src/cx/fbn/nevernote/utilities/StringUtils.java

   1 /*\r
   2  * This file is part of NeverNote \r
   3  * Copyright 2009 Randy Baumgarte\r
   4  * \r
   5  * This file may be licensed under the terms of of the\r
   6  * GNU General Public License Version 2 (the ``GPL'').\r
   7  *\r
   8  * Software distributed under the License is distributed\r
   9  * on an ``AS IS'' basis, WITHOUT WARRANTY OF ANY KIND, either\r
  10  * express or implied. See the GPL for the specific language\r
  11  * governing rights and limitations.\r
  12  *\r
  13  * You should have received a copy of the GPL along with this\r
  14  * program. If not, go to http://www.gnu.org/licenses/gpl.html\r
  15  * or write to the Free Software Foundation, Inc.,\r
  16  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.\r
  17  *\r
  18 */\r
  19 \r
  20 package cx.fbn.nevernote.utilities;\r
  21 \r
  22 import java.util.HashMap;\r
  23 \r
  24 public class StringUtils {\r
  25 \r
  26           private StringUtils() {}\r
  27           \r
  28           private static HashMap<String,String> htmlEntities;\r
  29           static {\r
  30             htmlEntities = new HashMap<String,String>();\r
  31             htmlEntities.put("&lt;","<")    ; htmlEntities.put("&gt;",">");\r
  32             htmlEntities.put("&amp;","&")   ; htmlEntities.put("&quot;","\"");\r
  33             htmlEntities.put("&agrave;","à"); htmlEntities.put("&agrave;","À");\r
  34             htmlEntities.put("&acirc;","â") ; htmlEntities.put("&auml;","ä");\r
  35             htmlEntities.put("&auml;","Ä")  ; htmlEntities.put("&acirc;","Â");\r
  36             htmlEntities.put("&aring;","å") ; htmlEntities.put("&aring;","Å");\r
  37             htmlEntities.put("&aelig;","æ") ; htmlEntities.put("&aElig;","Æ" );\r
  38             htmlEntities.put("&ccedil;","ç"); htmlEntities.put("&ccedil;","Ç");\r
  39             htmlEntities.put("&eacute;","é"); htmlEntities.put("&eacute;","É" );\r
  40             htmlEntities.put("&egrave;","è"); htmlEntities.put("&egrave;","È");\r
  41             htmlEntities.put("&ecirc;","ê") ; htmlEntities.put("&ecirc;","Ê");\r
  42             htmlEntities.put("&euml;","ë")  ; htmlEntities.put("&euml;","Ë");\r
  43             htmlEntities.put("&iuml;","ï")  ; htmlEntities.put("&iuml;","Ï");\r
  44             htmlEntities.put("&ocirc;","ô") ; htmlEntities.put("&ocirc;","Ô");\r
  45             htmlEntities.put("&ouml;","ö")  ; htmlEntities.put("&ouml;","Ö");\r
  46             htmlEntities.put("&oslash;","ø") ; htmlEntities.put("&oslash;","Ø");\r
  47             htmlEntities.put("&szlig;","ß") ; htmlEntities.put("&ugrave;","ù");\r
  48             htmlEntities.put("&ugrave;","Ù"); htmlEntities.put("&ucirc;","û");\r
  49             htmlEntities.put("&ucirc;","Û") ; htmlEntities.put("&uuml;","ü");\r
  50             htmlEntities.put("&uuml;","Ü")  ; htmlEntities.put("&nbsp;"," ");\r
  51             htmlEntities.put("&copy;","\u00a9"); htmlEntities.put("&apos;", "'");\r
  52             htmlEntities.put("&reg;","\u00ae"); htmlEntities.put("&iexcl;", "\u00a1");\r
  53             htmlEntities.put("&euro;","\u20a0"); htmlEntities.put("&cent;", "\u00a2");\r
  54             htmlEntities.put("&pound;", "\u00a3"); htmlEntities.put("&curen;", "\u00a4");\r
  55             htmlEntities.put("&yen;", "\u00a5"); htmlEntities.put("&brvbar;", "\u00a6");\r
  56             htmlEntities.put("&sect;", "\u00a7"); htmlEntities.put("&uml;", "\u00a8");\r
  57             htmlEntities.put("&copy;", "\u00a9"); htmlEntities.put("&ordf;", "\u00aa");\r
  58             htmlEntities.put("&laqo;", "\u00ab"); htmlEntities.put("&not;", "\u00ac");\r
  59             htmlEntities.put("&reg;", "\u00ae"); htmlEntities.put("&macr;", "\u00af");\r
  60           }\r
  61 \r
  62 \r
  63           \r
  64           public static final String unescapeHTML(String source, int start){\r
  65                      int i,j;\r
  66 \r
  67                      i = source.indexOf("&", start);\r
  68                      while (i>-1) {\r
  69                         j = source.indexOf(";" ,i);\r
  70                         if (j > i) {\r
  71                            String entityToLookFor = source.substring(i , j + 1);\r
  72                            String value = htmlEntities.get(entityToLookFor);\r
  73                            if (value != null) {\r
  74                                    value = " ";\r
  75                                    source = new StringBuffer().append(source.substring(0 , i).toLowerCase())\r
  76                                            .append(value)\r
  77                                            .append(source.substring(j + 1))\r
  78                                            .toString();\r
  79                                    i = source.indexOf("&", i+1);\r
  80                            }\r
  81                         }\r
  82                      }\r
  83                      return source;\r
  84                   }\r
  85 \r
  86           \r
  87           public static final String unescapeHTML2(String source, int start){\r
  88              int i,j;\r
  89 \r
  90              i = source.indexOf("&", start);\r
  91              if (i > -1) {\r
  92                 j = source.indexOf(";" ,i);\r
  93                 if (j > i) {\r
  94                    String entityToLookFor = source.substring(i , j + 1);\r
  95                    String value = htmlEntities.get(entityToLookFor);\r
  96                    if (value != null) {\r
  97                            value = " ";\r
  98                            source = new StringBuffer().append(source.substring(0 , i).toLowerCase())\r
  99                                    .append(value)\r
 100                                    .append(source.substring(j + 1))\r
 101                                    .toString();\r
 102                 return unescapeHTML(source, i + 1); // recursive call\r
 103                    }\r
 104                 }\r
 105              }\r
 106              return source;\r
 107           }\r
 108 }\r