|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of the ZBateson\MailMimeParser project. |
|
4
|
|
|
* |
|
5
|
|
|
* @license http => //opensource.org/licenses/bsd-license.php BSD |
|
6
|
|
|
*/ |
|
7
|
|
|
namespace ZBateson\MailMimeParser\Stream\Helper; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Helper class for converting strings between charsets. |
|
11
|
|
|
* |
|
12
|
|
|
* CharasetConverter tries to convert using mb_convert_encoding when possible, |
|
13
|
|
|
* defining as many aliases as possible for supported encodings. If not |
|
14
|
|
|
* supported, iconv is attempted. |
|
15
|
|
|
* |
|
16
|
|
|
* @author Zaahid Bateson |
|
17
|
|
|
*/ |
|
18
|
|
|
class CharsetConverter |
|
19
|
|
|
{ |
|
20
|
|
|
/** |
|
21
|
|
|
* @var array aliased charsets supported by mb_convert_encoding. |
|
22
|
|
|
* The alias is stripped of any non-alphanumeric characters (so CP367 |
|
23
|
|
|
* is equal to CP-367) when comparing. |
|
24
|
|
|
* Some of these translations are already supported by |
|
25
|
|
|
* mb_convert_encoding on "my" PHP 5.5.9, but may not be supported in |
|
26
|
|
|
* other implementations or versions since they're not part of |
|
27
|
|
|
* documented support. |
|
28
|
|
|
*/ |
|
29
|
|
|
public static $mbAliases = [ |
|
30
|
|
|
// supported but not included in mb_list_encodings for some reason... |
|
31
|
|
|
'CP850' => 'CP850', |
|
32
|
|
|
'GB2312' => 'GB2312', |
|
33
|
|
|
// aliases |
|
34
|
|
|
'646' => 'ASCII', |
|
35
|
|
|
'ANSIX341968' => 'ASCII', |
|
36
|
|
|
'ANSIX341986' => 'ASCII', |
|
37
|
|
|
'CP367' => 'ASCII', |
|
38
|
|
|
'CSASCII' => 'ASCII', |
|
39
|
|
|
'IBM367' => 'ASCII', |
|
40
|
|
|
'ISO646US' => 'ASCII', |
|
41
|
|
|
'ISO646IRV1991' => 'ASCII', |
|
42
|
|
|
'ISOIR6' => 'ASCII', |
|
43
|
|
|
'US' => 'ASCII', |
|
44
|
|
|
'USASCII' => 'ASCII', |
|
45
|
|
|
'BIG5' => 'BIG-5', |
|
46
|
|
|
'BIG5TW' => 'BIG-5', |
|
47
|
|
|
'CSBIG5' => 'BIG-5', |
|
48
|
|
|
'1251' => 'WINDOWS-1251', |
|
49
|
|
|
'CP1251' => 'WINDOWS-1251', |
|
50
|
|
|
'WINDOWS1251' => 'WINDOWS-1251', |
|
51
|
|
|
'1252' => 'WINDOWS-1252', |
|
52
|
|
|
'CP1252' => 'WINDOWS-1252', |
|
53
|
|
|
'WINDOWS1252' => 'WINDOWS-1252', |
|
54
|
|
|
'WE8MSWIN1252' => 'WINDOWS-1252', |
|
55
|
|
|
'1254' => 'WINDOWS-1254', |
|
56
|
|
|
'CP1254' => 'WINDOWS-1254', |
|
57
|
|
|
'WINDOWS1254' => 'WINDOWS-1254', |
|
58
|
|
|
'1255' => 'ISO-8859-8', |
|
59
|
|
|
'CP1255' => 'ISO-8859-8', |
|
60
|
|
|
'ISO88598I' => 'ISO-8859-8', |
|
61
|
|
|
'WINDOWS1255' => 'ISO-8859-8', |
|
62
|
|
|
'850' => 'CP850', |
|
63
|
|
|
'CSPC850MULTILINGUAL' => 'CP850', |
|
64
|
|
|
'IBM850' => 'CP850', |
|
65
|
|
|
'866' => 'CP866', |
|
66
|
|
|
'CSIBM866' => 'CP866', |
|
67
|
|
|
'IBM866' => 'CP866', |
|
68
|
|
|
'932' => 'CP932', |
|
69
|
|
|
'MS932' => 'CP932', |
|
70
|
|
|
'MSKANJI' => 'CP932', |
|
71
|
|
|
'950' => 'CP950', |
|
72
|
|
|
'MS950' => 'CP950', |
|
73
|
|
|
'EUCJP' => 'EUC-JP', |
|
74
|
|
|
'UJIS' => 'EUC-JP', |
|
75
|
|
|
'EUCKR' => 'EUC-KR', |
|
76
|
|
|
'KOREAN' => 'EUC-KR', |
|
77
|
|
|
'KSC5601' => 'EUC-KR', |
|
78
|
|
|
'KSC56011987' => 'EUC-KR', |
|
79
|
|
|
'KSX1001' => 'EUC-KR', |
|
80
|
|
|
'GB180302000' => 'GB18030', |
|
81
|
|
|
// GB2312 not listed but supported |
|
82
|
|
|
'CHINESE' => 'GB2312', |
|
83
|
|
|
'CSISO58GB231280' => 'GB2312', |
|
84
|
|
|
'EUCCN' => 'GB2312', |
|
85
|
|
|
'EUCGB2312CN' => 'GB2312', |
|
86
|
|
|
'GB23121980' => 'GB2312', |
|
87
|
|
|
'GB231280' => 'GB2312', |
|
88
|
|
|
'ISOIR58' => 'GB2312', |
|
89
|
|
|
'GBK' => 'CP936', |
|
90
|
|
|
'936' => 'CP936', |
|
91
|
|
|
'ms936' => 'CP936', |
|
92
|
|
|
'HZGB' => 'HZ', |
|
93
|
|
|
'HZGB2312' => 'HZ', |
|
94
|
|
|
'CSISO2022JP' => 'ISO-2022-JP', |
|
95
|
|
|
'ISO2022JP' => 'ISO-2022-JP', |
|
96
|
|
|
'ISO2022JP2004' => 'ISO-2022-JP-2004', |
|
97
|
|
|
'CSISO2022KR' => 'ISO-2022-KR', |
|
98
|
|
|
'ISO2022KR' => 'ISO-2022-KR', |
|
99
|
|
|
'CSISOLATIN6' => 'ISO-8859-10', |
|
100
|
|
|
'ISO885910' => 'ISO-8859-10', |
|
101
|
|
|
'ISO8859101992' => 'ISO-8859-10', |
|
102
|
|
|
'ISOIR157' => 'ISO-8859-10', |
|
103
|
|
|
'L6' => 'ISO-8859-10', |
|
104
|
|
|
'LATIN6' => 'ISO-8859-10', |
|
105
|
|
|
'ISO885913' => 'ISO-8859-13', |
|
106
|
|
|
'ISO885914' => 'ISO-8859-14', |
|
107
|
|
|
'ISO8859141998' => 'ISO-8859-14', |
|
108
|
|
|
'ISOCELTIC' => 'ISO-8859-14', |
|
109
|
|
|
'ISOIR199' => 'ISO-8859-14', |
|
110
|
|
|
'L8' => 'ISO-8859-14', |
|
111
|
|
|
'LATIN8' => 'ISO-8859-14', |
|
112
|
|
|
'ISO885915' => 'ISO-8859-15', |
|
113
|
|
|
'ISO885916' => 'ISO-8859-16', |
|
114
|
|
|
'ISO8859162001' => 'ISO-8859-16', |
|
115
|
|
|
'ISOIR226' => 'ISO-8859-16', |
|
116
|
|
|
'L10' => 'ISO-8859-16', |
|
117
|
|
|
'LATIN10' => 'ISO-8859-16', |
|
118
|
|
|
'CSISOLATIN2' => 'ISO-8859-2', |
|
119
|
|
|
'ISO88592' => 'ISO-8859-2', |
|
120
|
|
|
'ISO885921987' => 'ISO-8859-2', |
|
121
|
|
|
'ISOIR101' => 'ISO-8859-2', |
|
122
|
|
|
'L2' => 'ISO-8859-2', |
|
123
|
|
|
'LATIN2' => 'ISO-8859-2', |
|
124
|
|
|
'CSISOLATIN3' => 'ISO-8859-3', |
|
125
|
|
|
'ISO88593' => 'ISO-8859-3', |
|
126
|
|
|
'ISO885931988' => 'ISO-8859-3', |
|
127
|
|
|
'ISOIR109' => 'ISO-8859-3', |
|
128
|
|
|
'L3' => 'ISO-8859-3', |
|
129
|
|
|
'LATIN3' => 'ISO-8859-3', |
|
130
|
|
|
'CSISOLATIN4' => 'ISO-8859-4', |
|
131
|
|
|
'ISO88594' => 'ISO-8859-4', |
|
132
|
|
|
'ISO885941988' => 'ISO-8859-4', |
|
133
|
|
|
'ISOIR110' => 'ISO-8859-4', |
|
134
|
|
|
'L4' => 'ISO-8859-4', |
|
135
|
|
|
'LATIN4' => 'ISO-8859-4', |
|
136
|
|
|
'CSISOLATINCYRILLIC' => 'ISO-8859-5', |
|
137
|
|
|
'CYRILLIC' => 'ISO-8859-5', |
|
138
|
|
|
'ISO88595' => 'ISO-8859-5', |
|
139
|
|
|
'ISO885951988' => 'ISO-8859-5', |
|
140
|
|
|
'ISOIR144' => 'ISO-8859-5', |
|
141
|
|
|
'ARABIC' => 'ISO-8859-6', |
|
142
|
|
|
'ASMO708' => 'ISO-8859-6', |
|
143
|
|
|
'CSISOLATINARABIC' => 'ISO-8859-6', |
|
144
|
|
|
'ECMA114' => 'ISO-8859-6', |
|
145
|
|
|
'ISO88596' => 'ISO-8859-6', |
|
146
|
|
|
'ISO885961987' => 'ISO-8859-6', |
|
147
|
|
|
'ISOIR127' => 'ISO-8859-6', |
|
148
|
|
|
'CSISOLATINGREEK' => 'ISO-8859-7', |
|
149
|
|
|
'ECMA118' => 'ISO-8859-7', |
|
150
|
|
|
'ELOT928' => 'ISO-8859-7', |
|
151
|
|
|
'GREEK' => 'ISO-8859-7', |
|
152
|
|
|
'GREEK8' => 'ISO-8859-7', |
|
153
|
|
|
'ISO88597' => 'ISO-8859-7', |
|
154
|
|
|
'ISO885971987' => 'ISO-8859-7', |
|
155
|
|
|
'ISOIR126' => 'ISO-8859-7', |
|
156
|
|
|
'CSISOLATINHEBREW' => 'ISO-8859-8', |
|
157
|
|
|
'HEBREW' => 'ISO-8859-8', |
|
158
|
|
|
'ISO88598' => 'ISO-8859-8', |
|
159
|
|
|
'ISO885981988' => 'ISO-8859-8', |
|
160
|
|
|
'ISOIR138' => 'ISO-8859-8', |
|
161
|
|
|
'CSISOLATIN5' => 'ISO-8859-9', |
|
162
|
|
|
'ISO88599' => 'ISO-8859-9', |
|
163
|
|
|
'ISO885991989' => 'ISO-8859-9', |
|
164
|
|
|
'ISOIR148' => 'ISO-8859-9', |
|
165
|
|
|
'L5' => 'ISO-8859-9', |
|
166
|
|
|
'LATIN5' => 'ISO-8859-9', |
|
167
|
|
|
'CSKOI8R' => 'KOI8-R', |
|
168
|
|
|
'KOI8R' => 'KOI8-R', |
|
169
|
|
|
'8859' => 'ISO-8859-1', |
|
170
|
|
|
'CP819' => 'ISO-8859-1', |
|
171
|
|
|
'CSISOLATIN1' => 'ISO-8859-1', |
|
172
|
|
|
'IBM819' => 'ISO-8859-1', |
|
173
|
|
|
'ISO8859' => 'ISO-8859-1', |
|
174
|
|
|
'ISO88591' => 'ISO-8859-1', |
|
175
|
|
|
'ISO885911987' => 'ISO-8859-1', |
|
176
|
|
|
'ISOIR100' => 'ISO-8859-1', |
|
177
|
|
|
'L1' => 'ISO-8859-1', |
|
178
|
|
|
'LATIN' => 'ISO-8859-1', |
|
179
|
|
|
'LATIN1' => 'ISO-8859-1', |
|
180
|
|
|
'CSSHIFTJIS' => 'SJIS', |
|
181
|
|
|
'SHIFTJIS' => 'SJIS', |
|
182
|
|
|
'SHIFTJIS2004' => 'SJIS-2004', |
|
183
|
|
|
'SJIS2004' => 'SJIS-2004', |
|
184
|
|
|
// Microsoft charset values |
|
185
|
|
|
'0' => 'WINDOWS-1252', |
|
186
|
|
|
'128' => 'SJIS', |
|
187
|
|
|
'129' => 'EUC-KR', |
|
188
|
|
|
'134' => 'GB2312', |
|
189
|
|
|
'136' => 'BIG-5', |
|
190
|
|
|
'161' => 'WINDOWS-1253', |
|
191
|
|
|
'162' => 'WINDOWS-1254', |
|
192
|
|
|
'177' => 'WINDOWS-1255', |
|
193
|
|
|
'178' => 'WINDOWS-1256', |
|
194
|
|
|
'186' => 'WINDOWS-1257', |
|
195
|
|
|
'204' => 'WINDOWS-1251', |
|
196
|
|
|
'222' => 'WINDOWS-874', |
|
197
|
|
|
'238' => 'WINDOWS-1250', |
|
198
|
|
|
]; |
|
199
|
|
|
|
|
200
|
|
|
/** |
|
201
|
|
|
* @var array aliased charsets supported by iconv. |
|
202
|
|
|
*/ |
|
203
|
|
|
public static $iconvAliases = [ |
|
204
|
|
|
// iconv aliases -- a lot of these may already be supported |
|
205
|
|
|
'BIG5HKSCS' => 'BIG5HKSCS', |
|
206
|
|
|
'HKSCS' => 'BIG5HKSCS', |
|
207
|
|
|
'037' => 'CP037', |
|
208
|
|
|
'EBCDICCPCA' => 'CP037', |
|
209
|
|
|
'EBCDICCPNL' => 'CP037', |
|
210
|
|
|
'EBCDICCPUS' => 'CP037', |
|
211
|
|
|
'EBCDICCPWT' => 'CP037', |
|
212
|
|
|
'CSIBM037' => 'CP037', |
|
213
|
|
|
'IBM037' => 'CP037', |
|
214
|
|
|
'IBM039' => 'CP037', |
|
215
|
|
|
'1026' => 'CP1026', |
|
216
|
|
|
'CSIBM1026' => 'CP1026', |
|
217
|
|
|
'IBM1026' => 'CP1026', |
|
218
|
|
|
'1140' => 'CP1140', |
|
219
|
|
|
'IBM1140' => 'CP1140', |
|
220
|
|
|
'1250' => 'CP1250', |
|
221
|
|
|
'WINDOWS1250' => 'CP1250', |
|
222
|
|
|
'1253' => 'CP1253', |
|
223
|
|
|
'WINDOWS1253' => 'CP1253', |
|
224
|
|
|
'1256' => 'CP1256', |
|
225
|
|
|
'WINDOWS1256' => 'CP1256', |
|
226
|
|
|
'1257' => 'CP1257', |
|
227
|
|
|
'WINDOWS1257' => 'CP1257', |
|
228
|
|
|
'1258' => 'CP1258', |
|
229
|
|
|
'WINDOWS1258' => 'CP1258', |
|
230
|
|
|
'424' => 'CP424', |
|
231
|
|
|
'CSIBM424' => 'CP424', |
|
232
|
|
|
'EBCDICCPHE' => 'CP424', |
|
233
|
|
|
'IBM424' => 'CP424', |
|
234
|
|
|
'437' => 'CP437', |
|
235
|
|
|
'CSPC8CODEPAGE437' => 'CP437', |
|
236
|
|
|
'IBM437' => 'CP437', |
|
237
|
|
|
'500' => 'CP500', |
|
238
|
|
|
'CSIBM500' => 'CP500', |
|
239
|
|
|
'EBCDICCPBE' => 'CP500', |
|
240
|
|
|
'EBCDICCPCH' => 'CP500', |
|
241
|
|
|
'IBM500' => 'CP500', |
|
242
|
|
|
'775' => 'CP775', |
|
243
|
|
|
'CSPC775BALTIC' => 'CP775', |
|
244
|
|
|
'IBM775' => 'CP775', |
|
245
|
|
|
'860' => 'CP860', |
|
246
|
|
|
'CSIBM860' => 'CP860', |
|
247
|
|
|
'IBM860' => 'CP860', |
|
248
|
|
|
'861' => 'CP861', |
|
249
|
|
|
'CPIS' => 'CP861', |
|
250
|
|
|
'CSIBM861' => 'CP861', |
|
251
|
|
|
'IBM861' => 'CP861', |
|
252
|
|
|
'862' => 'CP862', |
|
253
|
|
|
'CSPC862LATINHEBREW' => 'CP862', |
|
254
|
|
|
'IBM862' => 'CP862', |
|
255
|
|
|
'863' => 'CP863', |
|
256
|
|
|
'CSIBM863' => 'CP863', |
|
257
|
|
|
'IBM863' => 'CP863', |
|
258
|
|
|
'864' => 'CP864', |
|
259
|
|
|
'CSIBM864' => 'CP864', |
|
260
|
|
|
'IBM864' => 'CP864', |
|
261
|
|
|
'865' => 'CP865', |
|
262
|
|
|
'CSIBM865' => 'CP865', |
|
263
|
|
|
'IBM865' => 'CP865', |
|
264
|
|
|
'869' => 'CP869', |
|
265
|
|
|
'CPGR' => 'CP869', |
|
266
|
|
|
'CSIBM869' => 'CP869', |
|
267
|
|
|
'IBM869' => 'CP869', |
|
268
|
|
|
'949' => 'CP949', |
|
269
|
|
|
'MS949' => 'CP949', |
|
270
|
|
|
'UHC' => 'CP949', |
|
271
|
|
|
'ROMAN8' => 'ROMAN8', |
|
272
|
|
|
'HPROMAN8' => 'ROMAN8', |
|
273
|
|
|
'R8' => 'ROMAN8', |
|
274
|
|
|
'CSHPROMAN8' => 'ROMAN8', |
|
275
|
|
|
'ISO2022JP2' => 'ISO2022JP2', |
|
276
|
|
|
'THAI' => 'ISO885911', |
|
277
|
|
|
'ISO885911' => 'ISO885911', |
|
278
|
|
|
'ISO8859112001' => 'ISO885911', |
|
279
|
|
|
'JOHAB' => 'CP1361', |
|
280
|
|
|
'MS1361' => 'CP1361', |
|
281
|
|
|
'MACCYRILLIC' => 'MACCYRILLIC', |
|
282
|
|
|
'CSPTCP154' => 'PT154', |
|
283
|
|
|
'PTCP154' => 'PT154', |
|
284
|
|
|
'CP154' => 'PT154', |
|
285
|
|
|
'CYRILLICASIAN' => 'PT154', |
|
286
|
|
|
'TIS620' => 'TIS620', |
|
287
|
|
|
'TIS6200' => 'TIS620', |
|
288
|
|
|
'TIS62025290' => 'TIS620', |
|
289
|
|
|
'TIS62025291' => 'TIS620', |
|
290
|
|
|
'ISOIR166' => 'TIS620', |
|
291
|
|
|
]; |
|
292
|
|
|
|
|
293
|
|
|
/** |
|
294
|
|
|
* @var string charset to convert from |
|
295
|
|
|
*/ |
|
296
|
|
|
protected $fromCharset; |
|
297
|
|
|
|
|
298
|
|
|
/** |
|
299
|
|
|
* @var string charset to convert to |
|
300
|
|
|
*/ |
|
301
|
|
|
protected $toCharset; |
|
302
|
|
|
|
|
303
|
|
|
/** |
|
304
|
|
|
* @var boolean indicates if $fromCharset is supported by |
|
305
|
|
|
* mb_convert_encoding |
|
306
|
|
|
*/ |
|
307
|
|
|
protected $fromCharsetMbSupported = true; |
|
308
|
|
|
|
|
309
|
|
|
/** |
|
310
|
|
|
* @var boolean indicates if $toCharset is supported by mb_convert_encoding |
|
311
|
|
|
*/ |
|
312
|
|
|
protected $toCharsetMbSupported = true; |
|
313
|
|
|
|
|
314
|
|
|
/** |
|
315
|
|
|
* Constructs the charset converter with source/destination charsets. |
|
316
|
|
|
* |
|
317
|
|
|
* @param string $fromCharset |
|
318
|
|
|
* @param string $toCharset |
|
319
|
|
|
*/ |
|
320
|
3 |
|
public function __construct($fromCharset, $toCharset) |
|
321
|
|
|
{ |
|
322
|
3 |
|
$this->fromCharset = $this->findSupportedCharset($fromCharset, $this->fromCharsetMbSupported); |
|
323
|
3 |
|
$this->toCharset = $this->findSupportedCharset($toCharset, $this->toCharsetMbSupported); |
|
324
|
3 |
|
} |
|
325
|
|
|
|
|
326
|
|
|
/** |
|
327
|
|
|
* Converts the passed string's charset from $this->fromCharset to |
|
328
|
|
|
* $this->toCharset. |
|
329
|
|
|
* |
|
330
|
|
|
* The function attempts to use mb_convert_encoding if possible, and falls |
|
331
|
|
|
* back to iconv if not. If the source or destination character sets aren't |
|
332
|
|
|
* supported, a blank string is returned. |
|
333
|
|
|
* |
|
334
|
|
|
* @param string $str |
|
335
|
|
|
* @return string |
|
336
|
|
|
*/ |
|
337
|
3 |
|
public function convert($str) |
|
338
|
|
|
{ |
|
339
|
|
|
// there may be some mb-supported encodings not supported by iconv (on my libiconv for instance |
|
340
|
|
|
// HZ isn't supported), and so it may happen that failing an mb_convert_encoding, an iconv |
|
341
|
|
|
// may also fail even though both support an encoding separately. |
|
342
|
|
|
// Unfortunately there's no great way of testing what charsets are available on iconv, and |
|
343
|
|
|
// attempting to blindly convert the string may be too costly, as could converting first |
|
344
|
|
|
// to an intermediate (ASSUMPTION: may be worth testing converting to an intermediate) |
|
345
|
3 |
|
if ($str !== '') { |
|
346
|
3 |
|
if ($this->fromCharsetMbSupported && $this->toCharsetMbSupported) { |
|
347
|
2 |
|
return mb_convert_encoding($str, $this->toCharset, $this->fromCharset); |
|
348
|
|
|
} |
|
349
|
2 |
|
return iconv($this->fromCharset, $this->toCharset . '//TRANSLIT//IGNORE', $str); |
|
350
|
|
|
} |
|
351
|
|
|
return $str; |
|
352
|
|
|
} |
|
353
|
|
|
|
|
354
|
|
|
/** |
|
355
|
|
|
* Looks up the passed $cs in mb_list_encodings, then strips non |
|
356
|
|
|
* alpha-numeric characters and tries again, then failing that calls |
|
357
|
|
|
* findAliasedCharset. The method returns the charset name that should be |
|
358
|
|
|
* used in calls to mb_convert_encoding or iconv. |
|
359
|
|
|
* |
|
360
|
|
|
* If the charset is part of mb_list_encodings, $mbSupported is set to true. |
|
361
|
|
|
* |
|
362
|
|
|
* @param string $cs |
|
363
|
|
|
* @param boolean $mbSupported |
|
364
|
|
|
* @return string the final charset name to use |
|
365
|
|
|
*/ |
|
366
|
3 |
|
private function findSupportedCharset($cs, &$mbSupported) |
|
367
|
|
|
{ |
|
368
|
3 |
|
$mbSupported = true; |
|
369
|
3 |
|
$comp = strtoupper($cs); |
|
370
|
3 |
|
$available = array_map('strtoupper', mb_list_encodings()); |
|
371
|
3 |
|
if (in_array($comp, $available)) { |
|
372
|
3 |
|
return $comp; |
|
373
|
|
|
} |
|
374
|
3 |
|
$stripped = preg_replace('/[^A-Z0-9]+/', '', $comp); |
|
375
|
3 |
|
$amb = preg_replace('/[^A-Z0-9]+/', '', $available); |
|
376
|
3 |
|
$index = array_search($stripped, $amb, true); |
|
377
|
3 |
|
if ($index !== false) { |
|
378
|
1 |
|
return $available[$index]; |
|
379
|
|
|
} |
|
380
|
3 |
|
return $this->findAliasedCharset($comp, $stripped, $mbSupported); |
|
381
|
|
|
} |
|
382
|
|
|
|
|
383
|
|
|
/** |
|
384
|
|
|
* Looks up the passed $comp and $stripped strings in self::$mbAliases, and |
|
385
|
|
|
* returns the mapped charset if applicable. Otherwise calls |
|
386
|
|
|
* $this->findAliasedIconvCharset. |
|
387
|
|
|
* |
|
388
|
|
|
* $mbSupported is set to false if the charset is not located in |
|
389
|
|
|
* self::$mbAliases. |
|
390
|
|
|
* |
|
391
|
|
|
* @param string $comp |
|
392
|
|
|
* @param string $stripped |
|
393
|
|
|
* @param boolean $mbSupported |
|
394
|
|
|
* @return string the mapped charset |
|
395
|
|
|
*/ |
|
396
|
3 |
|
private function findAliasedCharset($comp, $stripped, &$mbSupported) |
|
397
|
|
|
{ |
|
398
|
3 |
|
if (array_key_exists($comp, self::$mbAliases)) { |
|
399
|
1 |
|
return self::$mbAliases[$comp]; |
|
400
|
3 |
|
} elseif (array_key_exists($stripped, self::$mbAliases)) { |
|
401
|
2 |
|
return self::$mbAliases[$stripped]; |
|
402
|
|
|
} |
|
403
|
2 |
|
$mbSupported = false; |
|
404
|
2 |
|
return $this->findAliasedIconvCharset($comp, $stripped); |
|
405
|
|
|
} |
|
406
|
|
|
|
|
407
|
|
|
/** |
|
408
|
|
|
* Looks up the passed $comp and $stripped strings in self::$iconvAliases, |
|
409
|
|
|
* and returns the mapped charset if applicable. Otherwise returns $comp. |
|
410
|
|
|
* |
|
411
|
|
|
* @param string $comp |
|
412
|
|
|
* @param string $stripped |
|
413
|
|
|
* @return string the mapped charset (if mapped) or $comp otherwise |
|
414
|
|
|
*/ |
|
415
|
2 |
|
private function findAliasedIconvCharset($comp, $stripped) |
|
416
|
|
|
{ |
|
417
|
2 |
|
if (array_key_exists($comp, self::$iconvAliases)) { |
|
418
|
1 |
|
return static::$iconvAliases[$comp]; |
|
419
|
2 |
|
} elseif (array_key_exists($stripped, self::$iconvAliases)) { |
|
420
|
1 |
|
return static::$iconvAliases[$stripped]; |
|
421
|
|
|
} |
|
422
|
2 |
|
return $comp; |
|
423
|
|
|
} |
|
424
|
|
|
} |
|
425
|
|
|
|
zbateson /
mail-mime-parser
