Merge pull request #365 from orisano/fix/#364

fix: care surrogate-pair for non stream string decoder

decode_test.go

@@ -3931,3 +3931,16 @@ func TestIssue359(t *testing.T) {
 		t.Errorf("unexpected result: %v", string(v))
 	}
 }
+
+func TestIssue364(t *testing.T) {
+	var v struct {
+		Description string `json:"description"`
+	}
+	err := json.Unmarshal([]byte(`{"description":"\uD83D\uDE87 Toledo is a metro station"}`), &v)
+	if err != nil {
+		t.Errorf("unexpected error: %v", err)
+	}
+	if v.Description != "🚇 Toledo is a metro station" {
+		t.Errorf("unexpected result: %v", v.Description)
+	}
+}

internal/decoder/string.go

@@ -386,6 +386,19 @@ func unescapeString(buf []byte) int {
 				v3 := hexToInt[char(src, 4)]
 				v4 := hexToInt[char(src, 5)]
 				code := rune((v1 << 12) | (v2 << 8) | (v3 << 4) | v4)
+				if code >= 0xd800 && code < 0xdc00 && uintptr(unsafeAdd(src, 11)) < uintptr(end) {
+					if char(src, 6) == '\\' && char(src, 7) == 'u' {
+						v1 := hexToInt[char(src, 8)]
+						v2 := hexToInt[char(src, 9)]
+						v3 := hexToInt[char(src, 10)]
+						v4 := hexToInt[char(src, 11)]
+						lo := rune((v1 << 12) | (v2 << 8) | (v3 << 4) | v4)
+						if lo >= 0xdc00 && lo < 0xe000 {
+							code = (code-0xd800)<<10 | (lo - 0xdc00) + 0x10000
+							src = unsafeAdd(src, 6)
+						}
+					}
+				}
 				var b [utf8.UTFMax]byte
 				n := utf8.EncodeRune(b[:], code)
 				switch n {