libgo/go/mime/encodedword_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 mime
   6
   7 import (
   8         "errors"
   9         "io"
  10         "io/ioutil"
  11         "strings"
  12         "testing"
  13 )
  14
  15 func TestEncodeWord(t *testing.T) {
  16         utf8, iso88591 := "utf-8", "iso-8859-1"
  17         tests := []struct {
  18                 enc      WordEncoder
  19                 charset  string
  20                 src, exp string
  21         }{
  22                 {QEncoding, utf8, "François-Jérôme", "=?utf-8?q?Fran=C3=A7ois-J=C3=A9r=C3=B4me?="},
  23                 {BEncoding, utf8, "Café", "=?utf-8?b?Q2Fmw6k=?="},
  24                 {QEncoding, iso88591, "La Seleção", "=?iso-8859-1?q?La_Sele=C3=A7=C3=A3o?="},
  25                 {QEncoding, utf8, "", ""},
  26                 {QEncoding, utf8, "A", "A"},
  27                 {QEncoding, iso88591, "a", "a"},
  28                 {QEncoding, utf8, "123 456", "123 456"},
  29                 {QEncoding, utf8, "\t !\"#$%&'()*+,-./ :;<>?@[\\]^_`{|}~", "\t !\"#$%&'()*+,-./ :;<>?@[\\]^_`{|}~"},
  30                 {QEncoding, utf8, strings.Repeat("é", 10), "=?utf-8?q?" + strings.Repeat("=C3=A9", 10) + "?="},
  31                 {QEncoding, utf8, strings.Repeat("é", 11), "=?utf-8?q?" + strings.Repeat("=C3=A9", 10) + "?= =?utf-8?q?=C3=A9?="},
  32                 {QEncoding, iso88591, strings.Repeat("\xe9", 22), "=?iso-8859-1?q?" + strings.Repeat("=E9", 22) + "?="},
  33                 {QEncoding, utf8, strings.Repeat("\x80", 22), "=?utf-8?q?" + strings.Repeat("=80", 21) + "?= =?utf-8?q?=80?="},
  34                 {BEncoding, iso88591, strings.Repeat("\xe9", 45), "=?iso-8859-1?b?" + strings.Repeat("6enp", 15) + "?="},
  35                 {BEncoding, utf8, strings.Repeat("\x80", 48), "=?utf-8?b?" + strings.Repeat("gICA", 15) + "?= =?utf-8?b?gICA?="},
  36         }
  37
  38         for _, test := range tests {
  39                 if s := test.enc.Encode(test.charset, test.src); s != test.exp {
  40                         t.Errorf("Encode(%q) = %q, want %q", test.src, s, test.exp)
  41                 }
  42         }
  43 }
  44
  45 func TestEncodedWordLength(t *testing.T) {
  46         tests := []struct {
  47                 enc WordEncoder
  48                 src string
  49         }{
  50                 {QEncoding, strings.Repeat("à", 30)},
  51                 {QEncoding, strings.Repeat("é", 60)},
  52                 {BEncoding, strings.Repeat("ï", 25)},
  53                 {BEncoding, strings.Repeat("ô", 37)},
  54                 {BEncoding, strings.Repeat("\x80", 50)},
  55                 {QEncoding, "{$firstname} Bienvendio a Apostolica, aquà inicia el camino de tu"},
  56         }
  57
  58         for _, test := range tests {
  59                 s := test.enc.Encode("utf-8", test.src)
  60                 wordLen := 0
  61                 for i := 0; i < len(s); i++ {
  62                         if s[i] == ' ' {
  63                                 wordLen = 0
  64                                 continue
  65                         }
  66
  67                         wordLen++
  68                         if wordLen > maxEncodedWordLen {
  69                                 t.Errorf("Encode(%q) has more than %d characters: %q",
  70                                         test.src, maxEncodedWordLen, s)
  71                         }
  72                 }
  73         }
  74 }
  75
  76 func TestDecodeWord(t *testing.T) {
  77         tests := []struct {
  78                 src, exp string
  79                 hasErr   bool
  80         }{
  81                 {"=?UTF-8?Q?=C2=A1Hola,_se=C3=B1or!?=", "¡Hola, señor!", false},
  82                 {"=?UTF-8?Q?Fran=C3=A7ois-J=C3=A9r=C3=B4me?=", "François-Jérôme", false},
  83                 {"=?UTF-8?q?ascii?=", "ascii", false},
  84                 {"=?utf-8?B?QW5kcsOp?=", "André", false},
  85                 {"=?ISO-8859-1?Q?Rapha=EBl_Dupont?=", "Raphaël Dupont", false},
  86                 {"=?utf-8?b?IkFudG9uaW8gSm9zw6kiIDxqb3NlQGV4YW1wbGUub3JnPg==?=", `"Antonio José" <jose@example.org>`, false},
  87                 {"=?UTF-8?A?Test?=", "", true},
  88                 {"=?UTF-8?Q?A=B?=", "", true},
  89                 {"=?UTF-8?Q?=A?=", "", true},
  90                 {"=?UTF-8?A?A?=", "", true},
  91         }
  92
  93         for _, test := range tests {
  94                 dec := new(WordDecoder)
  95                 s, err := dec.Decode(test.src)
  96                 if test.hasErr && err == nil {
  97                         t.Errorf("Decode(%q) should return an error", test.src)
  98                         continue
  99                 }
 100                 if !test.hasErr && err != nil {
 101                         t.Errorf("Decode(%q): %v", test.src, err)
 102                         continue
 103                 }
 104                 if s != test.exp {
 105                         t.Errorf("Decode(%q) = %q, want %q", test.src, s, test.exp)
 106                 }
 107         }
 108 }
 109
 110 func TestDecodeHeader(t *testing.T) {
 111         tests := []struct {
 112                 src, exp string
 113         }{
 114                 {"=?UTF-8?Q?=C2=A1Hola,_se=C3=B1or!?=", "¡Hola, señor!"},
 115                 {"=?UTF-8?Q?Fran=C3=A7ois-J=C3=A9r=C3=B4me?=", "François-Jérôme"},
 116                 {"=?UTF-8?q?ascii?=", "ascii"},
 117                 {"=?utf-8?B?QW5kcsOp?=", "André"},
 118                 {"=?ISO-8859-1?Q?Rapha=EBl_Dupont?=", "Raphaël Dupont"},
 119                 {"Jean", "Jean"},
 120                 {"=?utf-8?b?IkFudG9uaW8gSm9zw6kiIDxqb3NlQGV4YW1wbGUub3JnPg==?=", `"Antonio José" <jose@example.org>`},
 121                 {"=?UTF-8?A?Test?=", "=?UTF-8?A?Test?="},
 122                 {"=?UTF-8?Q?A=B?=", "=?UTF-8?Q?A=B?="},
 123                 {"=?UTF-8?Q?=A?=", "=?UTF-8?Q?=A?="},
 124                 {"=?UTF-8?A?A?=", "=?UTF-8?A?A?="},
 125                 // Incomplete words
 126                 {"=?", "=?"},
 127                 {"=?UTF-8?", "=?UTF-8?"},
 128                 {"=?UTF-8?=", "=?UTF-8?="},
 129                 {"=?UTF-8?Q", "=?UTF-8?Q"},
 130                 {"=?UTF-8?Q?", "=?UTF-8?Q?"},
 131                 {"=?UTF-8?Q?=", "=?UTF-8?Q?="},
 132                 {"=?UTF-8?Q?A", "=?UTF-8?Q?A"},
 133                 {"=?UTF-8?Q?A?", "=?UTF-8?Q?A?"},
 134                 // Tests from RFC 2047
 135                 {"=?ISO-8859-1?Q?a?=", "a"},
 136                 {"=?ISO-8859-1?Q?a?= b", "a b"},
 137                 {"=?ISO-8859-1?Q?a?= =?ISO-8859-1?Q?b?=", "ab"},
 138                 {"=?ISO-8859-1?Q?a?=  =?ISO-8859-1?Q?b?=", "ab"},
 139                 {"=?ISO-8859-1?Q?a?= \r\n\t =?ISO-8859-1?Q?b?=", "ab"},
 140                 {"=?ISO-8859-1?Q?a_b?=", "a b"},
 141         }
 142
 143         for _, test := range tests {
 144                 dec := new(WordDecoder)
 145                 s, err := dec.DecodeHeader(test.src)
 146                 if err != nil {
 147                         t.Errorf("DecodeHeader(%q): %v", test.src, err)
 148                 }
 149                 if s != test.exp {
 150                         t.Errorf("DecodeHeader(%q) = %q, want %q", test.src, s, test.exp)
 151                 }
 152         }
 153 }
 154
 155 func TestCharsetDecoder(t *testing.T) {
 156         tests := []struct {
 157                 src      string
 158                 want     string
 159                 charsets []string
 160                 content  []string
 161         }{
 162                 {"=?utf-8?b?Q2Fmw6k=?=", "Café", nil, nil},
 163                 {"=?ISO-8859-1?Q?caf=E9?=", "café", nil, nil},
 164                 {"=?US-ASCII?Q?foo_bar?=", "foo bar", nil, nil},
 165                 {"=?utf-8?Q?=?=", "=?utf-8?Q?=?=", nil, nil},
 166                 {"=?utf-8?Q?=A?=", "=?utf-8?Q?=A?=", nil, nil},
 167                 {
 168                         "=?ISO-8859-15?Q?f=F5=F6?=  =?windows-1252?Q?b=E0r?=",
 169                         "f\xf5\xf6b\xe0r",
 170                         []string{"iso-8859-15", "windows-1252"},
 171                         []string{"f\xf5\xf6", "b\xe0r"},
 172                 },
 173         }
 174
 175         for _, test := range tests {
 176                 i := 0
 177                 dec := &WordDecoder{
 178                         CharsetReader: func(charset string, input io.Reader) (io.Reader, error) {
 179                                 if charset != test.charsets[i] {
 180                                         t.Errorf("DecodeHeader(%q), got charset %q, want %q", test.src, charset, test.charsets[i])
 181                                 }
 182                                 content, err := ioutil.ReadAll(input)
 183                                 if err != nil {
 184                                         t.Errorf("DecodeHeader(%q), error in reader: %v", test.src, err)
 185                                 }
 186                                 got := string(content)
 187                                 if got != test.content[i] {
 188                                         t.Errorf("DecodeHeader(%q), got content %q, want %q", test.src, got, test.content[i])
 189                                 }
 190                                 i++
 191
 192                                 return strings.NewReader(got), nil
 193                         },
 194                 }
 195                 got, err := dec.DecodeHeader(test.src)
 196                 if err != nil {
 197                         t.Errorf("DecodeHeader(%q): %v", test.src, err)
 198                 }
 199                 if got != test.want {
 200                         t.Errorf("DecodeHeader(%q) = %q, want %q", test.src, got, test.want)
 201                 }
 202         }
 203 }
 204
 205 func TestCharsetDecoderError(t *testing.T) {
 206         dec := &WordDecoder{
 207                 CharsetReader: func(charset string, input io.Reader) (io.Reader, error) {
 208                         return nil, errors.New("Test error")
 209                 },
 210         }
 211
 212         if _, err := dec.DecodeHeader("=?charset?Q?foo?="); err == nil {
 213                 t.Error("DecodeHeader should return an error")
 214         }
 215 }
 216
 217 func BenchmarkQEncodeWord(b *testing.B) {
 218         for i := 0; i < b.N; i++ {
 219                 QEncoding.Encode("UTF-8", "¡Hola, señor!")
 220         }
 221 }
 222
 223 func BenchmarkQDecodeWord(b *testing.B) {
 224         dec := new(WordDecoder)
 225
 226         for i := 0; i < b.N; i++ {
 227                 dec.Decode("=?utf-8?q?=C2=A1Hola,_se=C3=B1or!?=")
 228         }
 229 }
 230
 231 func BenchmarkQDecodeHeader(b *testing.B) {
 232         dec := new(WordDecoder)
 233
 234         for i := 0; i < b.N; i++ {
 235                 dec.DecodeHeader("=?utf-8?q?=C2=A1Hola,_se=C3=B1or!?=")
 236         }
 237 }