@@ -10,6 +10,8 @@ import (
1010 "reflect"
1111 "sort"
1212 "strings"
13+ "unicode"
14+ "unicode/utf8"
1315
1416 "github.com/golang/protobuf/proto"
1517 "github.com/golang/protobuf/protoc-gen-go/descriptor"
@@ -1656,7 +1658,7 @@ func (p *Printer) printOption(name string, optVal interface{}, w *writer, indent
16561658 case string :
16571659 fmt .Fprintf (w , "%s" , quotedString (optVal ))
16581660 case []byte :
1659- fmt .Fprintf (w , "%s" , quotedString (string (optVal )))
1661+ fmt .Fprintf (w , "%s" , quotedBytes (string (optVal )))
16601662 case bool :
16611663 fmt .Fprintf (w , "%v" , optVal )
16621664 case ident :
@@ -1978,12 +1980,12 @@ func optionsAsElementAddrs(optionsTag int32, order int, opts map[int32][]option)
19781980 return optAddrs
19791981}
19801982
1981- // quotedString implements the text format for string literals for protocol
1982- // buffers. This form is also acceptable for string literals in option values
1983- // by the protocol buffer compiler, protoc.
1984- func quotedString (s string ) string {
1983+ // quotedBytes implements the text format for string literals for protocol
1984+ // buffers. Since the underlying data is a bytes field, this encodes all
1985+ // bytes outside the 7-bit ASCII printable range. To preserve unicode strings
1986+ // without byte escapes, use quotedString.
1987+ func quotedBytes (s string ) string {
19851988 var b bytes.Buffer
1986- // use WriteByte here to get any needed indent
19871989 b .WriteByte ('"' )
19881990 // Loop over the bytes, not the runes.
19891991 for i := 0 ; i < len (s ); i ++ {
@@ -2014,6 +2016,62 @@ func quotedString(s string) string {
20142016 return b .String ()
20152017}
20162018
2019+ // quotedString implements the text format for string literals for protocol
2020+ // buffers. This form is also acceptable for string literals in option values
2021+ // by the protocol buffer compiler, protoc.
2022+ func quotedString (s string ) string {
2023+ var b bytes.Buffer
2024+ b .WriteByte ('"' )
2025+ // Loop over the bytes, not the runes.
2026+ for {
2027+ r , n := utf8 .DecodeRuneInString (s )
2028+ if n == 0 {
2029+ break // end of string
2030+ }
2031+ if r == utf8 .RuneError && n == 1 {
2032+ // Invalid UTF8! Use an octal byte escape to encode the bad byte.
2033+ fmt .Fprintf (& b , "\\ %03o" , s [0 ])
2034+ s = s [1 :]
2035+ continue
2036+ }
2037+
2038+ // Divergence from C++: we don't escape apostrophes.
2039+ // There's no need to escape them, and the C++ parser
2040+ // copes with a naked apostrophe.
2041+ switch r {
2042+ case '\n' :
2043+ b .WriteString ("\\ n" )
2044+ case '\r' :
2045+ b .WriteString ("\\ r" )
2046+ case '\t' :
2047+ b .WriteString ("\\ t" )
2048+ case '"' :
2049+ b .WriteString ("\\ " )
2050+ case '\\' :
2051+ b .WriteString ("\\ \\ " )
2052+ default :
2053+ if unicode .IsPrint (r ) {
2054+ b .WriteRune (r )
2055+ } else {
2056+ // if it's not printable, use a unicode escape
2057+ if r > 0xffff {
2058+ fmt .Fprintf (& b , "\\ U%08X" , r )
2059+ } else if r > 0x7F {
2060+ fmt .Fprintf (& b , "\\ u%04X" , r )
2061+ } else {
2062+ fmt .Fprintf (& b , "\\ %03o" , byte (r ))
2063+ }
2064+ }
2065+ }
2066+
2067+ s = s [n :]
2068+ }
2069+
2070+ b .WriteByte ('"' )
2071+
2072+ return b .String ()
2073+ }
2074+
20172075type elementAddr struct {
20182076 elementType int32
20192077 elementIndex int
0 commit comments