Cherry pick valyala/fastjson/pull/87

parser.go

@@ -267,28 +267,41 @@ func parseObject(s string, c *cache, depth int) (*Value, string, error) {
 }
 
 func escapeString(dst []byte, s string) []byte {
-	if !hasSpecialChars(s) {
-		// Fast path - nothing to escape.
-		dst = append(dst, '"')
-		dst = append(dst, s...)
-		dst = append(dst, '"')
-		return dst
-	}
-
-	// Slow path.
-	return strconv.AppendQuote(dst, s)
-}
-
-func hasSpecialChars(s string) bool {
-	if strings.IndexByte(s, '"') >= 0 || strings.IndexByte(s, '\\') >= 0 {
-		return true
-	}
+	dst = append(dst, '"')
 	for i := 0; i < len(s); i++ {
-		if s[i] < 0x20 {
-			return true
-		}
-	}
-	return false
+		c := s[i]
+		switch {
+		case c == '"':
+			// quotation mark
+			dst = append(dst, []byte{'\\', '"'}...)
+		case c == '\\':
+			// reverse solidus
+			dst = append(dst, []byte{'\\', '\\'}...)
+		case c >= 0x20:
+			// default, rest below are control chars
+			dst = append(dst, c)
+		case c == 0x08:
+			dst = append(dst, []byte{'\\', 'b'}...)
+		case c < 0x09:
+			dst = append(dst, []byte{'\\', 'u', '0', '0', '0', '0' + c}...)
+		case c == 0x09:
+			dst = append(dst, []byte{'\\', 't'}...)
+		case c == 0x0a:
+			dst = append(dst, []byte{'\\', 'n'}...)
+		case c == 0x0c:
+			dst = append(dst, []byte{'\\', 'f'}...)
+		case c == 0x0d:
+			dst = append(dst, []byte{'\\', 'r'}...)
+		case c < 0x10:
+			dst = append(dst, []byte{'\\', 'u', '0', '0', '0', 0x57 + c}...)
+		case c < 0x1a:
+			dst = append(dst, []byte{'\\', 'u', '0', '0', '1', 0x20 + c}...)
+		case c < 0x20:
+			dst = append(dst, []byte{'\\', 'u', '0', '0', '1', 0x47 + c}...)
+		}
+	}
+	dst = append(dst, '"')
+	return dst
 }
 
 func unescapeStringBestEffort(s string) string {

parser_test.go

@@ -1,6 +1,8 @@
 package fastjson
 
 import (
+	"bytes"
+	"encoding/json"
 	"fmt"
 	"math"
 	"strings"
@@ -1275,3 +1277,59 @@ func testParseGetSerial(s string) error {
 	}
 	return nil
 }
+
+// Tests for https://github.com/valyala/fastjson/issues/90
+// This was manifesting due to the use of strconv.AppendQuote
+func TestUTF8NonPrintableArtifacts(t *testing.T) {
+	testCases := []struct {
+		name string
+		s    string
+	}{
+		{
+			name: "problematic bytes",
+			s:    "data:\"\\xd6`\\xb76d\\xf6E\U000E8737(\\x91\\xb294\"",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			jsonEnvelope := struct {
+				Inner string
+			}{
+				Inner: tc.s,
+			}
+
+			m, err := json.Marshal(jsonEnvelope) // `m` is a full valid json
+			if err != nil {
+				t.Fatalf("unexpected error: %s", err)
+			}
+
+			// we will pass that `m` json through a fastjson marshal/unmarshal cycle
+
+			fastjsonValue, err := ParseBytes(m)
+			if err != nil {
+				t.Fatalf("unexpected error: %s", err)
+			}
+
+			o, err := fastjsonValue.Object()
+			if err != nil {
+				t.Fatalf("unexpected error: %s", err)
+			}
+
+			// In order to trigger the bug we need to visit all the keys and call Type() on them
+			o.Visit(func(k []byte, v *Value) {
+				v.Type()
+			})
+
+			res := fastjsonValue.MarshalTo(nil)
+			if !bytes.Equal(res, m) {
+				t.Fatalf("unexpected result; got %q; want %q", res, m)
+			}
+
+			err = ValidateBytes(res)
+			if err != nil {
+				t.Fatalf("unexpected error: %s", err)
+			}
+		})
+	}
+}