Joe Tsai | 879b18d | 2018-08-03 17:22:24 -0700 | [diff] [blame] | 1 | // Copyright 2018 The Go Authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style |
| 3 | // license that can be found in the LICENSE file. |
| 4 | |
| 5 | package json |
| 6 | |
| 7 | import ( |
| 8 | "io" |
| 9 | "math/bits" |
| 10 | "strconv" |
| 11 | "unicode" |
| 12 | "unicode/utf16" |
| 13 | "unicode/utf8" |
| 14 | |
Joe Tsai | 01ab296 | 2018-09-21 17:44:00 -0700 | [diff] [blame^] | 15 | "github.com/golang/protobuf/v2/internal/errors" |
Joe Tsai | 879b18d | 2018-08-03 17:22:24 -0700 | [diff] [blame] | 16 | ) |
| 17 | |
| 18 | func (p *encoder) marshalString(v Value) error { |
| 19 | var err error |
| 20 | p.out, err = appendString(p.out, v) |
| 21 | return err |
| 22 | } |
| 23 | func appendString(out []byte, v Value) ([]byte, error) { |
| 24 | if v.Type() != String { |
| 25 | return nil, errors.New("invalid type %v, expected string", v.Type()) |
| 26 | } |
| 27 | if len(v.raw) > 0 { |
| 28 | return append(out, v.raw...), nil |
| 29 | } |
| 30 | in := v.String() |
| 31 | |
| 32 | var nerr errors.NonFatal |
| 33 | out = append(out, '"') |
| 34 | i := indexNeedEscape(in) |
| 35 | in, out = in[i:], append(out, in[:i]...) |
| 36 | for len(in) > 0 { |
| 37 | switch r, n := utf8.DecodeRuneInString(in); { |
| 38 | case r == utf8.RuneError && n == 1: |
| 39 | nerr.AppendInvalidUTF8("") |
| 40 | in, out = in[1:], append(out, in[0]) // preserve invalid byte |
| 41 | case r < ' ' || r == '"' || r == '\\': |
| 42 | out = append(out, '\\') |
| 43 | switch r { |
| 44 | case '"', '\\': |
| 45 | out = append(out, byte(r)) |
| 46 | case '\b': |
| 47 | out = append(out, 'b') |
| 48 | case '\f': |
| 49 | out = append(out, 'f') |
| 50 | case '\n': |
| 51 | out = append(out, 'n') |
| 52 | case '\r': |
| 53 | out = append(out, 'r') |
| 54 | case '\t': |
| 55 | out = append(out, 't') |
| 56 | default: |
| 57 | out = append(out, 'u') |
| 58 | out = append(out, "0000"[1+(bits.Len32(uint32(r))-1)/4:]...) |
| 59 | out = strconv.AppendUint(out, uint64(r), 16) |
| 60 | } |
| 61 | in = in[n:] |
| 62 | default: |
| 63 | i := indexNeedEscape(in[n:]) |
| 64 | in, out = in[n+i:], append(out, in[:n+i]...) |
| 65 | } |
| 66 | } |
| 67 | out = append(out, '"') |
| 68 | return out, nerr.E |
| 69 | } |
| 70 | |
| 71 | func (p *decoder) unmarshalString() (Value, error) { |
| 72 | v, n, err := consumeString(p.in) |
| 73 | p.consume(n) |
| 74 | return v, err |
| 75 | } |
| 76 | func consumeString(in []byte) (Value, int, error) { |
| 77 | var nerr errors.NonFatal |
| 78 | in0 := in |
| 79 | if len(in) == 0 { |
| 80 | return Value{}, 0, io.ErrUnexpectedEOF |
| 81 | } |
| 82 | if in[0] != '"' { |
| 83 | return Value{}, 0, newSyntaxError("invalid character %q at start of string", in[0]) |
| 84 | } |
| 85 | in = in[1:] |
| 86 | i := indexNeedEscape(string(in)) |
| 87 | in, out := in[i:], in[:i:i] // set cap to prevent mutations |
| 88 | for len(in) > 0 { |
| 89 | switch r, n := utf8.DecodeRune(in); { |
| 90 | case r == utf8.RuneError && n == 1: |
| 91 | nerr.AppendInvalidUTF8("") |
| 92 | in, out = in[1:], append(out, in[0]) // preserve invalid byte |
| 93 | case r < ' ': |
| 94 | return Value{}, 0, newSyntaxError("invalid character %q in string", r) |
| 95 | case r == '"': |
| 96 | in = in[1:] |
| 97 | n := len(in0) - len(in) |
| 98 | v := rawValueOf(string(out), in0[:n:n]) |
| 99 | return v, n, nerr.E |
| 100 | case r == '\\': |
| 101 | if len(in) < 2 { |
| 102 | return Value{}, 0, io.ErrUnexpectedEOF |
| 103 | } |
| 104 | switch r := in[1]; r { |
| 105 | case '"', '\\', '/': |
| 106 | in, out = in[2:], append(out, r) |
| 107 | case 'b': |
| 108 | in, out = in[2:], append(out, '\b') |
| 109 | case 'f': |
| 110 | in, out = in[2:], append(out, '\f') |
| 111 | case 'n': |
| 112 | in, out = in[2:], append(out, '\n') |
| 113 | case 'r': |
| 114 | in, out = in[2:], append(out, '\r') |
| 115 | case 't': |
| 116 | in, out = in[2:], append(out, '\t') |
| 117 | case 'u': |
| 118 | if len(in) < 6 { |
| 119 | return Value{}, 0, io.ErrUnexpectedEOF |
| 120 | } |
| 121 | v, err := strconv.ParseUint(string(in[2:6]), 16, 16) |
| 122 | if err != nil { |
| 123 | return Value{}, 0, newSyntaxError("invalid escape code %q in string", in[:6]) |
| 124 | } |
| 125 | in = in[6:] |
| 126 | |
| 127 | r := rune(v) |
| 128 | if utf16.IsSurrogate(r) { |
| 129 | if len(in) < 6 { |
| 130 | return Value{}, 0, io.ErrUnexpectedEOF |
| 131 | } |
| 132 | v, err := strconv.ParseUint(string(in[2:6]), 16, 16) |
| 133 | r = utf16.DecodeRune(r, rune(v)) |
| 134 | if in[0] != '\\' || in[1] != 'u' || r == unicode.ReplacementChar || err != nil { |
| 135 | return Value{}, 0, newSyntaxError("invalid escape code %q in string", in[:6]) |
| 136 | } |
| 137 | in = in[6:] |
| 138 | } |
| 139 | out = append(out, string(r)...) |
| 140 | default: |
| 141 | return Value{}, 0, newSyntaxError("invalid escape code %q in string", in[:2]) |
| 142 | } |
| 143 | default: |
| 144 | i := indexNeedEscape(string(in[n:])) |
| 145 | in, out = in[n+i:], append(out, in[:n+i]...) |
| 146 | } |
| 147 | } |
| 148 | return Value{}, 0, io.ErrUnexpectedEOF |
| 149 | } |
| 150 | |
| 151 | // indexNeedEscape returns the index of the next character that needs escaping. |
| 152 | // If no characters need escaping, this returns the input length. |
| 153 | func indexNeedEscape(s string) int { |
| 154 | for i, r := range s { |
| 155 | if r < ' ' || r == '\\' || r == '"' || r == utf8.RuneError { |
| 156 | return i |
| 157 | } |
| 158 | } |
| 159 | return len(s) |
| 160 | } |