort"
@@ -172,17 +171,16 @@ func (r RegionMap) MarshalJSON() ([]byte, error) {
 	buf := &bytes.Buffer{}
 	buf.Grow(r.Len() * 30)
 	buf.WriteByte('{')
-	encoder := json.NewEncoder(buf)
 	for i, key := range r.keys {
 		if i > 0 {
 			buf.WriteByte(',')
 		}
-		// The key is assumed to be safe and need no escaping.
-		// Not calling encoder.Encode on the key saves many allocations.
-		buf.WriteString(`"` + key + `":`)
-		if err := encoder.Encode(r.values[key]); err != nil {
-			return nil, err
-		}
+		// A fully generic MarshalJSON would call encoder.Encode() to ensure both key and value
+		// are escaped and contain valid utf8. Since all regions are defined in the package, they
+		// are assumed to be valid, avoiding thousands of allocs when marshalling the format list.
+		buf.WriteString(`"` + key + `":"`)
+		buf.WriteString(r.values[key])
+		buf.WriteString(`"`)
 	}
 	buf.WriteByte('}')
 

address_test.go

@@ -7,6 +7,7 @@ import (
 	"encoding/json"
 	"reflect"
 	"testing"
+	"unicode/utf8"
 
 	"github.com/bojanz/address"
 )
@@ -339,6 +340,38 @@ func TestGetFormats(t *testing.T) {
 	}
 }
 
+func TestGetFormats_ValidRegionData(t *testing.T) {
+	// Confirm that all regions contain valid utf8.
+	// Avoids having the check at runtime, in RegionMap.MarshalJSON.
+	formats := address.GetFormats()
+	for countryCode, format := range formats {
+		if format.Regions.Len() > 0 {
+			keys := format.Regions.Keys()
+			for _, key := range keys {
+				value, _ := format.Regions.Get(key)
+				if !utf8.ValidString(key) {
+					t.Errorf("invalid key %v in %v regions", key, countryCode)
+				}
+				if !utf8.ValidString(value) {
+					t.Errorf("invalid value %v for key %v in %v regions", value, key, countryCode)
+				}
+			}
+		}
+		if format.LocalRegions.Len() > 0 {
+			keys := format.LocalRegions.Keys()
+			for _, key := range keys {
+				value, _ := format.LocalRegions.Get(key)
+				if !utf8.ValidString(key) {
+					t.Errorf("invalid key %v in %v local regions", key, countryCode)
+				}
+				if !utf8.ValidString(value) {
+					t.Errorf("invalid value %v for key %v in %v regions", value, key, countryCode)
+				}
+			}
+		}
+	}
+}
+
 func contains(a []string, x string) bool {
 	for _, v := range a {
 		if v == x {