1
2
3
4
5 package charset
6
7 import (
8 "bytes"
9 "encoding/xml"
10 "io"
11 "os"
12 "runtime"
13 "strings"
14 "testing"
15
16 "golang.org/x/text/transform"
17 )
18
19 func transformString(t transform.Transformer, s string) (string, error) {
20 r := transform.NewReader(strings.NewReader(s), t)
21 b, err := io.ReadAll(r)
22 return string(b), err
23 }
24
25 type testCase struct {
26 utf8, other, otherEncoding string
27 }
28
29
30 var testCases = []testCase{
31 {"Résumé", "Résumé", "utf8"},
32 {"Résumé", "R\xe9sum\xe9", "latin1"},
33 {"これは漢字です。", "S0\x8c0o0\"oW[g0Y0\x020", "UTF-16LE"},
34 {"これは漢字です。", "0S0\x8c0oo\"[W0g0Y0\x02", "UTF-16BE"},
35 {"Hello, world", "Hello, world", "ASCII"},
36 {"Gdańsk", "Gda\xf1sk", "ISO-8859-2"},
37 {"Ââ Čč Đđ Ŋŋ Õõ Šš Žž Åå Ää", "\xc2\xe2 \xc8\xe8 \xa9\xb9 \xaf\xbf \xd5\xf5 \xaa\xba \xac\xbc \xc5\xe5 \xc4\xe4", "ISO-8859-10"},
38 {"สำหรับ", "\xca\xd3\xcb\xc3\u047a", "ISO-8859-11"},
39 {"latviešu", "latvie\xf0u", "ISO-8859-13"},
40 {"Seònaid", "Se\xf2naid", "ISO-8859-14"},
41 {"€1 is cheap", "\xa41 is cheap", "ISO-8859-15"},
42 {"românește", "rom\xe2ne\xbate", "ISO-8859-16"},
43 {"nutraĵo", "nutra\xbco", "ISO-8859-3"},
44 {"Kalâdlit", "Kal\xe2dlit", "ISO-8859-4"},
45 {"русский", "\xe0\xe3\xe1\xe1\xda\xd8\xd9", "ISO-8859-5"},
46 {"ελληνικά", "\xe5\xeb\xeb\xe7\xed\xe9\xea\xdc", "ISO-8859-7"},
47 {"Kağan", "Ka\xf0an", "ISO-8859-9"},
48 {"Résumé", "R\x8esum\x8e", "macintosh"},
49 {"Gdańsk", "Gda\xf1sk", "windows-1250"},
50 {"русский", "\xf0\xf3\xf1\xf1\xea\xe8\xe9", "windows-1251"},
51 {"Résumé", "R\xe9sum\xe9", "windows-1252"},
52 {"ελληνικά", "\xe5\xeb\xeb\xe7\xed\xe9\xea\xdc", "windows-1253"},
53 {"Kağan", "Ka\xf0an", "windows-1254"},
54 {"עִבְרִית", "\xf2\xc4\xe1\xc0\xf8\xc4\xe9\xfa", "windows-1255"},
55 {"العربية", "\xc7\xe1\xda\xd1\xc8\xed\xc9", "windows-1256"},
56 {"latviešu", "latvie\xf0u", "windows-1257"},
57 {"Việt", "Vi\xea\xf2t", "windows-1258"},
58 {"สำหรับ", "\xca\xd3\xcb\xc3\u047a", "windows-874"},
59 {"русский", "\xd2\xd5\xd3\xd3\xcb\xc9\xca", "KOI8-R"},
60 {"українська", "\xd5\xcb\xd2\xc1\xa7\xce\xd3\xd8\xcb\xc1", "KOI8-U"},
61 {"Hello 常用國字標準字體表", "Hello \xb1`\xa5\u03b0\xea\xa6r\xbc\u0437\u01e6r\xc5\xe9\xaa\xed", "big5"},
62 {"Hello 常用國字標準字體表", "Hello \xb3\xa3\xd3\xc3\x87\xf8\xd7\xd6\x98\xcb\x9c\xca\xd7\xd6\xf3\x77\xb1\xed", "gbk"},
63 {"Hello 常用國字標準字體表", "Hello \xb3\xa3\xd3\xc3\x87\xf8\xd7\xd6\x98\xcb\x9c\xca\xd7\xd6\xf3\x77\xb1\xed", "gb18030"},
64 {"עִבְרִית", "\x81\x30\xfb\x30\x81\x30\xf6\x34\x81\x30\xf9\x33\x81\x30\xf6\x30\x81\x30\xfb\x36\x81\x30\xf6\x34\x81\x30\xfa\x31\x81\x30\xfb\x38", "gb18030"},
65 {"㧯", "\x82\x31\x89\x38", "gb18030"},
66 {"これは漢字です。", "\x82\xb1\x82\xea\x82\xcd\x8a\xbf\x8e\x9a\x82\xc5\x82\xb7\x81B", "SJIS"},
67 {"Hello, 世界!", "Hello, \x90\xa2\x8aE!", "SJIS"},
68 {"ｲｳｴｵｶ", "\xb2\xb3\xb4\xb5\xb6", "SJIS"},
69 {"これは漢字です。", "\xa4\xb3\xa4\xec\xa4\u03f4\xc1\xbb\xfa\xa4\u01e4\xb9\xa1\xa3", "EUC-JP"},
70 {"Hello, 世界!", "Hello, \x1b$B@$3&\x1b(B!", "ISO-2022-JP"},
71 {"다음과 같은 조건을 따라야 합니다: 저작자표시", "\xb4\xd9\xc0\xbd\xb0\xfa \xb0\xb0\xc0\xba \xc1\xb6\xb0\xc7\xc0\xbb \xb5\xfb\xb6\xf3\xbe\xdf \xc7մϴ\xd9: \xc0\xfa\xc0\xdb\xc0\xdaǥ\xbd\xc3", "EUC-KR"},
72 }
73
74 func TestDecode(t *testing.T) {
75 testCases := append(testCases, []testCase{
76
77
78 {"Rés\ufffdumé", "Rés\xe1\x80umé", "utf8"},
79 }...)
80 for _, tc := range testCases {
81 e, _ := Lookup(tc.otherEncoding)
82 if e == nil {
83 t.Errorf("%s: not found", tc.otherEncoding)
84 continue
85 }
86 s, err := transformString(e.NewDecoder(), tc.other)
87 if err != nil {
88 t.Errorf("%s: decode %q: %v", tc.otherEncoding, tc.other, err)
89 continue
90 }
91 if s != tc.utf8 {
92 t.Errorf("%s: got %q, want %q", tc.otherEncoding, s, tc.utf8)
93 }
94 }
95 }
96
97 func TestEncode(t *testing.T) {
98 testCases := append(testCases, []testCase{
99
100 {"Rés\xe1\x80umé", "Rés\ufffd\ufffdumé", "utf8"},
101
102 {"Gdańsk", "Gdańsk", "ISO-8859-11"},
103 {"\ufffd", "�", "ISO-8859-11"},
104 {"a\xe1\x80b", "a��b", "ISO-8859-11"},
105 }...)
106 for _, tc := range testCases {
107 e, _ := Lookup(tc.otherEncoding)
108 if e == nil {
109 t.Errorf("%s: not found", tc.otherEncoding)
110 continue
111 }
112 s, err := transformString(e.NewEncoder(), tc.utf8)
113 if err != nil {
114 t.Errorf("%s: encode %q: %s", tc.otherEncoding, tc.utf8, err)
115 continue
116 }
117 if s != tc.other {
118 t.Errorf("%s: got %q, want %q", tc.otherEncoding, s, tc.other)
119 }
120 }
121 }
122
123 var sniffTestCases = []struct {
124 filename, declared, want string
125 }{
126 {"HTTP-charset.html", "text/html; charset=iso-8859-15", "iso-8859-15"},
127 {"UTF-16LE-BOM.html", "", "utf-16le"},
128 {"UTF-16BE-BOM.html", "", "utf-16be"},
129 {"meta-content-attribute.html", "text/html", "iso-8859-15"},
130 {"meta-charset-attribute.html", "text/html", "iso-8859-15"},
131 {"No-encoding-declaration.html", "text/html", "utf-8"},
132 {"HTTP-vs-UTF-8-BOM.html", "text/html; charset=iso-8859-15", "utf-8"},
133 {"HTTP-vs-meta-content.html", "text/html; charset=iso-8859-15", "iso-8859-15"},
134 {"HTTP-vs-meta-charset.html", "text/html; charset=iso-8859-15", "iso-8859-15"},
135 {"UTF-8-BOM-vs-meta-content.html", "text/html", "utf-8"},
136 {"UTF-8-BOM-vs-meta-charset.html", "text/html", "utf-8"},
137 }
138
139 func TestSniff(t *testing.T) {
140 switch runtime.GOOS {
141 case "nacl":
142 t.Skipf("not supported on %q", runtime.GOOS)
143 }
144
145 for _, tc := range sniffTestCases {
146 content, err := os.ReadFile("testdata/" + tc.filename)
147 if err != nil {
148 t.Errorf("%s: error reading file: %v", tc.filename, err)
149 continue
150 }
151
152 _, name, _ := DetermineEncoding(content, tc.declared)
153 if name != tc.want {
154 t.Errorf("%s: got %q, want %q", tc.filename, name, tc.want)
155 continue
156 }
157 }
158 }
159
160 func TestReader(t *testing.T) {
161 switch runtime.GOOS {
162 case "nacl":
163 t.Skipf("not supported on %q", runtime.GOOS)
164 }
165
166 for _, tc := range sniffTestCases {
167 content, err := os.ReadFile("testdata/" + tc.filename)
168 if err != nil {
169 t.Errorf("%s: error reading file: %v", tc.filename, err)
170 continue
171 }
172
173 r, err := NewReader(bytes.NewReader(content), tc.declared)
174 if err != nil {
175 t.Errorf("%s: error creating reader: %v", tc.filename, err)
176 continue
177 }
178
179 got, err := io.ReadAll(r)
180 if err != nil {
181 t.Errorf("%s: error reading from charset.NewReader: %v", tc.filename, err)
182 continue
183 }
184
185 e, _ := Lookup(tc.want)
186 want, err := io.ReadAll(transform.NewReader(bytes.NewReader(content), e.NewDecoder()))
187 if err != nil {
188 t.Errorf("%s: error decoding with hard-coded charset name: %v", tc.filename, err)
189 continue
190 }
191
192 if !bytes.Equal(got, want) {
193 t.Errorf("%s: got %q, want %q", tc.filename, got, want)
194 continue
195 }
196 }
197 }
198
199 var metaTestCases = []struct {
200 meta, want string
201 }{
202 {"", ""},
203 {"text/html", ""},
204 {"text/html; charset utf-8", ""},
205 {"text/html; charset=latin-2", "latin-2"},
206 {"text/html; charset; charset = utf-8", "utf-8"},
207 {`charset="big5"`, "big5"},
208 {"charset='shift_jis'", "shift_jis"},
209 }
210
211 func TestFromMeta(t *testing.T) {
212 for _, tc := range metaTestCases {
213 got := fromMetaElement(tc.meta)
214 if got != tc.want {
215 t.Errorf("%q: got %q, want %q", tc.meta, got, tc.want)
216 }
217 }
218 }
219
220 func TestXML(t *testing.T) {
221 const s = "<?xml version=\"1.0\" encoding=\"windows-1252\"?><a><Word>r\xe9sum\xe9</Word></a>"
222
223 d := xml.NewDecoder(strings.NewReader(s))
224 d.CharsetReader = NewReaderLabel
225
226 var a struct {
227 Word string
228 }
229 if err := d.Decode(&a); err != nil {
230 t.Fatalf("Decode: %v", err)
231 }
232
233 want := "résumé"
234 if a.Word != want {
235 t.Errorf("got %q, want %q", a.Word, want)
236 }
237 }
238
View as plain text