| Matthias Andreas Benkard | b382b10 | 2021-01-02 15:32:21 +0100 | [diff] [blame^] | 1 | <?php namespace PhpMimeMailParser; |
| 2 | |
| 3 | use PhpMimeMailParser\Contracts\CharsetManager; |
| 4 | |
| 5 | class Charset implements CharsetManager |
| 6 | { |
| 7 | /** |
| 8 | * Charset Aliases |
| 9 | */ |
| 10 | private $charsetAlias = [ |
| 11 | 'ascii' => 'us-ascii', |
| 12 | 'us-ascii' => 'us-ascii', |
| 13 | 'ansi_x3.4-1968' => 'us-ascii', |
| 14 | '646' => 'us-ascii', |
| 15 | 'iso-8859-1' => 'iso-8859-1', |
| 16 | 'iso-8859-2' => 'iso-8859-2', |
| 17 | 'iso-8859-3' => 'iso-8859-3', |
| 18 | 'iso-8859-4' => 'iso-8859-4', |
| 19 | 'iso-8859-5' => 'iso-8859-5', |
| 20 | 'iso-8859-6' => 'iso-8859-6', |
| 21 | 'iso-8859-6-i' => 'iso-8859-6-i', |
| 22 | 'iso-8859-6-e' => 'iso-8859-6-e', |
| 23 | 'iso-8859-7' => 'iso-8859-7', |
| 24 | 'iso-8859-8' => 'iso-8859-8', |
| 25 | 'iso-8859-8-i' => 'iso-8859-8', |
| 26 | 'iso-8859-8-e' => 'iso-8859-8-e', |
| 27 | 'iso-8859-9' => 'iso-8859-9', |
| 28 | 'iso-8859-10' => 'iso-8859-10', |
| 29 | 'iso-8859-11' => 'iso-8859-11', |
| 30 | 'iso-8859-13' => 'iso-8859-13', |
| 31 | 'iso-8859-14' => 'iso-8859-14', |
| 32 | 'iso-8859-15' => 'iso-8859-15', |
| 33 | 'iso-8859-16' => 'iso-8859-16', |
| 34 | 'iso-ir-111' => 'iso-ir-111', |
| 35 | 'iso-2022-cn' => 'iso-2022-cn', |
| 36 | 'iso-2022-cn-ext' => 'iso-2022-cn', |
| 37 | 'iso-2022-kr' => 'iso-2022-kr', |
| 38 | 'iso-2022-jp' => 'iso-2022-jp', |
| 39 | 'utf-16be' => 'utf-16be', |
| 40 | 'utf-16le' => 'utf-16le', |
| 41 | 'utf-16' => 'utf-16', |
| 42 | 'windows-1250' => 'windows-1250', |
| 43 | 'windows-1251' => 'windows-1251', |
| 44 | 'windows-1252' => 'windows-1252', |
| 45 | 'windows-1253' => 'windows-1253', |
| 46 | 'windows-1254' => 'windows-1254', |
| 47 | 'windows-1255' => 'windows-1255', |
| 48 | 'windows-1256' => 'windows-1256', |
| 49 | 'windows-1257' => 'windows-1257', |
| 50 | 'windows-1258' => 'windows-1258', |
| 51 | 'ibm866' => 'ibm866', |
| 52 | 'ibm850' => 'ibm850', |
| 53 | 'ibm852' => 'ibm852', |
| 54 | 'ibm855' => 'ibm855', |
| 55 | 'ibm857' => 'ibm857', |
| 56 | 'ibm862' => 'ibm862', |
| 57 | 'ibm864' => 'ibm864', |
| 58 | 'utf-8' => 'utf-8', |
| 59 | 'utf-7' => 'utf-7', |
| 60 | 'shift_jis' => 'shift_jis', |
| 61 | 'big5' => 'big5', |
| 62 | 'euc-jp' => 'euc-jp', |
| 63 | 'euc-kr' => 'euc-kr', |
| 64 | 'gb2312' => 'gb2312', |
| 65 | 'gb18030' => 'gb18030', |
| 66 | 'viscii' => 'viscii', |
| 67 | 'koi8-r' => 'koi8-r', |
| 68 | 'koi8_r' => 'koi8-r', |
| 69 | 'cskoi8r' => 'koi8-r', |
| 70 | 'koi' => 'koi8-r', |
| 71 | 'koi8' => 'koi8-r', |
| 72 | 'koi8-u' => 'koi8-u', |
| 73 | 'tis-620' => 'tis-620', |
| 74 | 't.61-8bit' => 't.61-8bit', |
| 75 | 'hz-gb-2312' => 'hz-gb-2312', |
| 76 | 'big5-hkscs' => 'big5-hkscs', |
| 77 | 'gbk' => 'gbk', |
| 78 | 'cns11643' => 'x-euc-tw', |
| 79 | 'x-imap4-modified-utf7' => 'x-imap4-modified-utf7', |
| 80 | 'x-euc-tw' => 'x-euc-tw', |
| 81 | 'x-mac-ce' => 'macce', |
| 82 | 'x-mac-turkish' => 'macturkish', |
| 83 | 'x-mac-greek' => 'macgreek', |
| 84 | 'x-mac-icelandic' => 'macicelandic', |
| 85 | 'x-mac-croatian' => 'maccroatian', |
| 86 | 'x-mac-romanian' => 'macromanian', |
| 87 | 'x-mac-cyrillic' => 'maccyrillic', |
| 88 | 'x-mac-ukrainian' => 'macukrainian', |
| 89 | 'x-mac-hebrew' => 'machebrew', |
| 90 | 'x-mac-arabic' => 'macarabic', |
| 91 | 'x-mac-farsi' => 'macfarsi', |
| 92 | 'x-mac-devanagari' => 'macdevanagari', |
| 93 | 'x-mac-gujarati' => 'macgujarati', |
| 94 | 'x-mac-gurmukhi' => 'macgurmukhi', |
| 95 | 'armscii-8' => 'armscii-8', |
| 96 | 'x-viet-tcvn5712' => 'x-viet-tcvn5712', |
| 97 | 'x-viet-vps' => 'x-viet-vps', |
| 98 | 'iso-10646-ucs-2' => 'utf-16be', |
| 99 | 'x-iso-10646-ucs-2-be' => 'utf-16be', |
| 100 | 'x-iso-10646-ucs-2-le' => 'utf-16le', |
| 101 | 'x-user-defined' => 'x-user-defined', |
| 102 | 'x-johab' => 'x-johab', |
| 103 | 'latin1' => 'iso-8859-1', |
| 104 | 'iso_8859-1' => 'iso-8859-1', |
| 105 | 'iso8859-1' => 'iso-8859-1', |
| 106 | 'iso8859-2' => 'iso-8859-2', |
| 107 | 'iso8859-3' => 'iso-8859-3', |
| 108 | 'iso8859-4' => 'iso-8859-4', |
| 109 | 'iso8859-5' => 'iso-8859-5', |
| 110 | 'iso8859-6' => 'iso-8859-6', |
| 111 | 'iso8859-7' => 'iso-8859-7', |
| 112 | 'iso8859-8' => 'iso-8859-8', |
| 113 | 'iso8859-9' => 'iso-8859-9', |
| 114 | 'iso8859-10' => 'iso-8859-10', |
| 115 | 'iso8859-11' => 'iso-8859-11', |
| 116 | 'iso8859-13' => 'iso-8859-13', |
| 117 | 'iso8859-14' => 'iso-8859-14', |
| 118 | 'iso8859-15' => 'iso-8859-15', |
| 119 | 'iso_8859-1:1987' => 'iso-8859-1', |
| 120 | 'iso-ir-100' => 'iso-8859-1', |
| 121 | 'l1' => 'iso-8859-1', |
| 122 | 'ibm819' => 'iso-8859-1', |
| 123 | 'cp819' => 'iso-8859-1', |
| 124 | 'csisolatin1' => 'iso-8859-1', |
| 125 | 'latin2' => 'iso-8859-2', |
| 126 | 'iso_8859-2' => 'iso-8859-2', |
| 127 | 'iso_8859-2:1987' => 'iso-8859-2', |
| 128 | 'iso-ir-101' => 'iso-8859-2', |
| 129 | 'l2' => 'iso-8859-2', |
| 130 | 'csisolatin2' => 'iso-8859-2', |
| 131 | 'latin3' => 'iso-8859-3', |
| 132 | 'iso_8859-3' => 'iso-8859-3', |
| 133 | 'iso_8859-3:1988' => 'iso-8859-3', |
| 134 | 'iso-ir-109' => 'iso-8859-3', |
| 135 | 'l3' => 'iso-8859-3', |
| 136 | 'csisolatin3' => 'iso-8859-3', |
| 137 | 'latin4' => 'iso-8859-4', |
| 138 | 'iso_8859-4' => 'iso-8859-4', |
| 139 | 'iso_8859-4:1988' => 'iso-8859-4', |
| 140 | 'iso-ir-110' => 'iso-8859-4', |
| 141 | 'l4' => 'iso-8859-4', |
| 142 | 'csisolatin4' => 'iso-8859-4', |
| 143 | 'cyrillic' => 'iso-8859-5', |
| 144 | 'iso_8859-5' => 'iso-8859-5', |
| 145 | 'iso_8859-5:1988' => 'iso-8859-5', |
| 146 | 'iso-ir-144' => 'iso-8859-5', |
| 147 | 'csisolatincyrillic' => 'iso-8859-5', |
| 148 | 'arabic' => 'iso-8859-6', |
| 149 | 'iso_8859-6' => 'iso-8859-6', |
| 150 | 'iso_8859-6:1987' => 'iso-8859-6', |
| 151 | 'iso-ir-127' => 'iso-8859-6', |
| 152 | 'ecma-114' => 'iso-8859-6', |
| 153 | 'asmo-708' => 'iso-8859-6', |
| 154 | 'csisolatinarabic' => 'iso-8859-6', |
| 155 | 'csiso88596i' => 'iso-8859-6-i', |
| 156 | 'csiso88596e' => 'iso-8859-6-e', |
| 157 | 'greek' => 'iso-8859-7', |
| 158 | 'greek8' => 'iso-8859-7', |
| 159 | 'sun_eu_greek' => 'iso-8859-7', |
| 160 | 'iso_8859-7' => 'iso-8859-7', |
| 161 | 'iso_8859-7:1987' => 'iso-8859-7', |
| 162 | 'iso-ir-126' => 'iso-8859-7', |
| 163 | 'elot_928' => 'iso-8859-7', |
| 164 | 'ecma-118' => 'iso-8859-7', |
| 165 | 'csisolatingreek' => 'iso-8859-7', |
| 166 | 'hebrew' => 'iso-8859-8', |
| 167 | 'iso_8859-8' => 'iso-8859-8', |
| 168 | 'visual' => 'iso-8859-8', |
| 169 | 'iso_8859-8:1988' => 'iso-8859-8', |
| 170 | 'iso-ir-138' => 'iso-8859-8', |
| 171 | 'csisolatinhebrew' => 'iso-8859-8', |
| 172 | 'csiso88598i' => 'iso-8859-8', |
| 173 | 'iso-8859-8i' => 'iso-8859-8', |
| 174 | 'logical' => 'iso-8859-8', |
| 175 | 'csiso88598e' => 'iso-8859-8-e', |
| 176 | 'latin5' => 'iso-8859-9', |
| 177 | 'iso_8859-9' => 'iso-8859-9', |
| 178 | 'iso_8859-9:1989' => 'iso-8859-9', |
| 179 | 'iso-ir-148' => 'iso-8859-9', |
| 180 | 'l5' => 'iso-8859-9', |
| 181 | 'csisolatin5' => 'iso-8859-9', |
| 182 | 'unicode-1-1-utf-8' => 'utf-8', |
| 183 | 'utf8' => 'utf-8', |
| 184 | 'x-sjis' => 'shift_jis', |
| 185 | 'shift-jis' => 'shift_jis', |
| 186 | 'ms_kanji' => 'shift_jis', |
| 187 | 'csshiftjis' => 'shift_jis', |
| 188 | 'windows-31j' => 'shift_jis', |
| 189 | 'cp932' => 'shift_jis', |
| 190 | 'sjis' => 'shift_jis', |
| 191 | 'cseucpkdfmtjapanese' => 'euc-jp', |
| 192 | 'x-euc-jp' => 'euc-jp', |
| 193 | 'csiso2022jp' => 'iso-2022-jp', |
| 194 | 'iso-2022-jp-2' => 'iso-2022-jp', |
| 195 | 'csiso2022jp2' => 'iso-2022-jp', |
| 196 | 'csbig5' => 'big5', |
| 197 | 'cn-big5' => 'big5', |
| 198 | 'x-x-big5' => 'big5', |
| 199 | 'zh_tw-big5' => 'big5', |
| 200 | 'cseuckr' => 'euc-kr', |
| 201 | 'ks_c_5601-1987' => 'euc-kr', |
| 202 | 'iso-ir-149' => 'euc-kr', |
| 203 | 'ks_c_5601-1989' => 'euc-kr', |
| 204 | 'ksc_5601' => 'euc-kr', |
| 205 | 'ksc5601' => 'euc-kr', |
| 206 | 'korean' => 'euc-kr', |
| 207 | 'csksc56011987' => 'euc-kr', |
| 208 | '5601' => 'euc-kr', |
| 209 | 'windows-949' => 'euc-kr', |
| 210 | 'gb_2312-80' => 'gb2312', |
| 211 | 'iso-ir-58' => 'gb2312', |
| 212 | 'chinese' => 'gb2312', |
| 213 | 'csiso58gb231280' => 'gb2312', |
| 214 | 'csgb2312' => 'gb2312', |
| 215 | 'zh_cn.euc' => 'gb2312', |
| 216 | 'gb_2312' => 'gb2312', |
| 217 | 'x-cp1250' => 'windows-1250', |
| 218 | 'x-cp1251' => 'windows-1251', |
| 219 | 'x-cp1252' => 'windows-1252', |
| 220 | 'x-cp1253' => 'windows-1253', |
| 221 | 'x-cp1254' => 'windows-1254', |
| 222 | 'x-cp1255' => 'windows-1255', |
| 223 | 'x-cp1256' => 'windows-1256', |
| 224 | 'x-cp1257' => 'windows-1257', |
| 225 | 'x-cp1258' => 'windows-1258', |
| 226 | 'windows-874' => 'windows-874', |
| 227 | 'ibm874' => 'windows-874', |
| 228 | 'dos-874' => 'windows-874', |
| 229 | 'macintosh' => 'macintosh', |
| 230 | 'x-mac-roman' => 'macintosh', |
| 231 | 'mac' => 'macintosh', |
| 232 | 'csmacintosh' => 'macintosh', |
| 233 | 'cp866' => 'ibm866', |
| 234 | 'cp-866' => 'ibm866', |
| 235 | '866' => 'ibm866', |
| 236 | 'csibm866' => 'ibm866', |
| 237 | 'cp850' => 'ibm850', |
| 238 | '850' => 'ibm850', |
| 239 | 'csibm850' => 'ibm850', |
| 240 | 'cp852' => 'ibm852', |
| 241 | '852' => 'ibm852', |
| 242 | 'csibm852' => 'ibm852', |
| 243 | 'cp855' => 'ibm855', |
| 244 | '855' => 'ibm855', |
| 245 | 'csibm855' => 'ibm855', |
| 246 | 'cp857' => 'ibm857', |
| 247 | '857' => 'ibm857', |
| 248 | 'csibm857' => 'ibm857', |
| 249 | 'cp862' => 'ibm862', |
| 250 | '862' => 'ibm862', |
| 251 | 'csibm862' => 'ibm862', |
| 252 | 'cp864' => 'ibm864', |
| 253 | '864' => 'ibm864', |
| 254 | 'csibm864' => 'ibm864', |
| 255 | 'ibm-864' => 'ibm864', |
| 256 | 't.61' => 't.61-8bit', |
| 257 | 'iso-ir-103' => 't.61-8bit', |
| 258 | 'csiso103t618bit' => 't.61-8bit', |
| 259 | 'x-unicode-2-0-utf-7' => 'utf-7', |
| 260 | 'unicode-2-0-utf-7' => 'utf-7', |
| 261 | 'unicode-1-1-utf-7' => 'utf-7', |
| 262 | 'csunicode11utf7' => 'utf-7', |
| 263 | 'csunicode' => 'utf-16be', |
| 264 | 'csunicode11' => 'utf-16be', |
| 265 | 'iso-10646-ucs-basic' => 'utf-16be', |
| 266 | 'csunicodeascii' => 'utf-16be', |
| 267 | 'iso-10646-unicode-latin1' => 'utf-16be', |
| 268 | 'csunicodelatin1' => 'utf-16be', |
| 269 | 'iso-10646' => 'utf-16be', |
| 270 | 'iso-10646-j-1' => 'utf-16be', |
| 271 | 'latin6' => 'iso-8859-10', |
| 272 | 'iso-ir-157' => 'iso-8859-10', |
| 273 | 'l6' => 'iso-8859-10', |
| 274 | 'csisolatin6' => 'iso-8859-10', |
| 275 | 'iso_8859-15' => 'iso-8859-15', |
| 276 | 'csisolatin9' => 'iso-8859-15', |
| 277 | 'l9' => 'iso-8859-15', |
| 278 | 'ecma-cyrillic' => 'iso-ir-111', |
| 279 | 'csiso111ecmacyrillic' => 'iso-ir-111', |
| 280 | 'csiso2022kr' => 'iso-2022-kr', |
| 281 | 'csviscii' => 'viscii', |
| 282 | 'zh_tw-euc' => 'x-euc-tw', |
| 283 | 'iso88591' => 'iso-8859-1', |
| 284 | 'iso88592' => 'iso-8859-2', |
| 285 | 'iso88593' => 'iso-8859-3', |
| 286 | 'iso88594' => 'iso-8859-4', |
| 287 | 'iso88595' => 'iso-8859-5', |
| 288 | 'iso88596' => 'iso-8859-6', |
| 289 | 'iso88597' => 'iso-8859-7', |
| 290 | 'iso88598' => 'iso-8859-8', |
| 291 | 'iso88599' => 'iso-8859-9', |
| 292 | 'iso885910' => 'iso-8859-10', |
| 293 | 'iso885911' => 'iso-8859-11', |
| 294 | 'iso885912' => 'iso-8859-12', |
| 295 | 'iso885913' => 'iso-8859-13', |
| 296 | 'iso885914' => 'iso-8859-14', |
| 297 | 'iso885915' => 'iso-8859-15', |
| 298 | 'tis620' => 'tis-620', |
| 299 | 'cp1250' => 'windows-1250', |
| 300 | 'cp1251' => 'windows-1251', |
| 301 | 'cp1252' => 'windows-1252', |
| 302 | 'cp1253' => 'windows-1253', |
| 303 | 'cp1254' => 'windows-1254', |
| 304 | 'cp1255' => 'windows-1255', |
| 305 | 'cp1256' => 'windows-1256', |
| 306 | 'cp1257' => 'windows-1257', |
| 307 | 'cp1258' => 'windows-1258', |
| 308 | 'x-gbk' => 'gbk', |
| 309 | 'windows-936' => 'gbk', |
| 310 | 'ansi-1251' => 'windows-1251', |
| 311 | ]; |
| 312 | |
| 313 | /** |
| 314 | * {@inheritdoc} |
| 315 | */ |
| 316 | public function decodeCharset($encodedString, $charset) |
| 317 | { |
| 318 | $charset = $this->getCharsetAlias($charset); |
| 319 | |
| 320 | if ($charset == 'utf-8' || $charset == 'us-ascii') { |
| 321 | return $encodedString; |
| 322 | } |
| 323 | |
| 324 | if (function_exists('mb_convert_encoding')) { |
| 325 | if ($charset == 'iso-2022-jp') { |
| 326 | return mb_convert_encoding($encodedString, 'utf-8', 'iso-2022-jp-ms'); |
| 327 | } |
| 328 | |
| 329 | if (array_search($charset, $this->getSupportedEncodings())) { |
| 330 | return mb_convert_encoding($encodedString, 'utf-8', $charset); |
| 331 | } |
| 332 | } |
| 333 | |
| 334 | return iconv($charset, 'utf-8//translit//ignore', $encodedString); |
| 335 | } |
| 336 | |
| 337 | /** |
| 338 | * {@inheritdoc} |
| 339 | */ |
| 340 | public function getCharsetAlias($charset) |
| 341 | { |
| 342 | $charset = strtolower($charset); |
| 343 | |
| 344 | if (array_key_exists($charset, $this->charsetAlias)) { |
| 345 | return $this->charsetAlias[$charset]; |
| 346 | } |
| 347 | |
| 348 | return 'us-ascii'; |
| 349 | } |
| 350 | |
| 351 | private function getSupportedEncodings() |
| 352 | { |
| 353 | return |
| 354 | array_map( |
| 355 | 'strtolower', |
| 356 | array_unique( |
| 357 | array_merge( |
| 358 | $enc = mb_list_encodings(), |
| 359 | call_user_func_array( |
| 360 | 'array_merge', |
| 361 | array_map( |
| 362 | "mb_encoding_aliases", |
| 363 | $enc |
| 364 | ) |
| 365 | ) |
| 366 | ) |
| 367 | ) |
| 368 | ); |
| 369 | } |
| 370 | } |