vendor/golang.org/x/text/encoding/japanese/all_test.go

   1 // Copyright 2015 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 japanese
   6
   7 import (
   8         "fmt"
   9         "strings"
  10         "testing"
  11
  12         "golang.org/x/text/encoding"
  13         "golang.org/x/text/encoding/internal"
  14         "golang.org/x/text/encoding/internal/enctest"
  15         "golang.org/x/text/transform"
  16 )
  17
  18 func dec(e encoding.Encoding) (dir string, t transform.Transformer, err error) {
  19         return "Decode", e.NewDecoder(), nil
  20 }
  21 func enc(e encoding.Encoding) (dir string, t transform.Transformer, err error) {
  22         return "Encode", e.NewEncoder(), internal.ErrASCIIReplacement
  23 }
  24
  25 func TestNonRepertoire(t *testing.T) {
  26         // Pick n to cause the destination buffer in transform.String to overflow.
  27         const n = 100
  28         long := strings.Repeat(".", n)
  29         testCases := []struct {
  30                 init      func(e encoding.Encoding) (string, transform.Transformer, error)
  31                 e         encoding.Encoding
  32                 src, want string
  33         }{
  34                 {enc, EUCJP, "갂", ""},
  35                 {enc, EUCJP, "a갂", "a"},
  36                 {enc, EUCJP, "丌갂", "\x8f\xb0\xa4"},
  37
  38                 {enc, ISO2022JP, "갂", ""},
  39                 {enc, ISO2022JP, "a갂", "a"},
  40                 {enc, ISO2022JP, "朗갂", "\x1b$BzF\x1b(B"}, // switch back to ASCII mode at end
  41
  42                 {enc, ShiftJIS, "갂", ""},
  43                 {enc, ShiftJIS, "a갂", "a"},
  44                 {enc, ShiftJIS, "\u2190갂", "\x81\xa9"},
  45
  46                 // Continue correctly after errors
  47                 {dec, EUCJP, "\x8e\xa0", "\ufffd\ufffd"},
  48                 {dec, EUCJP, "\x8e\xe0", "\ufffd"},
  49                 {dec, EUCJP, "\x8e\xff", "\ufffd\ufffd"},
  50                 {dec, EUCJP, "\x8ea", "\ufffda"},
  51                 {dec, EUCJP, "\x8f\xa0", "\ufffd\ufffd"},
  52                 {dec, EUCJP, "\x8f\xa1\xa0", "\ufffd\ufffd"},
  53                 {dec, EUCJP, "\x8f\xa1a", "\ufffda"},
  54                 {dec, EUCJP, "\x8f\xa1a", "\ufffda"},
  55                 {dec, EUCJP, "\x8f\xa1a", "\ufffda"},
  56                 {dec, EUCJP, "\x8f\xa2\xa2", "\ufffd"},
  57                 {dec, EUCJP, "\xfe", "\ufffd"},
  58                 {dec, EUCJP, "\xfe\xfc", "\ufffd"},
  59                 {dec, EUCJP, "\xfe\xff", "\ufffd\ufffd"},
  60                 // Correct handling of end of source
  61                 {dec, EUCJP, strings.Repeat("\x8e", n), strings.Repeat("\ufffd", n)},
  62                 {dec, EUCJP, strings.Repeat("\x8f", n), strings.Repeat("\ufffd", n)},
  63                 {dec, EUCJP, strings.Repeat("\x8f\xa0", n), strings.Repeat("\ufffd", 2*n)},
  64                 {dec, EUCJP, "a" + strings.Repeat("\x8f\xa1", n), "a" + strings.Repeat("\ufffd", n)},
  65                 {dec, EUCJP, "a" + strings.Repeat("\x8f\xa1\xff", n), "a" + strings.Repeat("\ufffd", 2*n)},
  66
  67                 // Continue correctly after errors
  68                 {dec, ShiftJIS, "\x80", "\u0080"}, // It's what the spec says.
  69                 {dec, ShiftJIS, "\x81", "\ufffd"},
  70                 {dec, ShiftJIS, "\x81\x7f", "\ufffd\u007f"},
  71                 {dec, ShiftJIS, "\xe0", "\ufffd"},
  72                 {dec, ShiftJIS, "\xe0\x39", "\ufffd\u0039"},
  73                 {dec, ShiftJIS, "\xe0\x9f", "燹"},
  74                 {dec, ShiftJIS, "\xe0\xfd", "\ufffd"},
  75                 {dec, ShiftJIS, "\xef\xfc", "\ufffd"},
  76                 {dec, ShiftJIS, "\xfc\xfc", "\ufffd"},
  77                 {dec, ShiftJIS, "\xfc\xfd", "\ufffd"},
  78                 {dec, ShiftJIS, "\xfdaa", "\ufffdaa"},
  79
  80                 {dec, ShiftJIS, strings.Repeat("\x81\x81", n), strings.Repeat("＝", n)},
  81                 {dec, ShiftJIS, strings.Repeat("\xe0\xfd", n), strings.Repeat("\ufffd", n)},
  82                 {dec, ShiftJIS, "a" + strings.Repeat("\xe0\xfd", n), "a" + strings.Repeat("\ufffd", n)},
  83
  84                 {dec, ISO2022JP, "\x1b$", "\ufffd$"},
  85                 {dec, ISO2022JP, "\x1b(", "\ufffd("},
  86                 {dec, ISO2022JP, "\x1b@", "\ufffd@"},
  87                 {dec, ISO2022JP, "\x1bZ", "\ufffdZ"},
  88                 // incomplete escapes
  89                 {dec, ISO2022JP, "\x1b$", "\ufffd$"},
  90                 {dec, ISO2022JP, "\x1b$J.", "\ufffd$J."},             // illegal
  91                 {dec, ISO2022JP, "\x1b$B.", "\ufffd"},                // JIS208
  92                 {dec, ISO2022JP, "\x1b$(", "\ufffd$("},               // JIS212
  93                 {dec, ISO2022JP, "\x1b$(..", "\ufffd$(.."},           // JIS212
  94                 {dec, ISO2022JP, "\x1b$(" + long, "\ufffd$(" + long}, // JIS212
  95                 {dec, ISO2022JP, "\x1b$(D.", "\ufffd"},               // JIS212
  96                 {dec, ISO2022JP, "\x1b$(D..", "\ufffd"},              // JIS212
  97                 {dec, ISO2022JP, "\x1b$(D...", "\ufffd\ufffd"},       // JIS212
  98                 {dec, ISO2022JP, "\x1b(B.", "."},                     // ascii
  99                 {dec, ISO2022JP, "\x1b(B..", ".."},                   // ascii
 100                 {dec, ISO2022JP, "\x1b(J.", "."},                     // roman
 101                 {dec, ISO2022JP, "\x1b(J..", ".."},                   // roman
 102                 {dec, ISO2022JP, "\x1b(I\x20", "\ufffd"},             // katakana
 103                 {dec, ISO2022JP, "\x1b(I\x20\x20", "\ufffd\ufffd"},   // katakana
 104                 // recover to same state
 105                 {dec, ISO2022JP, "\x1b(B\x1b.", "\ufffd."},
 106                 {dec, ISO2022JP, "\x1b(I\x1b.", "\ufffdｮ"},
 107                 {dec, ISO2022JP, "\x1b(I\x1b$.", "\ufffd､ｮ"},
 108                 {dec, ISO2022JP, "\x1b(I\x1b(.", "\ufffdｨｮ"},
 109                 {dec, ISO2022JP, "\x1b$B\x7e\x7e", "\ufffd"},
 110                 {dec, ISO2022JP, "\x1b$@\x0a.", "\x0a."},
 111                 {dec, ISO2022JP, "\x1b$B\x0a.", "\x0a."},
 112                 {dec, ISO2022JP, "\x1b$(D\x0a.", "\x0a."},
 113                 {dec, ISO2022JP, "\x1b$(D\x7e\x7e", "\ufffd"},
 114                 {dec, ISO2022JP, "\x80", "\ufffd"},
 115
 116                 // TODO: according to https://encoding.spec.whatwg.org/#iso-2022-jp,
 117                 // these should all be correct.
 118                 // {dec, ISO2022JP, "\x1b(B\x0E", "\ufffd"},
 119                 // {dec, ISO2022JP, "\x1b(B\x0F", "\ufffd"},
 120                 {dec, ISO2022JP, "\x1b(B\x5C", "\u005C"},
 121                 {dec, ISO2022JP, "\x1b(B\x7E", "\u007E"},
 122                 // {dec, ISO2022JP, "\x1b(J\x0E", "\ufffd"},
 123                 // {dec, ISO2022JP, "\x1b(J\x0F", "\ufffd"},
 124                 // {dec, ISO2022JP, "\x1b(J\x5C", "\u00A5"},
 125                 // {dec, ISO2022JP, "\x1b(J\x7E", "\u203E"},
 126         }
 127         for _, tc := range testCases {
 128                 dir, tr, wantErr := tc.init(tc.e)
 129                 t.Run(fmt.Sprintf("%s/%v/%q", dir, tc.e, tc.src), func(t *testing.T) {
 130                         dst := make([]byte, 100000)
 131                         src := []byte(tc.src)
 132                         for i := 0; i <= len(tc.src); i++ {
 133                                 nDst, nSrc, err := tr.Transform(dst, src[:i], false)
 134                                 if err != nil && err != transform.ErrShortSrc && err != wantErr {
 135                                         t.Fatalf("error on first call to Transform: %v", err)
 136                                 }
 137                                 n, _, err := tr.Transform(dst[nDst:], src[nSrc:], true)
 138                                 nDst += n
 139                                 if err != wantErr {
 140                                         t.Fatalf("(%q|%q): got %v; want %v", tc.src[:i], tc.src[i:], err, wantErr)
 141                                 }
 142                                 if got := string(dst[:nDst]); got != tc.want {
 143                                         t.Errorf("(%q|%q):\ngot  %q\nwant %q", tc.src[:i], tc.src[i:], got, tc.want)
 144                                 }
 145                         }
 146                 })
 147         }
 148 }
 149
 150 func TestCorrect(t *testing.T) {
 151         testCases := []struct {
 152                 init      func(e encoding.Encoding) (string, transform.Transformer, error)
 153                 e         encoding.Encoding
 154                 src, want string
 155         }{
 156                 {dec, ShiftJIS, "\x9f\xfc", "滌"},
 157                 {dec, ShiftJIS, "\xfb\xfc", "髙"},
 158                 {dec, ShiftJIS, "\xfa\xb1", "﨑"},
 159                 {enc, ShiftJIS, "滌", "\x9f\xfc"},
 160                 {enc, ShiftJIS, "﨑", "\xed\x95"},
 161         }
 162         for _, tc := range testCases {
 163                 dir, tr, _ := tc.init(tc.e)
 164
 165                 dst, _, err := transform.String(tr, tc.src)
 166                 if err != nil {
 167                         t.Errorf("%s %v(%q): got %v; want %v", dir, tc.e, tc.src, err, nil)
 168                 }
 169                 if got := string(dst); got != tc.want {
 170                         t.Errorf("%s %v(%q):\ngot  %q\nwant %q", dir, tc.e, tc.src, got, tc.want)
 171                 }
 172         }
 173 }
 174
 175 func TestBasics(t *testing.T) {
 176         // The encoded forms can be verified by the iconv program:
 177         // $ echo 月日は百代 | iconv -f UTF-8 -t SHIFT-JIS | xxd
 178         testCases := []struct {
 179                 e         encoding.Encoding
 180                 encPrefix string
 181                 encSuffix string
 182                 encoded   string
 183                 utf8      string
 184         }{{
 185                 // "A｡ｶﾟ 0208: etc 0212: etc" is a nonsense string that contains ASCII, half-width
 186                 // kana, JIS X 0208 (including two near the kink in the Shift JIS second byte
 187                 // encoding) and JIS X 0212 encodable codepoints.
 188                 //
 189                 // "月日は百代の過客にして、行かふ年も又旅人也。" is from the 17th century poem
 190                 // "Oku no Hosomichi" and contains both hiragana and kanji.
 191                 e: EUCJP,
 192                 encoded: "A\x8e\xa1\x8e\xb6\x8e\xdf " +
 193                         "0208: \xa1\xa1\xa1\xa2\xa1\xdf\xa1\xe0\xa1\xfd\xa1\xfe\xa2\xa1\xa2\xa2\xf4\xa6 " +
 194                         "0212: \x8f\xa2\xaf\x8f\xed\xe3",
 195                 utf8: "A｡ｶﾟ " +
 196                         "0208: \u3000\u3001\u00d7\u00f7\u25ce\u25c7\u25c6\u25a1\u7199 " +
 197                         "0212: \u02d8\u9fa5",
 198         }, {
 199                 e: EUCJP,
 200                 encoded: "\xb7\xee\xc6\xfc\xa4\xcf\xc9\xb4\xc2\xe5\xa4\xce\xb2\xe1\xb5\xd2" +
 201                         "\xa4\xcb\xa4\xb7\xa4\xc6\xa1\xa2\xb9\xd4\xa4\xab\xa4\xd5\xc7\xaf" +
 202                         "\xa4\xe2\xcb\xf4\xce\xb9\xbf\xcd\xcc\xe9\xa1\xa3",
 203                 utf8: "月日は百代の過客にして、行かふ年も又旅人也。",
 204         }, {
 205                 e:         ISO2022JP,
 206                 encSuffix: "\x1b\x28\x42",
 207                 encoded: "\x1b\x28\x49\x21\x36\x5f\x1b\x28\x42 " +
 208                         "0208: \x1b\x24\x42\x21\x21\x21\x22\x21\x5f\x21\x60\x21\x7d\x21\x7e\x22\x21\x22\x22\x74\x26",
 209                 utf8: "｡ｶﾟ " +
 210                         "0208: \u3000\u3001\u00d7\u00f7\u25ce\u25c7\u25c6\u25a1\u7199",
 211         }, {
 212                 e:         ISO2022JP,
 213                 encPrefix: "\x1b\x24\x42",
 214                 encSuffix: "\x1b\x28\x42",
 215                 encoded: "\x37\x6e\x46\x7c\x24\x4f\x49\x34\x42\x65\x24\x4e\x32\x61\x35\x52" +
 216                         "\x24\x4b\x24\x37\x24\x46\x21\x22\x39\x54\x24\x2b\x24\x55\x47\x2f" +
 217                         "\x24\x62\x4b\x74\x4e\x39\x3f\x4d\x4c\x69\x21\x23",
 218                 utf8: "月日は百代の過客にして、行かふ年も又旅人也。",
 219         }, {
 220                 e: ShiftJIS,
 221                 encoded: "A\xa1\xb6\xdf " +
 222                         "0208: \x81\x40\x81\x41\x81\x7e\x81\x80\x81\x9d\x81\x9e\x81\x9f\x81\xa0\xea\xa4",
 223                 utf8: "A｡ｶﾟ " +
 224                         "0208: \u3000\u3001\u00d7\u00f7\u25ce\u25c7\u25c6\u25a1\u7199",
 225         }, {
 226                 e: ShiftJIS,
 227                 encoded: "\x8c\x8e\x93\xfa\x82\xcd\x95\x53\x91\xe3\x82\xcc\x89\xdf\x8b\x71" +
 228                         "\x82\xc9\x82\xb5\x82\xc4\x81\x41\x8d\x73\x82\xa9\x82\xd3\x94\x4e" +
 229                         "\x82\xe0\x96\x94\x97\xb7\x90\x6c\x96\xe7\x81\x42",
 230                 utf8: "月日は百代の過客にして、行かふ年も又旅人也。",
 231         }}
 232
 233         for _, tc := range testCases {
 234                 enctest.TestEncoding(t, tc.e, tc.encoded, tc.utf8, tc.encPrefix, tc.encSuffix)
 235         }
 236 }
 237
 238 func TestFiles(t *testing.T) {
 239         enctest.TestFile(t, EUCJP)
 240         enctest.TestFile(t, ISO2022JP)
 241         enctest.TestFile(t, ShiftJIS)
 242 }
 243
 244 func BenchmarkEncoding(b *testing.B) {
 245         enctest.Benchmark(b, EUCJP)
 246         enctest.Benchmark(b, ISO2022JP)
 247         enctest.Benchmark(b, ShiftJIS)
 248 }