reporting/includes/html_entity_decode_php4.php

   1 <?php
   2 //============================================================+
   3 // File name   : html_entity_decode_php4.php
   4 // Begin       : 2006-06-22
   5 // Last Update : 2008-04-01
   6 // Author      : Nicola Asuni
   7 // Version     : 1.0.000
   8 // License     : GNU LGPL (http://www.gnu.org/copyleft/lesser.html)
   9 //      ----------------------------------------------------------------------------
  10 //  Copyright (C) 2002-2008  Nicola Asuni - Tecnick.com S.r.l.
  11 //
  12 //      This program is free software: you can redistribute it and/or modify
  13 //      it under the terms of the GNU Lesser General Public License as published by
  14 //      the Free Software Foundation, either version 2.1 of the License, or
  15 //      (at your option) any later version.
  16 //
  17 //      This program is distributed in the hope that it will be useful,
  18 //      but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 //      MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20 //      GNU Lesser General Public License for more details.
  21 //
  22 //      You should have received a copy of the GNU Lesser General Public License
  23 //      along with this program.  If not, see <http://www.gnu.org/licenses/>.
  24 //
  25 //      See LICENSE.TXT file for more information.
  26 //  ----------------------------------------------------------------------------
  27 //
  28 //
  29 // Description : This is a PHP4 function that redefine the
  30 //               standard html_entity_decode function to support
  31 //               UTF-8 encoding.
  32 //
  33 //
  34 // Author: Nicola Asuni
  35 //
  36 // (c) Copyright:
  37 //               Nicola Asuni
  38 //               Tecnick.com S.r.l.
  39 //               Via della Pace, 11
  40 //               09044 Quartucciu (CA)
  41 //               ITALY
  42 //               www.tecnick.com
  43 //               info@tecnick.com
  44 //============================================================+
  45
  46 /**
  47  * This is a PHP4 function that redefine the standard
  48  * html_entity_decode function to support UTF-8 encoding.
  49  * @package com.tecnick.tcpdf
  50  * @author Nicola Asuni
  51  * @copyright 2004-2008 Nicola Asuni - Tecnick.com S.r.l (www.tecnick.com) Via Della Pace, 11 - 09044 - Quartucciu (CA) - ITALY - www.tecnick.com - info@tecnick.com
  52  * @link http://www.tcpdf.org
  53  * @license http://www.gnu.org/copyleft/lesser.html LGPL
  54  */
  55
  56 /**
  57  * Returns the UTF-8 string corresponding to unicode value.
  58  * @param $num unicode value to convert.
  59  * @return string converted
  60  */
  61 function code_to_utf8($num) {
  62         if ($num <= 0x7F) {
  63                 return chr($num);
  64         } elseif ($num <= 0x7FF) {
  65                 return chr(($num >> 0x06) + 0xC0).chr(($num & 0x3F) + 128);
  66         } elseif ($num <= 0xFFFF) {
  67                 return chr(($num >> 0x0C) + 0xE0).chr((($num >> 0x06) & 0x3F) + 0x80).chr(($num & 0x3F) + 0x80);
  68         } elseif ($num <= 0x1FFFFF) {
  69                 return chr(($num >> 0x12) + 0xF0).chr((($num >> 0x0C) & 0x3F) + 0x80).chr((($num >> 0x06) & 0x3F) + 0x80).chr(($num & 0x3F) + 0x80);
  70         }
  71         return ' '; // default value
  72 }
  73
  74 /**
  75  * Reverse function for htmlentities.
  76  * Convert entities in UTF-8.
  77  * @param $text_to_convert Text to convert.
  78  * @return string converted
  79  */
  80 function html_entity_decode_php4($text_to_convert) {
  81         $htmlentities_table = array (
  82                 "&Aacute;" => "".chr(195).chr(129)."",
  83                 "&aacute;" => "".chr(195).chr(161)."",
  84                 "&Acirc;" => "".chr(195).chr(130)."",
  85                 "&acirc;" => "".chr(195).chr(162)."",
  86                 "&acute;" => "".chr(194).chr(180)."",
  87                 "&AElig;" => "".chr(195).chr(134)."",
  88                 "&aelig;" => "".chr(195).chr(166)."",
  89                 "&Agrave;" => "".chr(195).chr(128)."",
  90                 "&agrave;" => "".chr(195).chr(160)."",
  91                 "&alefsym;" => "".chr(226).chr(132).chr(181)."",
  92                 "&Alpha;" => "".chr(206).chr(145)."",
  93                 "&alpha;" => "".chr(206).chr(177)."",
  94                 "&amp;" => "".chr(38)."",
  95                 "&and;" => "".chr(226).chr(136).chr(167)."",
  96                 "&ang;" => "".chr(226).chr(136).chr(160)."",
  97                 "&Aring;" => "".chr(195).chr(133)."",
  98                 "&aring;" => "".chr(195).chr(165)."",
  99                 "&asymp;" => "".chr(226).chr(137).chr(136)."",
 100                 "&Atilde;" => "".chr(195).chr(131)."",
 101                 "&atilde;" => "".chr(195).chr(163)."",
 102                 "&Auml;" => "".chr(195).chr(132)."",
 103                 "&auml;" => "".chr(195).chr(164)."",
 104                 "&bdquo;" => "".chr(226).chr(128).chr(158)."",
 105                 "&Beta;" => "".chr(206).chr(146)."",
 106                 "&beta;" => "".chr(206).chr(178)."",
 107                 "&brvbar;" => "".chr(194).chr(166)."",
 108                 "&bull;" => "".chr(226).chr(128).chr(162)."",
 109                 "&cap;" => "".chr(226).chr(136).chr(169)."",
 110                 "&Ccedil;" => "".chr(195).chr(135)."",
 111                 "&ccedil;" => "".chr(195).chr(167)."",
 112                 "&cedil;" => "".chr(194).chr(184)."",
 113                 "&cent;" => "".chr(194).chr(162)."",
 114                 "&Chi;" => "".chr(206).chr(167)."",
 115                 "&chi;" => "".chr(207).chr(135)."",
 116                 "&circ;" => "".chr(203).chr(134)."",
 117                 "&clubs;" => "".chr(226).chr(153).chr(163)."",
 118                 "&cong;" => "".chr(226).chr(137).chr(133)."",
 119                 "&copy;" => "".chr(194).chr(169)."",
 120                 "&crarr;" => "".chr(226).chr(134).chr(181)."",
 121                 "&cup;" => "".chr(226).chr(136).chr(170)."",
 122                 "&curren;" => "".chr(194).chr(164)."",
 123                 "&dagger;" => "".chr(226).chr(128).chr(160)."",
 124                 "&Dagger;" => "".chr(226).chr(128).chr(161)."",
 125                 "&darr;" => "".chr(226).chr(134).chr(147)."",
 126                 "&dArr;" => "".chr(226).chr(135).chr(147)."",
 127                 "&deg;" => "".chr(194).chr(176)."",
 128                 "&Delta;" => "".chr(206).chr(148)."",
 129                 "&delta;" => "".chr(206).chr(180)."",
 130                 "&diams;" => "".chr(226).chr(153).chr(166)."",
 131                 "&divide;" => "".chr(195).chr(183)."",
 132                 "&Eacute;" => "".chr(195).chr(137)."",
 133                 "&eacute;" => "".chr(195).chr(169)."",
 134                 "&Ecirc;" => "".chr(195).chr(138)."",
 135                 "&ecirc;" => "".chr(195).chr(170)."",
 136                 "&Egrave;" => "".chr(195).chr(136)."",
 137                 "&egrave;" => "".chr(195).chr(168)."",
 138                 "&empty;" => "".chr(226).chr(136).chr(133)."",
 139                 "&emsp;" => "".chr(226).chr(128).chr(131)."",
 140                 "&ensp;" => "".chr(226).chr(128).chr(130)."",
 141                 "&Epsilon;" => "".chr(206).chr(149)."",
 142                 "&epsilon;" => "".chr(206).chr(181)."",
 143                 "&equiv;" => "".chr(226).chr(137).chr(161)."",
 144                 "&Eta;" => "".chr(206).chr(151)."",
 145                 "&eta;" => "".chr(206).chr(183)."",
 146                 "&ETH;" => "".chr(195).chr(144)."",
 147                 "&eth;" => "".chr(195).chr(176)."",
 148                 "&Euml;" => "".chr(195).chr(139)."",
 149                 "&euml;" => "".chr(195).chr(171)."",
 150                 "&euro;" => "".chr(226).chr(130).chr(172)."",
 151                 "&exist;" => "".chr(226).chr(136).chr(131)."",
 152                 "&fnof;" => "".chr(198).chr(146)."",
 153                 "&forall;" => "".chr(226).chr(136).chr(128)."",
 154                 "&frac12;" => "".chr(194).chr(189)."",
 155                 "&frac14;" => "".chr(194).chr(188)."",
 156                 "&frac34;" => "".chr(194).chr(190)."",
 157                 "&frasl;" => "".chr(226).chr(129).chr(132)."",
 158                 "&Gamma;" => "".chr(206).chr(147)."",
 159                 "&gamma;" => "".chr(206).chr(179)."",
 160                 "&ge;" => "".chr(226).chr(137).chr(165)."",
 161                 "&harr;" => "".chr(226).chr(134).chr(148)."",
 162                 "&hArr;" => "".chr(226).chr(135).chr(148)."",
 163                 "&hearts;" => "".chr(226).chr(153).chr(165)."",
 164                 "&hellip;" => "".chr(226).chr(128).chr(166)."",
 165                 "&Iacute;" => "".chr(195).chr(141)."",
 166                 "&iacute;" => "".chr(195).chr(173)."",
 167                 "&Icirc;" => "".chr(195).chr(142)."",
 168                 "&icirc;" => "".chr(195).chr(174)."",
 169                 "&iexcl;" => "".chr(194).chr(161)."",
 170                 "&Igrave;" => "".chr(195).chr(140)."",
 171                 "&igrave;" => "".chr(195).chr(172)."",
 172                 "&image;" => "".chr(226).chr(132).chr(145)."",
 173                 "&infin;" => "".chr(226).chr(136).chr(158)."",
 174                 "&int;" => "".chr(226).chr(136).chr(171)."",
 175                 "&Iota;" => "".chr(206).chr(153)."",
 176                 "&iota;" => "".chr(206).chr(185)."",
 177                 "&iquest;" => "".chr(194).chr(191)."",
 178                 "&isin;" => "".chr(226).chr(136).chr(136)."",
 179                 "&Iuml;" => "".chr(195).chr(143)."",
 180                 "&iuml;" => "".chr(195).chr(175)."",
 181                 "&Kappa;" => "".chr(206).chr(154)."",
 182                 "&kappa;" => "".chr(206).chr(186)."",
 183                 "&Lambda;" => "".chr(206).chr(155)."",
 184                 "&lambda;" => "".chr(206).chr(187)."",
 185                 "&lang;" => "".chr(226).chr(140).chr(169)."",
 186                 "&laquo;" => "".chr(194).chr(171)."",
 187                 "&larr;" => "".chr(226).chr(134).chr(144)."",
 188                 "&lArr;" => "".chr(226).chr(135).chr(144)."",
 189                 "&lceil;" => "".chr(226).chr(140).chr(136)."",
 190                 "&ldquo;" => "".chr(226).chr(128).chr(156)."",
 191                 "&le;" => "".chr(226).chr(137).chr(164)."",
 192                 "&lfloor;" => "".chr(226).chr(140).chr(138)."",
 193                 "&lowast;" => "".chr(226).chr(136).chr(151)."",
 194                 "&loz;" => "".chr(226).chr(151).chr(138)."",
 195                 "&lrm;" => "".chr(226).chr(128).chr(142)."",
 196                 "&lsaquo;" => "".chr(226).chr(128).chr(185)."",
 197                 "&lsquo;" => "".chr(226).chr(128).chr(152)."",
 198                 "&macr;" => "".chr(194).chr(175)."",
 199                 "&mdash;" => "".chr(226).chr(128).chr(148)."",
 200                 "&micro;" => "".chr(194).chr(181)."",
 201                 "&middot;" => "".chr(194).chr(183)."",
 202                 "&minus;" => "".chr(226).chr(136).chr(146)."",
 203                 "&Mu;" => "".chr(206).chr(156)."",
 204                 "&mu;" => "".chr(206).chr(188)."",
 205                 "&nabla;" => "".chr(226).chr(136).chr(135)."",
 206                 "&nbsp;" => "".chr(194).chr(160)."",
 207                 "&ndash;" => "".chr(226).chr(128).chr(147)."",
 208                 "&ne;" => "".chr(226).chr(137).chr(160)."",
 209                 "&ni;" => "".chr(226).chr(136).chr(139)."",
 210                 "&not;" => "".chr(194).chr(172)."",
 211                 "&notin;" => "".chr(226).chr(136).chr(137)."",
 212                 "&nsub;" => "".chr(226).chr(138).chr(132)."",
 213                 "&Ntilde;" => "".chr(195).chr(145)."",
 214                 "&ntilde;" => "".chr(195).chr(177)."",
 215                 "&Nu;" => "".chr(206).chr(157)."",
 216                 "&nu;" => "".chr(206).chr(189)."",
 217                 "&Oacute;" => "".chr(195).chr(147)."",
 218                 "&oacute;" => "".chr(195).chr(179)."",
 219                 "&Ocirc;" => "".chr(195).chr(148)."",
 220                 "&ocirc;" => "".chr(195).chr(180)."",
 221                 "&OElig;" => "".chr(197).chr(146)."",
 222                 "&oelig;" => "".chr(197).chr(147)."",
 223                 "&Ograve;" => "".chr(195).chr(146)."",
 224                 "&ograve;" => "".chr(195).chr(178)."",
 225                 "&oline;" => "".chr(226).chr(128).chr(190)."",
 226                 "&Omega;" => "".chr(206).chr(169)."",
 227                 "&omega;" => "".chr(207).chr(137)."",
 228                 "&Omicron;" => "".chr(206).chr(159)."",
 229                 "&omicron;" => "".chr(206).chr(191)."",
 230                 "&oplus;" => "".chr(226).chr(138).chr(149)."",
 231                 "&or;" => "".chr(226).chr(136).chr(168)."",
 232                 "&ordf;" => "".chr(194).chr(170)."",
 233                 "&ordm;" => "".chr(194).chr(186)."",
 234                 "&Oslash;" => "".chr(195).chr(152)."",
 235                 "&oslash;" => "".chr(195).chr(184)."",
 236                 "&Otilde;" => "".chr(195).chr(149)."",
 237                 "&otilde;" => "".chr(195).chr(181)."",
 238                 "&otimes;" => "".chr(226).chr(138).chr(151)."",
 239                 "&Ouml;" => "".chr(195).chr(150)."",
 240                 "&ouml;" => "".chr(195).chr(182)."",
 241                 "&para;" => "".chr(194).chr(182)."",
 242                 "&part;" => "".chr(226).chr(136).chr(130)."",
 243                 "&permil;" => "".chr(226).chr(128).chr(176)."",
 244                 "&perp;" => "".chr(226).chr(138).chr(165)."",
 245                 "&Phi;" => "".chr(206).chr(166)."",
 246                 "&phi;" => "".chr(207).chr(134)."",
 247                 "&Pi;" => "".chr(206).chr(160)."",
 248                 "&pi;" => "".chr(207).chr(128)."",
 249                 "&piv;" => "".chr(207).chr(150)."",
 250                 "&plusmn;" => "".chr(194).chr(177)."",
 251                 "&pound;" => "".chr(194).chr(163)."",
 252                 "&prime;" => "".chr(226).chr(128).chr(178)."",
 253                 "&Prime;" => "".chr(226).chr(128).chr(179)."",
 254                 "&prod;" => "".chr(226).chr(136).chr(143)."",
 255                 "&prop;" => "".chr(226).chr(136).chr(157)."",
 256                 "&Psi;" => "".chr(206).chr(168)."",
 257                 "&psi;" => "".chr(207).chr(136)."",
 258                 "&radic;" => "".chr(226).chr(136).chr(154)."",
 259                 "&rang;" => "".chr(226).chr(140).chr(170)."",
 260                 "&raquo;" => "".chr(194).chr(187)."",
 261                 "&rarr;" => "".chr(226).chr(134).chr(146)."",
 262                 "&rArr;" => "".chr(226).chr(135).chr(146)."",
 263                 "&rceil;" => "".chr(226).chr(140).chr(137)."",
 264                 "&rdquo;" => "".chr(226).chr(128).chr(157)."",
 265                 "&real;" => "".chr(226).chr(132).chr(156)."",
 266                 "&reg;" => "".chr(194).chr(174)."",
 267                 "&rfloor;" => "".chr(226).chr(140).chr(139)."",
 268                 "&Rho;" => "".chr(206).chr(161)."",
 269                 "&rho;" => "".chr(207).chr(129)."",
 270                 "&rlm;" => "".chr(226).chr(128).chr(143)."",
 271                 "&rsaquo;" => "".chr(226).chr(128).chr(186)."",
 272                 "&rsquo;" => "".chr(226).chr(128).chr(153)."",
 273                 "&sbquo;" => "".chr(226).chr(128).chr(154)."",
 274                 "&Scaron;" => "".chr(197).chr(160)."",
 275                 "&scaron;" => "".chr(197).chr(161)."",
 276                 "&sdot;" => "".chr(226).chr(139).chr(133)."",
 277                 "&sect;" => "".chr(194).chr(167)."",
 278                 "&shy;" => "".chr(194).chr(173)."",
 279                 "&Sigma;" => "".chr(206).chr(163)."",
 280                 "&sigma;" => "".chr(207).chr(131)."",
 281                 "&sigmaf;" => "".chr(207).chr(130)."",
 282                 "&sim;" => "".chr(226).chr(136).chr(188)."",
 283                 "&spades;" => "".chr(226).chr(153).chr(160)."",
 284                 "&sub;" => "".chr(226).chr(138).chr(130)."",
 285                 "&sube;" => "".chr(226).chr(138).chr(134)."",
 286                 "&sum;" => "".chr(226).chr(136).chr(145)."",
 287                 "&sup1;" => "".chr(194).chr(185)."",
 288                 "&sup2;" => "".chr(194).chr(178)."",
 289                 "&sup3;" => "".chr(194).chr(179)."",
 290                 "&sup;" => "".chr(226).chr(138).chr(131)."",
 291                 "&supe;" => "".chr(226).chr(138).chr(135)."",
 292                 "&szlig;" => "".chr(195).chr(159)."",
 293                 "&Tau;" => "".chr(206).chr(164)."",
 294                 "&tau;" => "".chr(207).chr(132)."",
 295                 "&there4;" => "".chr(226).chr(136).chr(180)."",
 296                 "&Theta;" => "".chr(206).chr(152)."",
 297                 "&theta;" => "".chr(206).chr(184)."",
 298                 "&thetasym;" => "".chr(207).chr(145)."",
 299                 "&thinsp;" => "".chr(226).chr(128).chr(137)."",
 300                 "&THORN;" => "".chr(195).chr(158)."",
 301                 "&thorn;" => "".chr(195).chr(190)."",
 302                 "&tilde;" => "".chr(203).chr(156)."",
 303                 "&times;" => "".chr(195).chr(151)."",
 304                 "&trade;" => "".chr(226).chr(132).chr(162)."",
 305                 "&Uacute;" => "".chr(195).chr(154)."",
 306                 "&uacute;" => "".chr(195).chr(186)."",
 307                 "&uarr;" => "".chr(226).chr(134).chr(145)."",
 308                 "&uArr;" => "".chr(226).chr(135).chr(145)."",
 309                 "&Ucirc;" => "".chr(195).chr(155)."",
 310                 "&ucirc;" => "".chr(195).chr(187)."",
 311                 "&Ugrave;" => "".chr(195).chr(153)."",
 312                 "&ugrave;" => "".chr(195).chr(185)."",
 313                 "&uml;" => "".chr(194).chr(168)."",
 314                 "&upsih;" => "".chr(207).chr(146)."",
 315                 "&Upsilon;" => "".chr(206).chr(165)."",
 316                 "&upsilon;" => "".chr(207).chr(133)."",
 317                 "&Uuml;" => "".chr(195).chr(156)."",
 318                 "&uuml;" => "".chr(195).chr(188)."",
 319                 "&weierp;" => "".chr(226).chr(132).chr(152)."",
 320                 "&Xi;" => "".chr(206).chr(158)."",
 321                 "&xi;" => "".chr(206).chr(190)."",
 322                 "&Yacute;" => "".chr(195).chr(157)."",
 323                 "&yacute;" => "".chr(195).chr(189)."",
 324                 "&yen;" => "".chr(194).chr(165)."",
 325                 "&yuml;" => "".chr(195).chr(191)."",
 326                 "&Yuml;" => "".chr(197).chr(184)."",
 327                 "&Zeta;" => "".chr(206).chr(150)."",
 328                 "&zeta;" => "".chr(206).chr(182)."",
 329                 "&zwj;" => "".chr(226).chr(128).chr(141)."",
 330                 "&zwnj;" => "".chr(226).chr(128).chr(140)."",
 331                 "&gt;" => ">",
 332                 "&lt;" => "<"
 333         );
 334         $return_text = strtr($text_to_convert, $htmlentities_table);
 335
 336         if (version_compare(PHP_VERSION, '5.3.0') >= 0) // 07.11.2014, from php 5.3.0 fixed deprecated preg_replace with the /e flag. Joe.
 337         {
 338                 $return_text = preg_replace_callback('~&#x([0-9a-f]+);~i', function ($m){ return chr(hexdec($m[1]));}, $return_text);
 339                 $return_text = preg_replace_callback('~&#([0-9]+);~', function ($m){ return chr($m[1]);}, $return_text);
 340         }
 341         else
 342         {
 343                 $return_text = preg_replace('~&#x([0-9a-f]+);~ei', 'code_to_utf8(hexdec("\\1"))', $return_text);
 344                 $return_text = preg_replace('~&#([0-9]+);~e', 'code_to_utf8(\\1)', $return_text);
 345         }
 346         return $return_text;
 347 }
 348
 349 //============================================================+
 350 // END OF FILE
 351 //============================================================+
 352 ?>