source/mir/misc/StringUtil.java

   1 /*
   2  * put your module comment here
   3  */
   4
   5
   6 package  mir.misc;
   7
   8 import  java.io.*;
   9 import  java.lang.*;
  10 import  java.util.*;
  11 import  gnu.regexp.*;
  12
  13 /**
  14  * Statische Hilfsmethoden zur Stringbehandlung
  15  *
  16  * @version 29.6.99
  17  * @author RK
  18  */
  19 public final class StringUtil {
  20
  21         private static RE   re_newline2br, re_brbr2p, re_mail, re_url, re_tags;
  22
  23         private StringUtil() { }  // this avoids contruction
  24
  25         static {
  26                 try {
  27                         //precompile regex
  28                         re_newline2br = new RE("(\r?\n){1}");
  29                         re_brbr2p     = new RE("(<br>\r?\n<br>){1,}");
  30                         re_mail       = new RE("([a-zA-Z0-9_.-]+)@([a-zA-Z0-9_-]+).([a-zA-Z0-9_.-]+)");
  31                         re_url        = new RE("((https://)|(http://)|(ftp://)){1}([a-zA-Z0-9_-]+).([a-zA-Z0-9_.:-]+)/?([^ \t\r\n<>\\)\\]]+[^ \t\r\n.,<>\\)\\]])");
  32                         re_tags       = new RE("<[^>]*>",RE.REG_ICASE);
  33                 }
  34                 catch (REException e){
  35                         System.err.println("FATAL: StringUtil: could not precompile REGEX: "+e.toString());
  36                 }
  37         }
  38
  39
  40         /**
  41          * Wandelt Datum in einen 8-ziffrigen String um (yyyymmdd)
  42          * @param theDate
  43          * @return 8-ziffriger String (yyyymmdd)
  44          */
  45
  46         public static final String date2webdbDate (GregorianCalendar theDate) {
  47                 StringBuffer webdbDate = new StringBuffer();
  48                 webdbDate.append(String.valueOf(theDate.get(Calendar.YEAR)));
  49                 webdbDate.append(pad2(theDate.get(Calendar.MONTH) + 1));
  50                 webdbDate.append(pad2(theDate.get(Calendar.DATE)));
  51                 return  webdbDate.toString();
  52         }
  53
  54         /**
  55          * Wandelt Calendar in einen 12-ziffrigen String um (yyyymmddhhmm)
  56          * @param theDate
  57          * @return 12-ziffriger String (yyyymmdd)
  58          */
  59
  60         public static final String date2webdbDateTime (GregorianCalendar theDate) {
  61                 StringBuffer webdbDate = new StringBuffer();
  62                 webdbDate.append(String.valueOf(theDate.get(Calendar.YEAR)));
  63                 webdbDate.append(pad2(theDate.get(Calendar.MONTH) + 1));
  64                 webdbDate.append(pad2(theDate.get(Calendar.DATE)));
  65                 webdbDate.append(pad2(theDate.get(Calendar.HOUR)));
  66                 webdbDate.append(pad2(theDate.get(Calendar.MINUTE)));
  67                 return  webdbDate.toString();
  68         }
  69
  70         /**
  71          * wandelt Calendar in dd.mm.yyyy / hh.mm um
  72          * @param theDate
  73          * @return String mit (dd.mm.yyyy / hh.mm um)
  74          */
  75         public static String date2readableDateTime (GregorianCalendar theDate) {
  76                 String readable = "";
  77                 int hour;
  78                 readable += pad2(theDate.get(Calendar.DATE));
  79                 readable += "." + pad2(theDate.get(Calendar.MONTH) + 1);
  80                 readable += "." + String.valueOf(theDate.get(Calendar.YEAR));
  81                 hour = theDate.get(Calendar.HOUR);
  82                 if (theDate.get(Calendar.AM_PM) == Calendar.PM)
  83                         hour += 12;
  84                 readable += " / " + pad2(hour);
  85                 readable += ":" + pad2(theDate.get(Calendar.MINUTE));
  86                 return  readable;
  87         }
  88
  89         /**
  90          * wandelt eine Datum in einen 8-buchstabigen String, der durch <code>/</code>
  91          * getrennt ist.
  92          *
  93          * @param webdbDate
  94          * @return String mit <code>/yyyy/mm/dd</code>
  95          */
  96         public static final String webdbDate2path (String webdbDate) {
  97                 StringBuffer path = new StringBuffer();
  98                 path.append("/").append(webdbDate.substring(0, 4));
  99                 path.append("/").append(webdbDate.substring(4, 6));
 100                 path.append("/");
 101                 //who did this?
 102                 //path.append("/").append(webdbDate.substring(6, 8));
 103                 return  path.toString();
 104         }
 105
 106         /**
 107          * wandelt Calendar in dd.mm.yyyy um
 108          *
 109          * @param theDate
 110          * @return String mit  <code>dd.mm.yyyy</code>
 111          */
 112         public static final String webdbDate2readableDate (String webdbDate) {
 113                 String date = "";
 114                 date += webdbDate.substring(6, 8);
 115                 date += "." + webdbDate.substring(4, 6);
 116                 date += "." + webdbDate.substring(0, 4);
 117                 return  date;
 118         }
 119
 120
 121         /**
 122          * converts string from format: yyyy-mm-dd__hh:mm:ss.d
 123          * to dd.mm.yyyy hh:mm
 124          */
 125         public static String dateToReadableDate(String date) {
 126                 StringBuffer returnDate = new StringBuffer();
 127                 if (date!=null) {
 128
 129                         returnDate.append(date.substring(8,10)).append('.');
 130                         returnDate.append(date.substring(5,7)).append('.');
 131                         returnDate.append(date.substring(0,4)).append(' ');
 132                         returnDate.append(date.substring(11,16));
 133                 }
 134                 return returnDate.toString();
 135         }
 136
 137         /**
 138          * converts string from format: yyyy-mm-dd__hh:mm:ss.d
 139          * to yyyy
 140          */
 141         public static String dateToYear (String date) {
 142                 StringBuffer returnDate = new StringBuffer();
 143                 if (date!=null) {
 144
 145                         returnDate.append(date.substring(0,4));
 146                 }
 147                 return returnDate.toString();
 148         }
 149
 150         /**
 151          * converts string from format: yyyy-mm-dd__hh:mm:ss.d
 152          * to [m]m
 153          */
 154         public static String dateToMonth (String date) {
 155                 StringBuffer returnDate = new StringBuffer();
 156                 if (date!=null) {
 157                         if (!date.substring(5,6).equalsIgnoreCase("0")) returnDate.append(date.substring(5,7));
 158                         else returnDate.append(date.substring(6,7));
 159                 }
 160                 return returnDate.toString();
 161         }
 162
 163         /**
 164          * converts string from format: yyyy-mm-dd__hh:mm:ss.d
 165          * to [d]d
 166          */
 167         public static String dateToDayOfMonth (String date) {
 168                 StringBuffer returnDate = new StringBuffer();
 169                 if (date!=null) {
 170                         if (!date.substring(8,9).equalsIgnoreCase("0")) returnDate.append(date.substring(8,10));
 171                         else returnDate.append(date.substring(9,10));
 172                 }
 173                 return returnDate.toString();
 174         }
 175
 176         /**
 177          * converts string from format: yyyy-mm-dd__hh:mm:ss.d
 178          * to hh:mm
 179          */
 180         public static String dateToTime (String date) {
 181                 StringBuffer returnDate = new StringBuffer();
 182                 if (date!=null) {
 183                         returnDate.append(date.substring(11,16));
 184                 }
 185                 return returnDate.toString();
 186         }
 187
 188
 189         /**
 190          * schließt einen String in Anführungsszeichen ein, falls er Leerzeichen o.ä. enthält
 191          *
 192          * @return gequoteter String
 193          */
 194          public static String quoteIfNecessary(String s) {
 195                 for (int i = 0; i < s.length(); i++)
 196                         if (!(Character.isLetterOrDigit(s.charAt(i)) || s.charAt(i) == '.'))
 197                                 return quote(s, '"');
 198                 return s;
 199         }
 200
 201          /**
 202          * schließt <code>s</code> in <code>'</code> ein und setzt Backslashes vor
 203          * "gefährliche" Zeichen innerhalb des Strings
 204          * Quotes special SQL-characters in <code>s</code>
 205          *
 206          * @return geqoteter String
 207          */
 208         public static String quote(String s)
 209         {
 210                 //String s2 = quote(s, '\'');
 211
 212                 //Quickhack
 213                 //Because of '?-Bug in Postgresql-JDBC-Driver
 214                 StringBuffer temp = new StringBuffer();
 215                 for(int i=0;i<s.length();i++){
 216                         if(s.charAt(i)=='\''){
 217                                 temp.append("&acute;");
 218                         } else {
 219                                 temp.append(s.charAt(i));
 220                         }
 221                 }
 222                 String s2 = temp.toString();
 223                 //Quickhack end
 224
 225                 s2 = quote(s2, '\"');
 226                 return s2;
 227         }
 228
 229         /**
 230          * schließt <code>s</code> in <code>'</code> ein und setzt Backslashes vor
 231          * "gefährliche" Zeichen innerhalb des Strings
 232          *
 233          * @param s String, der gequoted werden soll
 234          * @param quoteChar zu quotendes Zeichen
 235          * @return gequoteter String
 236          */
 237         public static String quote(String s, char quoteChar)
 238         {
 239                 StringBuffer buf = new StringBuffer(s.length());
 240                 int pos = 0;
 241                 while (pos < s.length()) {
 242                         int i = s.indexOf(quoteChar, pos);
 243                         if (i < 0) i = s.length();
 244                         buf.append(s.substring(pos, i));
 245                         pos = i;
 246                         if (pos < s.length()) {
 247                                 buf.append('\\');
 248                                 buf.append(quoteChar);
 249                                 pos++;
 250                         }
 251                 }
 252                 return buf.toString();
 253         }
 254
 255         /**
 256          * replaces dangerous characters in <code>s</code>
 257          *
 258          */
 259
 260         public static String unquote(String s)
 261         {
 262                 char quoteChar='\'';
 263                 StringBuffer buf = new StringBuffer(s.length());
 264                 int pos = 0;
 265                 String searchString = "\\"+quoteChar;
 266                 while (pos < s.length()) {
 267                         int i = s.indexOf(searchString, pos);
 268                         if (i < 0) i = s.length();
 269                         buf.append(s.substring(pos, i));
 270                         pos = i+1;
 271                 }
 272                 return buf.toString();
 273         }
 274
 275         /**
 276          * Wandelet String in byte[] um.
 277          * @param s
 278          * @return byte[] des String
 279          */
 280
 281         public static byte[] stringToBytes(String s) {
 282                 String crlf = System.getProperty("line.separator");
 283                 if (!crlf.equals("\n"))
 284                         s = replace(s, "\n", crlf);
 285                 // byte[] buf = new byte[s.length()];
 286                 byte[] buf = s.getBytes();
 287                 return buf;
 288         }
 289
 290                 /**
 291          * Ersetzt in String <code>s</code> das <code>pattern</code> durch <code>substitute</code>
 292          * @param s
 293          * @param pattern
 294          * @param substitute
 295          * @return String mit den Ersetzungen
 296          */
 297         public static String replace(String s, String pattern, String substitute) {
 298                 int i = 0, pLen = pattern.length(), sLen = substitute.length();
 299                 StringBuffer buf = new StringBuffer(s.length());
 300                 while (true) {
 301                         int j = s.indexOf(pattern, i);
 302                         if (j < 0) {
 303                                 buf.append(s.substring(i));
 304                                 break;
 305                         } else {
 306                                 buf.append(s.substring(i, j));
 307                                 buf.append(substitute);
 308                                 i = j+pLen;
 309                         }
 310                 }
 311                 return buf.toString();
 312         }
 313
 314         /**
 315          * Ersetzt in String <code>s</code> das Regexp <code>pattern</code> durch <code>substitute</code>
 316          * @param s
 317          * @param pattern
 318          * @param substitute
 319          * @return String mit den Ersetzungen
 320          */
 321         public static String regexpReplace(String haystack, String pattern, String substitute) {
 322                 try {
 323                         RE regex = new RE(pattern);
 324                         return regex.substituteAll(haystack,substitute);
 325                 } catch(REException ex){
 326                         return null;
 327                 }
 328         }
 329
 330
 331
 332
 333         /**
 334          * Fügt einen Separator an den Pfad an
 335          * @param path
 336          * @return Pfad mit Separator am Ende
 337          */
 338         public static final String addSeparator (String path) {
 339                 return  path.length() == 0 || path.endsWith(File.separator) ? path : path
 340                                 + File.separatorChar;
 341         }
 342
 343         /**
 344          * Fügt ein <code>/</code> ans ende des Strings and
 345          * @param path
 346          * @return Pfad mit <code>/</code> am Ende
 347          */
 348         public static final String addSlash (String path) {
 349                 return  path.length() == 0 || path.endsWith("/") ? path : path + '/';
 350         }
 351
 352         /**
 353          * Löscht <code>/</code> am Ende des Strings, falls vorhanden
 354          * @param path
 355          * @return String ohne <code>/</code> am Ende
 356          */
 357         public static final String removeSlash (String path) {
 358                 return  path.length() > 1 && path.endsWith("/") ? path.substring(0, path.length()
 359                                 - 1) : path;
 360         }
 361
 362         /**
 363          * Checks to see if the path is absolute by looking for a leading file
 364          * separater
 365          * @param path
 366          * @return
 367          */
 368         public static boolean isAbsolutePath (String path) {
 369                 return  path.startsWith(File.separator);
 370         }
 371
 372         /**
 373          * Löscht Slash am Anfang des Strings
 374          * @param path
 375          * @return
 376          */
 377         public static String removeFirstSlash (String path) {
 378                 return  path.startsWith("/") ? path.substring(1) : path;
 379         }
 380
 381         /**
 382          * formatiert eine Zahl (0-99) zweistellig (z.B. 5 -> 05)
 383          * @return zwistellige Zahl
 384          */
 385         public static String pad2 (int number) {
 386                 return  number < 10 ? "0" + number : String.valueOf(number);
 387         }
 388
 389         /**
 390          * formatiert eine Zahl (0-999) dreistellig (z.B. 7 -> 007)
 391          *
 392          * @return 3-stellige Zahl
 393          */
 394         public static String pad3 (int number) {
 395                 return  number < 10 ? "00" + number : number < 100 ? "0" + number : String.valueOf(number);
 396         }
 397
 398         /**
 399          * Konvertiert Unix-Linefeeds in Win-Linefeeds
 400          * @param s
 401          * @return Konvertierter String
 402          */
 403         public static String unixLineFeedsToWin(String s) {
 404                 int i = -1;
 405                 while (true) {
 406                         i = s.indexOf('\n', i+1);
 407                         if (i < 0) break;
 408                         if ((i == 0 || s.charAt(i-1) != '\r') &&
 409                                 (i == s.length()-1 || s.charAt(i+1) != '\r')) {
 410                                 s = s.substring(0, i)+'\r'+s.substring(i);
 411                                 i++;
 412                         }
 413                 }
 414                 return s;
 415         }
 416
 417
 418         /**
 419          * verwandelt einen String in eine gültige Url, konvertiert Sonderzeichen
 420          * und Spaces werden zu Underscores
 421          *
 422          * @return gültige Url
 423          */
 424         public static String convert2url(String s) {
 425                 s = toLowerCase(s);
 426                 StringBuffer buf = new StringBuffer();
 427                 for(int i = 0; i < s.length(); i++ ) {
 428                                 switch( s.charAt( i ) ) {
 429                                 case 'ö':
 430                         buf.append( "oe" ); break;
 431                                 case 'ä':
 432                         buf.append( "ae" ); break;
 433                                 case 'ü':
 434                         buf.append( "ue" ); break;
 435                                 case 'ã':
 436                         buf.append( "a" ); break;
 437                                 case '´':
 438                                 case '.':
 439                         buf.append( "_" ); break;
 440                                 case ' ':
 441                         if( buf.charAt( buf.length() - 1 ) != '_' ) {
 442                                         buf.append( "_" );
 443                         }
 444                         break;
 445                                 default:
 446                         buf.append( s.charAt( i ) );
 447                                 }
 448                 }
 449                 return buf.toString();
 450         }
 451
 452         /**
 453          * wandelt Sonderzeichen in Quotes um
 454          *
 455          * @return Kovertierter String
 456          */
 457         public static String encodeHtml(String s) {
 458                 StringBuffer buf = new StringBuffer();
 459                 for(int i=0;i < s.length(); i++ ) {
 460
 461                         /** @todo looks inefficient, to ask for index of every char, in
 462                          *  case of failure it runs to the end.*/
 463                         if (s.charAt(i)=='&') {
 464                                 // convert html to xml-parsable representation
 465                                 if( s.indexOf( "&ouml;", i ) == i ) {
 466                                         buf.append( "&#246;" ); i += 5;
 467                                         continue;
 468                                 }
 469                                 if( s.indexOf( "&auml;", i ) == i ) {
 470                                         buf.append( "&#228;" ); i += 5;
 471                                         continue;
 472                                 }
 473                                 if( s.indexOf( "&uuml;", i ) == i ) {
 474                                         buf.append( "&#252;" ); i += 5;
 475                                         continue;
 476                                 }
 477                                 if( s.indexOf( "&Ouml;", i ) == i ) {
 478                                         buf.append( "&#214;" ); i += 5;
 479                                         continue;
 480                                 }
 481                                 if( s.indexOf( "&Auml;", i ) == i ) {
 482                                         buf.append( "&#196;" ); i += 5;
 483                                         continue;
 484                                 }
 485                                 if( s.indexOf( "&Uuml;", i ) == i ) {
 486                                         buf.append( "&#220;" ); i += 5;
 487                                         continue;
 488                                 }
 489                                 if( s.indexOf( "&szlig;", i ) == i ) {
 490                                         buf.append( "&#223;" ); i += 6;
 491                                         continue;
 492                                 }
 493
 494                                 /** @todo should only escape outside of tags */
 495
 496                                 if( s.indexOf( "&quot;", i ) == i ) {
 497                                         buf.append( "&#223;" ); i += 5;
 498                                         continue;
 499                                 }
 500                                 if( s.indexOf( "&ndash;", i ) == i ) {
 501                                         buf.append( "&#8211;" ); i += 6;
 502                                         continue;
 503                                 }
 504                                 if( s.indexOf( "&mdash;", i ) == i ) {
 505                                         buf.append( "&#8212;" ); i += 6;
 506                                         continue;
 507                                 }
 508                                 if( s.indexOf( "&ldquo;", i ) == i ) {
 509                                         buf.append( "&#8220;" ); i += 6;
 510                                         continue;
 511                                 }
 512                                 if( s.indexOf( "&rdquo;", i ) == i ) {
 513                                         buf.append( "&#8221;" ); i += 6;
 514                                         continue;
 515                                 }
 516                                 if( s.indexOf( "&bdquo;", i ) == i ) {
 517                                         buf.append( "&#8222;" ); i += 6;
 518                                         continue;
 519                                 }
 520
 521                                 //looks pretty stupid
 522                                 if( s.indexOf( "&lt;", i ) == i ) {
 523                                         buf.append( "&lt;" ); i += 3;
 524                                         continue;
 525                                 }
 526                                 if( s.indexOf( "&gt;", i ) == i ) {
 527                                         buf.append( "&gt;" ); i += 3;
 528                                         continue;
 529                                 }
 530                                 if( s.indexOf( "&acute;", i ) == i ) {
 531                                         buf.append( "&acute;" ); i += 6;
 532                                         continue;
 533                                 }
 534                                 if( s.indexOf( "&nbsp;", i ) == i ) {
 535                                         buf.append( "&nbsp;" ); i += 5;
 536                                         continue;
 537                                 }
 538                                 //has to be the last
 539                                 //if( s.indexOf( "&", i ) == i ) {
 540                                 //  buf.append( "&#38;" ); i += 0;
 541                                 //  continue;
 542                                 //}
 543                         }
 544                         // convert umlauts an other special charakters
 545                         switch( s.charAt(i) ) {
 546                                 case 'ö': buf.append( "&#246;" ); break;
 547                                 case 'ä': buf.append( "&#228;" ); break;
 548                                 case 'ü': buf.append( "&#252;" ); break;
 549                                 case 'Ö': buf.append( "&#214;" ); break;
 550                                 case 'Ä': buf.append( "&#196;" ); break;
 551                                 case 'Ü': buf.append( "&#220;" ); break;
 552                                 case 'ß': buf.append( "&#223;" ); break;
 553                                 case 'é': buf.append( "&#233;" ); break;
 554                                 case 'è': buf.append( "&#232;" ); break;
 555                                 case 'á': buf.append( "&#225;" ); break;
 556                                 case 'à': buf.append( "&#224;" ); break;
 557                                 case 'â': buf.append( "&#226;" ); break;
 558                                 case 'ã': buf.append( "&#227;" ); break;
 559                                 case '¬': buf.append( "&#172;" ); break;
 560                                 case '¹': buf.append( "&#185;" ); break;
 561                                 case '²': buf.append( "&#178;" ); break;
 562                                 case '³': buf.append( "&#179;" ); break;
 563                                 case '¼': buf.append( "&#188;" ); break;
 564                                 case '½': buf.append( "&#189;" ); break;
 565                                 case '¾': buf.append( "&#190;" ); break;
 566                                 case '¶': buf.append( "&#182;" ); break;
 567                                 case 'æ': buf.append( "&#230;" ); break;
 568                                 case 'ð': buf.append( "&#240;" ); break;
 569                                 case '|': buf.append( "&#166;" ); break;
 570                                 case '·': buf.append( "&#183;" ); break;
 571                                 case '°': buf.append( "&#176;" ); break;
 572                                 case '§': buf.append( "&#167;" ); break;
 573                                 case 'ø': buf.append( "&#248;" ); break;
 574                                 case 'ç': buf.append( "&#231;" ); break;
 575                                 case '¤': buf.append( "&#164;" ); break;
 576                                 case 'ª': buf.append( "&#170;" ); break;
 577                                 case 'Ç': buf.append( "&#199;" ); break;
 578                                 case 'Ã': buf.append( "&#195;" ); break;
 579                                 case 'Â': buf.append( "&#194;" ); break;
 580                                 case 'Æ': buf.append( "&#198;" ); break;
 581                                 case '©': buf.append( "&#169;" ); break;
 582                                 case '®': buf.append( "&#174;" ); break;
 583                                 case '¥': buf.append( "&#165;" ); break;
 584                                 case 'Þ': buf.append( "&#254;" ); break;
 585                                 case '¯': buf.append( "&#175;" ); break;
 586                                 case 'Ð': buf.append( "&#208;" ); break;
 587                                 case 'º': buf.append( "&#186;" ); break;
 588                                 case '¡': buf.append( "&#161;" ); break;
 589                                 case '£': buf.append( "&#163;" ); break;
 590                                 case '±': buf.append( "&#177;" ); break;
 591                                 case '¿': buf.append( "&#191;" ); break;
 592                                 case 'Ø': buf.append( "&#216;" ); break;
 593                                 case 'Á': buf.append( "&#192;" ); break;
 594                                 case 'À': buf.append( "&#193;" ); break;
 595                                 case 'É': buf.append( "&#200;" ); break;
 596                                 case 'È': buf.append( "&#201;" ); break;
 597                                 case 'ù': buf.append( "&#250;" ); break;
 598                                 case 'ñ': buf.append( "&#241;" ); break;
 599                                 case 'Ñ': buf.append( "&#209;" ); break;
 600                                 case 'µ': buf.append( "&#181;" ); break;
 601                                 case 'Í': buf.append( "&#204;" ); break;
 602                                 case 'Ì': buf.append( "&#205;" ); break;
 603                                 case 'í': buf.append( "&#236;" ); break;
 604                                 case 'ì': buf.append( "&#237;" ); break;
 605                                 case 'î': buf.append( "&#238;" ); break;
 606                                 case 'Î': buf.append( "&#206;" ); break;
 607                                 case 'ó': buf.append( "&#243;" ); break;
 608                                 case 'Ó': buf.append( "&#210;" ); break;
 609                                 case 'ò': buf.append( "&#206;" ); break;
 610                                 case 'Ò': buf.append( "&#211;" ); break;
 611                                 case 'ô': buf.append( "&#244;" ); break;
 612                                 case 'Ô': buf.append( "&#212;" ); break;
 613                                 case 'õ': buf.append( "&#245;" ); break;
 614                                 case 'Õ': buf.append( "&#213;" ); break;
 615                                 case 'ý': buf.append( "&#253;" ); break;
 616                                 case 'Ý': buf.append( "&#221;" ); break;
 617                                 case 'û': buf.append( "&#251;" ); break;
 618                                 case 'Û': buf.append( "&#219;" ); break;
 619                                 case 'ú': buf.append( "&#249;" ); break;
 620                                 case 'Ú': buf.append( "&#217;" ); break;
 621                                 case 'Ù': buf.append( "&#218;" ); break;
 622                                 case 'Ê': buf.append( "&#202;" ); break;
 623                                 case 'ê': buf.append( "&#234;" ); break;
 624                                 case 'å': buf.append( "&#229;" ); break;
 625                                 case 'Å': buf.append( "&#197;" ); break;
 626                                 case 'ë': buf.append( "&#235;" ); break;
 627                                 case 'Ë': buf.append( "&#203;" ); break;
 628                                 case 'ÿ': buf.append( "&#255;" ); break;
 629                                 case 'ï': buf.append( "&#239;" ); break;
 630                                 case 'Ï': buf.append( "&#207;" ); break;
 631                                 case '«': buf.append( "&#171;" ); break;
 632                                 case '»': buf.append( "&#187;" ); break;
 633                                 case '\'': buf.append( "&acute;" ); break;
 634                                 case '\"': buf.append( "&quot;" ); break;
 635                                 //case '\u8211': buf.append( "&#8211;" ); break;
 636                                 //case '\u8212': buf.append( "&#8212;" ); break;
 637                                 //case '\u8220': buf.append( "&#8220;" ); break;
 638                                 //case '\u8221': buf.append( "&#8221;" ); break;
 639                                 //case '\u8222': buf.append( "&#8222;" ); break;
 640                                 //case '\"': buf.append( "&#34;" ); break;
 641                                 default: buf.append( s.charAt(i) );
 642                         }
 643
 644                 }
 645                 return buf.toString();
 646         }
 647
 648
 649         public static String decodeHTMLinTags(String s){
 650                 StringBuffer buffer = new StringBuffer();
 651                 boolean start = false;
 652                 boolean stop = false;
 653                 int startIndex = 0;
 654                 int stopIndex = 0;
 655                 int temp = 0;
 656
 657                 for(int i=0;i<s.length();i++){
 658                         if(s.charAt(i)=='<'){
 659                                 start = true;
 660                                 startIndex = i;
 661                         } else if(s.charAt(i)=='>'){
 662                                 stop = true;
 663                                 stopIndex = i;
 664
 665                                 if(start && stop){
 666                                         buffer.append(s.substring(temp,startIndex));
 667                                         buffer.append(replaceQuot(s.substring(startIndex,stopIndex+1)));
 668                                         i= temp= stopIndex+1;
 669                                         start= stop= false;
 670                                 }
 671                         }
 672                 }
 673                 if(stopIndex>0){
 674                         buffer.append(s.substring(stopIndex+1));
 675                         return buffer.toString();
 676                 } else {
 677                         return s;
 678                 }
 679         }
 680
 681         public static String replaceQuot(String s) {
 682                 StringBuffer buffer = new StringBuffer();
 683                 for(int j = 0; j < s.length();j++){
 684                         if(s.charAt(j)=='&'){
 685                                 if(s.indexOf( "&quot;",j) == j) {
 686                                         buffer.append( "\"" );
 687                                         j += 5;
 688                                 }//if
 689                         } else {
 690                                 buffer.append(s.charAt(j));
 691                         }//else
 692                 }//for
 693                 return buffer.toString();
 694         }
 695
 696         /** wandelt Quotes in Sonderzeichen um
 697          */
 698         /**
 699         public static String decodeHtml(String s) {
 700                 StringBuffer buf = new StringBuffer();
 701                 for(int i=0;i < s.length(); i++ ) {
 702                         if( s.indexOf( "&ouml;", i ) == i ) {
 703                                 buf.append( "ö" ); i += 5;
 704                                 continue;
 705                         }
 706                         if( s.indexOf( "&auml;", i ) == i ) {
 707                                 buf.append( "ä" ); i += 5;
 708                                 continue;
 709                         }
 710                         if( s.indexOf( "&uuml;", i ) == i ) {
 711                                 buf.append( "ü" ); i += 5;
 712                                 continue;
 713                         }
 714                         if( s.indexOf( "&Ouml;", i ) == i ) {
 715                                 buf.append( "Ö" ); i += 5;
 716                                 continue;
 717                         }
 718                         if( s.indexOf( "&Auml;", i ) == i ) {
 719                                 buf.append( "Ä" ); i += 5;
 720                                 continue;
 721                         }
 722                         if( s.indexOf( "&Uuml;", i ) == i ) {
 723                                 buf.append( "Ü" ); i += 5;
 724                                 continue;
 725                         }
 726                         if( s.indexOf( "&szlig;", i ) == i ) {
 727                                 buf.append( "ß" ); i += 6;
 728                                 continue;
 729                         }
 730                         if( s.indexOf( "&quot;", i ) == i ) {
 731                                 buf.append( "\"" ); i += 5;
 732                                 continue;
 733                         }
 734                         buf.append( s.charAt(i) );
 735                 }
 736                 return buf.toString();
 737         }
 738          */
 739
 740         /**
 741          * schnellere Variante der String.toLowerCase()-Routine
 742          *
 743          * @return String in Kleinbuchsten
 744          */
 745         public static String toLowerCase(String s) {
 746                 int l = s.length();
 747                 char[] a = new char[l];
 748                 for (int i = 0; i < l; i++)
 749                         a[i] = Character.toLowerCase(s.charAt(i));
 750                 return new String(a);
 751         }
 752
 753                 /**
 754          * Findet <code>element</code> im String-Array <code>array</code>
 755          * @param array
 756          * @param element
 757          * @return Fundstelle als int oder -1
 758          */
 759         public static int indexOf(String[] array, String element) {
 760                 if (array != null)
 761                         for (int i = 0; i < array.length; i++)
 762                                 if (array[i].equals(element))
 763                                         return i;
 764                 return -1;
 765         }
 766
 767         /**
 768          * Testet auf Vorkommen von <code>element</code> in <code>array</code>
 769          * @param array String-Array
 770          * @param element
 771          * @return true wenn <code>element</code> vorkommt, sonst false
 772          */
 773         public static boolean contains(String[] array, String element) {
 774                 return indexOf(array, element) >= 0;
 775         }
 776
 777                 /**
 778          * Ermittelt CRC-Prüfsumme von String <code>s</code>
 779          * @param s
 780          * @return CRC-Prüfsumme
 781          */
 782         public static int getCRC(String s) {
 783                 int h = 0;
 784                 char val[] = s.toCharArray();
 785                 int len = val.length;
 786
 787                 for (int i = 0 ; i < len; i++) {
 788                         h &= 0x7fffffff;
 789                         h = (((h >> 30) | (h << 1)) ^ (val[i]+i));
 790                 }
 791
 792                 return (h << 8) | (len & 0xff);
 793         }
 794
 795                 /**
 796          * Liefert Default-Wert def zurück, wenn String <code>s</code>
 797          * kein Integer ist.
 798          *
 799          * @param s
 800          * @param def
 801          * @return geparster int aus s oder def
 802          */
 803         public static int parseInt(String s, int def) {
 804                 if (s == null) return def;
 805                 try {
 806                         return Integer.parseInt(s);
 807                 } catch (NumberFormatException e) {
 808                         return def;
 809                 }
 810         }
 811
 812         /**
 813          * Liefert Defaultwert def zurück, wenn s nicht zu einem float geparsed werden kann.
 814          * @param s
 815          * @param def
 816          * @return geparster float oder def
 817          */
 818         public static float parseFloat(String s, float def) {
 819                 if (s == null) return def;
 820                 try {
 821                         return new Float(s).floatValue();
 822                 } catch (NumberFormatException e) {
 823                         return def;
 824                 }
 825         }
 826
 827                 /**
 828          * Findet Ende eines Satzes in String <code>text</code>
 829          * @param text
 830          * @param startIndex
 831          * @return index des Satzendes, oder -1
 832          */
 833         public static int findEndOfSentence(String text, int startIndex) {
 834                  while (true) {
 835                          int i = text.indexOf('.', startIndex);
 836                          if (i < 0) return -1;
 837                          if (i > 0 && !Character.isDigit(text.charAt(i-1)) &&
 838                                         (i+1 >= text.length()
 839                                         || text.charAt(i+1) == ' '
 840                                         || text.charAt(i+1) == '\n'
 841                                         || text.charAt(i+1) == '\t'))
 842                                         return i+1;
 843                          startIndex = i+1;
 844                  }
 845         }
 846
 847                 /**
 848          * Findet Wortende in String <code>text</code> ab <code>startIndex</code>
 849          * @param text
 850          * @param startIndex
 851          * @return Index des Wortendes, oder -1
 852          */
 853         public static int findEndOfWord(String text, int startIndex) {
 854                 int i = text.indexOf(' ', startIndex),
 855                         j = text.indexOf('\n', startIndex);
 856                 if (i < 0) i = text.length();
 857                 if (j < 0) j = text.length();
 858                 return Math.min(i, j);
 859         }
 860
 861
 862         /**
 863          *  convertNewline2P ist eine regex-routine zum umwandeln von 2 oder mehr newlines (\n)
 864          *  in den html-tag <p>
 865          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 866          */
 867         public static String convertNewline2P(String haystack) {
 868                         return re_brbr2p.substituteAll(haystack,"\n</p><p>");
 869         }
 870
 871         /**
 872          *  convertNewline2Break ist eine regex-routine zum umwandeln von 1 newline (\n)
 873          *  in den html-tag <br>
 874          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 875          */
 876         public static String convertNewline2Break(String haystack) {
 877                 return re_newline2br.substituteAll(haystack,"$0<br>");
 878         }
 879
 880         /**
 881          *  createMailLinks wandelt text im email-adressenformat
 882          *  in einen klickbaren link um
 883          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 884          */
 885         public static String createMailLinks(String haystack) {
 886                         return re_mail.substituteAll(haystack,"<a href=\"mailto:$0\">$0</a>");
 887         }
 888
 889
 890         /**
 891          *  createMailLinks wandelt text im email-adressenformat
 892          *  in einen klickbaren link um
 893          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 894          */
 895         public static String createMailLinks(String haystack, String imageRoot, String mailImage) {
 896                 return re_mail.substituteAll(haystack,"<img src=\""+imageRoot+"/"+mailImage+"\" border=\"0\"/>&#160;<a href=\"mailto:$0\">$0</a>");
 897         }
 898
 899
 900         /**
 901          *  createURLLinks wandelt text im url-format
 902          *  in einen klickbaren link um
 903          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 904          */
 905         public static String createURLLinks(String haystack) {
 906                 return re_url.substituteAll(haystack,"<a href=\"$0\">$0</a>");
 907         }
 908
 909         /**
 910          * this routine takes text in url format and makes
 911          * a clickaeble "<href>" link removing any "illegal" html tags
 912          * @param haystack, the url
 913          * @param title, the href link text
 914          * @param imagRoot, the place to find icons
 915          * @param extImage, the url of the icon to show next to the link
 916          * @return a String containing the url
 917          */
 918         public static String createURLLinks(String haystack, String title, String imageRoot,String extImage) {
 919                 if (title == null) {
 920                         return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/>&#160;<a href=\"$0\">$0</a>");
 921                 } else {
 922                         title = removeHTMLTags(title);
 923                         return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/>&#160;<a href=\"$0\">"+title+"</a>");
 924                 }
 925         }
 926
 927         /**
 928          * this routine takes text in url format and makes
 929          * a clickaeble "<href>" link removing any "illegal" html tags
 930          * @param haystack, the url
 931          * @param imageRoot, the place to find icons
 932          * @param extImage, the url of the icon to show next to the link
 933          * @param intImage, unused
 934          * @return a String containing the url
 935          */
 936         public static String createURLLinks(String haystack, String title, String imageRoot,String extImage,String intImage) {
 937                 return createURLLinks(haystack, title, imageRoot, extImage);
 938         }
 939
 940         /**
 941          * this routine takes text in url format and makes
 942          * an image link removing any "illegal" html tags
 943          * @param haystack, the url
 944          * @param title, the image alt text, can be null
 945          * @param height, height of the image
 946          * @param width, width of the image
 947          * @return a String containing the url
 948          */
 949         public static String createIMGLinks(String haystack, String title, String height,String width) {
 950                 if (title != null) {
 951                         title = removeHTMLTags(title);
 952                         return re_url.substituteAll(haystack,"<img hspace=\"10\" vspace=\"6\" align=\"left\" src=\"$0\" width=\""+width+"\" height=\""+height+"\" alt=\""+title+"\"/>&#160;<br><i>"+title+"</i>");
 953                 } else {
 954                         return re_url.substituteAll(haystack,"<img hspace=\"10\" vspace=\"6\" align=\"left\" src=\"$0\" width=\""+width+"\" height=\""+height+"\" alt=\"\"/>&#160;");
 955                 }
 956         }
 957
 958
 959          /**
 960          *  deleteForbiddenTags
 961          *  this method deletes all <script>, <body> and <head>-tags
 962          */
 963         public static final String deleteForbiddenTags(String haystack) {
 964                 try {
 965                         RE regex = new RE("<[ \t\r\n](.*?)script(.*?)/script(.*?)>",RE.REG_ICASE);
 966                         haystack = regex.substituteAll(haystack,"");
 967                         regex = new RE("<head>(.*?)</head>");
 968                         haystack = regex.substituteAll(haystack,"");
 969                         regex = new RE("<[ \t\r\n/]*body(.*?)>");
 970                         haystack = regex.substituteAll(haystack,"");
 971                         return haystack;
 972                 } catch(REException ex){
 973                         return null;
 974                 }
 975         }
 976
 977         /**
 978          * this method deletes all html tags
 979          */
 980         public static final String removeHTMLTags(String haystack){
 981                         return re_tags.substituteAll(haystack,"");
 982         }
 983
 984
 985         /**
 986          * this method deletes all but the approved tags html tags
 987          * it also deletes approved tags which contain malicious-looking attributes and doesn't work at all
 988          */
 989         public static String approveHTMLTags(String haystack){
 990                 try {
 991                         String approvedTags="a|img|h1|h2|h3|h4|h5|h6|br|b|i|strong|p";
 992                         String badAttributes="onAbort|onBlur|onChange|onClick|onDblClick|onDragDrop|onError|onFocus|onKeyDown|onKeyPress|onKeyUp|onLoad|onMouseDown|onMouseMove|onMouseOut|onMouseOver|onMouseUp|onMove|onReset|onResize|onSelect|onSubmit|onUnload";
 993                         String approvedProtocols="rtsp|http|ftp|https|freenet|mailto";
 994
 995                         // kill all the bad tags that have attributes
 996                         String s = "<\\s*/?\\s*(?!(("+approvedTags+")\\s))\\w+\\s[^>]*>";
 997                         RE regex = new RE(s,RE.REG_ICASE);
 998                         haystack = regex.substituteAll(haystack,"");
 999
1000                         // kill all the bad tags that are attributeless
1001                         regex = new RE("<\\s*/?\\s*(?!(("+approvedTags+")\\s*>))\\w+\\s*>",RE.REG_ICASE);
1002                         haystack = regex.substituteAll(haystack,"");
1003
1004                         // kill all the tags which have a javascript attribute like onLoad
1005                         regex = new RE("<[^>]*("+badAttributes+")[^>]*>",RE.REG_ICASE);
1006                         haystack = regex.substituteAll(haystack,"");
1007
1008                         // kill all the tags which include a url to an unacceptable protocol
1009                         regex = new RE("<\\s*a\\s+[^>]*href=(?!(\'|\")?("+approvedProtocols+"))[^>]*>",RE.REG_ICASE);
1010                         haystack = regex.substituteAll(haystack,"");
1011
1012                         return haystack;
1013                 } catch(REException ex){
1014                         ex.printStackTrace();
1015                         return null;
1016                 }
1017         }
1018
1019
1020         /**
1021          *  createHTML ruft alle regex-methoden zum unwandeln eines nicht
1022          *  htmlcodierten string auf und returnt einen htmlcodierten String
1023          */
1024         public static String createHTML(String content){
1025                 content=convertNewline2Break(content);
1026                 content=convertNewline2P(content);
1027                 content=createMailLinks(content);
1028                 content=createURLLinks(content);
1029                 return content;
1030         }
1031
1032
1033         /**
1034          *  createHTML ruft alle regex-methoden zum unwandeln eines nicht
1035          *  htmlcodierten string auf und returnt einen htmlcodierten String
1036          */
1037         public static String createHTML(String content,String producerDocRoot,String mailImage,String extImage,String intImage){
1038                 content=convertNewline2Break(content);
1039                 content=convertNewline2P(content);
1040                 content=createMailLinks(content,producerDocRoot,mailImage);
1041                 content=createURLLinks(content,null,producerDocRoot,extImage,intImage);
1042                 return content;
1043         }
1044
1045 }
1046