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      * Splits the provided CSV text into a list. stolen wholesale from
 190      * from Jakarta Turbine StrinUtils.java -mh
 191      *
 192      * @param text      The CSV list of values to split apart.
 193      * @param separator The separator character.
 194      * @return          The list of values.
 195      */
 196     public static String[] split(String text, String separator)
 197     {
 198         StringTokenizer st = new StringTokenizer(text, separator);
 199         String[] values = new String[st.countTokens()];
 200         int pos = 0;
 201         while (st.hasMoreTokens())
 202         {
 203             values[pos++] = st.nextToken();
 204         }
 205         return values;
 206     }
 207
 208     /**
 209      * Joins the elements of the provided array into a single string
 210      * containing a list of CSV elements. Stolen wholesale from Jakarta
 211      * Turbine StringUtils.java. -mh
 212      *
 213      * @param list      The list of values to join together.
 214      * @param separator The separator character.
 215      * @return          The CSV text.
 216      */
 217     public static String join(String[] list, String separator)
 218     {
 219         StringBuffer csv = new StringBuffer();
 220         for (int i = 0; i < list.length; i++)
 221         {
 222             if (i > 0)
 223             {
 224                 csv.append(separator);
 225             }
 226             csv.append(list[i]);
 227         }
 228         return csv.toString();
 229     }
 230
 231
 232         /**
 233          * schließt einen String in Anführungsszeichen ein, falls er Leerzeichen o.ä. enthält
 234          *
 235          * @return gequoteter String
 236          */
 237          public static String quoteIfNecessary(String s) {
 238                 for (int i = 0; i < s.length(); i++)
 239                         if (!(Character.isLetterOrDigit(s.charAt(i)) || s.charAt(i) == '.'))
 240                                 return quote(s, '"');
 241                 return s;
 242         }
 243
 244          /**
 245          * schließt <code>s</code> in <code>'</code> ein und setzt Backslashes vor
 246          * "gefährliche" Zeichen innerhalb des Strings
 247          * Quotes special SQL-characters in <code>s</code>
 248          *
 249          * @return geqoteter String
 250          */
 251         public static String quote(String s)
 252         {
 253                 String s2 = quote(s, '\'');
 254                 s2 = quote(s2, '\"');
 255                 return s2;
 256         }
 257
 258         /**
 259          * schließt <code>s</code> in <code>'</code> ein und setzt Backslashes vor
 260          * "gefährliche" Zeichen innerhalb des Strings
 261          *
 262          * @param s String, der gequoted werden soll
 263          * @param quoteChar zu quotendes Zeichen
 264          * @return gequoteter String
 265          */
 266         public static String quote(String s, char quoteChar)
 267         {
 268                 StringBuffer buf = new StringBuffer(s.length());
 269                 int pos = 0;
 270                 while (pos < s.length()) {
 271                         int i = s.indexOf(quoteChar, pos);
 272                         if (i < 0) i = s.length();
 273                         buf.append(s.substring(pos, i));
 274                         pos = i;
 275                         if (pos < s.length()) {
 276                                 buf.append('\\');
 277                                 buf.append(quoteChar);
 278                                 pos++;
 279                         }
 280                 }
 281                 return buf.toString();
 282         }
 283
 284         /**
 285          * replaces dangerous characters in <code>s</code>
 286          *
 287          */
 288
 289         public static String unquote(String s)
 290         {
 291                 char quoteChar='\'';
 292                 StringBuffer buf = new StringBuffer(s.length());
 293                 int pos = 0;
 294                 String searchString = "\\"+quoteChar;
 295                 while (pos < s.length()) {
 296                         int i = s.indexOf(searchString, pos);
 297                         if (i < 0) i = s.length();
 298                         buf.append(s.substring(pos, i));
 299                         pos = i+1;
 300                 }
 301                 return buf.toString();
 302         }
 303
 304         /**
 305          * Wandelet String in byte[] um.
 306          * @param s
 307          * @return byte[] des String
 308          */
 309
 310         public static byte[] stringToBytes(String s) {
 311                 String crlf = System.getProperty("line.separator");
 312                 if (!crlf.equals("\n"))
 313                         s = replace(s, "\n", crlf);
 314                 // byte[] buf = new byte[s.length()];
 315                 byte[] buf = s.getBytes();
 316                 return buf;
 317         }
 318
 319                 /**
 320          * Ersetzt in String <code>s</code> das <code>pattern</code> durch <code>substitute</code>
 321          * @param s
 322          * @param pattern
 323          * @param substitute
 324          * @return String mit den Ersetzungen
 325          */
 326         public static String replace(String s, String pattern, String substitute) {
 327                 int i = 0, pLen = pattern.length(), sLen = substitute.length();
 328                 StringBuffer buf = new StringBuffer(s.length());
 329                 while (true) {
 330                         int j = s.indexOf(pattern, i);
 331                         if (j < 0) {
 332                                 buf.append(s.substring(i));
 333                                 break;
 334                         } else {
 335                                 buf.append(s.substring(i, j));
 336                                 buf.append(substitute);
 337                                 i = j+pLen;
 338                         }
 339                 }
 340                 return buf.toString();
 341         }
 342
 343         /**
 344          * Ersetzt in String <code>s</code> das Regexp <code>pattern</code> durch <code>substitute</code>
 345          * @param s
 346          * @param pattern
 347          * @param substitute
 348          * @return String mit den Ersetzungen
 349          */
 350         public static String regexpReplace(String haystack, String pattern, String substitute) {
 351                 try {
 352                         RE regex = new RE(pattern);
 353                         return regex.substituteAll(haystack,substitute);
 354                 } catch(REException ex){
 355                         return null;
 356                 }
 357         }
 358
 359
 360
 361
 362         /**
 363          * Fügt einen Separator an den Pfad an
 364          * @param path
 365          * @return Pfad mit Separator am Ende
 366          */
 367         public static final String addSeparator (String path) {
 368                 return  path.length() == 0 || path.endsWith(File.separator) ? path : path
 369                                 + File.separatorChar;
 370         }
 371
 372         /**
 373          * Fügt ein <code>/</code> ans ende des Strings and
 374          * @param path
 375          * @return Pfad mit <code>/</code> am Ende
 376          */
 377         public static final String addSlash (String path) {
 378                 return  path.length() == 0 || path.endsWith("/") ? path : path + '/';
 379         }
 380
 381         /**
 382          * Löscht <code>/</code> am Ende des Strings, falls vorhanden
 383          * @param path
 384          * @return String ohne <code>/</code> am Ende
 385          */
 386         public static final String removeSlash (String path) {
 387                 return  path.length() > 1 && path.endsWith("/") ? path.substring(0, path.length()
 388                                 - 1) : path;
 389         }
 390
 391         /**
 392          * Checks to see if the path is absolute by looking for a leading file
 393          * separater
 394          * @param path
 395          * @return
 396          */
 397         public static boolean isAbsolutePath (String path) {
 398                 return  path.startsWith(File.separator);
 399         }
 400
 401         /**
 402          * Löscht Slash am Anfang des Strings
 403          * @param path
 404          * @return
 405          */
 406         public static String removeFirstSlash (String path) {
 407                 return  path.startsWith("/") ? path.substring(1) : path;
 408         }
 409
 410         /**
 411          * formatiert eine Zahl (0-99) zweistellig (z.B. 5 -> 05)
 412          * @return zwistellige Zahl
 413          */
 414         public static String pad2 (int number) {
 415                 return  number < 10 ? "0" + number : String.valueOf(number);
 416         }
 417
 418         /**
 419          * formatiert eine Zahl (0-999) dreistellig (z.B. 7 -> 007)
 420          *
 421          * @return 3-stellige Zahl
 422          */
 423         public static String pad3 (int number) {
 424                 return  number < 10 ? "00" + number : number < 100 ? "0" + number : String.valueOf(number);
 425         }
 426
 427         /**
 428          * Konvertiert Unix-Linefeeds in Win-Linefeeds
 429          * @param s
 430          * @return Konvertierter String
 431          */
 432         public static String unixLineFeedsToWin(String s) {
 433                 int i = -1;
 434                 while (true) {
 435                         i = s.indexOf('\n', i+1);
 436                         if (i < 0) break;
 437                         if ((i == 0 || s.charAt(i-1) != '\r') &&
 438                                 (i == s.length()-1 || s.charAt(i+1) != '\r')) {
 439                                 s = s.substring(0, i)+'\r'+s.substring(i);
 440                                 i++;
 441                         }
 442                 }
 443                 return s;
 444         }
 445
 446
 447         /**
 448          * verwandelt einen String in eine gültige Url, konvertiert Sonderzeichen
 449          * und Spaces werden zu Underscores
 450          *
 451          * @return gültige Url
 452          */
 453         public static String convert2url(String s) {
 454                 s = toLowerCase(s);
 455                 StringBuffer buf = new StringBuffer();
 456                 for(int i = 0; i < s.length(); i++ ) {
 457                                 switch( s.charAt( i ) ) {
 458                                 case 'ö':
 459                         buf.append( "oe" ); break;
 460                                 case 'ä':
 461                         buf.append( "ae" ); break;
 462                                 case 'ü':
 463                         buf.append( "ue" ); break;
 464                                 case 'ã':
 465                         buf.append( "a" ); break;
 466                                 case '´':
 467                                 case '.':
 468                         buf.append( "_" ); break;
 469                                 case ' ':
 470                         if( buf.charAt( buf.length() - 1 ) != '_' ) {
 471                                         buf.append( "_" );
 472                         }
 473                         break;
 474                                 default:
 475                         buf.append( s.charAt( i ) );
 476                                 }
 477                 }
 478                 return buf.toString();
 479         }
 480
 481         /**
 482          * wandelt Sonderzeichen in Quotes um
 483          *
 484          * @return Kovertierter String
 485          */
 486         public static String encodeHtml(String s) {
 487                 StringBuffer buf = new StringBuffer();
 488                 for(int i=0;i < s.length(); i++ ) {
 489
 490                         /** @todo looks inefficient, to ask for index of every char, in
 491                          *  case of failure it runs to the end.*/
 492                         if (s.charAt(i)=='&') {
 493                                 // convert html to xml-parsable representation
 494                                 if( s.indexOf( "&ouml;", i ) == i ) {
 495                                         buf.append( "&#246;" ); i += 5;
 496                                         continue;
 497                                 }
 498                                 if( s.indexOf( "&auml;", i ) == i ) {
 499                                         buf.append( "&#228;" ); i += 5;
 500                                         continue;
 501                                 }
 502                                 if( s.indexOf( "&uuml;", i ) == i ) {
 503                                         buf.append( "&#252;" ); i += 5;
 504                                         continue;
 505                                 }
 506                                 if( s.indexOf( "&Ouml;", i ) == i ) {
 507                                         buf.append( "&#214;" ); i += 5;
 508                                         continue;
 509                                 }
 510                                 if( s.indexOf( "&Auml;", i ) == i ) {
 511                                         buf.append( "&#196;" ); i += 5;
 512                                         continue;
 513                                 }
 514                                 if( s.indexOf( "&Uuml;", i ) == i ) {
 515                                         buf.append( "&#220;" ); i += 5;
 516                                         continue;
 517                                 }
 518                                 if( s.indexOf( "&szlig;", i ) == i ) {
 519                                         buf.append( "&#223;" ); i += 6;
 520                                         continue;
 521                                 }
 522
 523                                 /** @todo should only escape outside of tags */
 524
 525                                 if( s.indexOf( "&quot;", i ) == i ) {
 526                                         buf.append( "&#223;" ); i += 5;
 527                                         continue;
 528                                 }
 529                                 if( s.indexOf( "&ndash;", i ) == i ) {
 530                                         buf.append( "&#8211;" ); i += 6;
 531                                         continue;
 532                                 }
 533                                 if( s.indexOf( "&mdash;", i ) == i ) {
 534                                         buf.append( "&#8212;" ); i += 6;
 535                                         continue;
 536                                 }
 537                                 if( s.indexOf( "&ldquo;", i ) == i ) {
 538                                         buf.append( "&#8220;" ); i += 6;
 539                                         continue;
 540                                 }
 541                                 if( s.indexOf( "&rdquo;", i ) == i ) {
 542                                         buf.append( "&#8221;" ); i += 6;
 543                                         continue;
 544                                 }
 545                                 if( s.indexOf( "&bdquo;", i ) == i ) {
 546                                         buf.append( "&#8222;" ); i += 6;
 547                                         continue;
 548                                 }
 549
 550                                 //looks pretty stupid
 551                                 if( s.indexOf( "&lt;", i ) == i ) {
 552                                         buf.append( "&lt;" ); i += 3;
 553                                         continue;
 554                                 }
 555                                 if( s.indexOf( "&gt;", i ) == i ) {
 556                                         buf.append( "&gt;" ); i += 3;
 557                                         continue;
 558                                 }
 559                                 if( s.indexOf( "&acute;", i ) == i ) {
 560                                         buf.append( "&acute;" ); i += 6;
 561                                         continue;
 562                                 }
 563                                 if( s.indexOf( "&nbsp;", i ) == i ) {
 564                                         buf.append( "&nbsp;" ); i += 5;
 565                                         continue;
 566                                 }
 567                                 //has to be the last
 568                                 //if( s.indexOf( "&", i ) == i ) {
 569                                 //  buf.append( "&#38;" ); i += 0;
 570                                 //  continue;
 571                                 //}
 572                         }
 573                         // convert umlauts an other special charakters
 574                         switch( s.charAt(i) ) {
 575                                 case 'ö': buf.append( "&#246;" ); break;
 576                                 case 'ä': buf.append( "&#228;" ); break;
 577                                 case 'ü': buf.append( "&#252;" ); break;
 578                                 case 'Ö': buf.append( "&#214;" ); break;
 579                                 case 'Ä': buf.append( "&#196;" ); break;
 580                                 case 'Ü': buf.append( "&#220;" ); break;
 581                                 case 'ß': buf.append( "&#223;" ); break;
 582                                 case 'é': buf.append( "&#233;" ); break;
 583                                 case 'è': buf.append( "&#232;" ); break;
 584                                 case 'á': buf.append( "&#225;" ); break;
 585                                 case 'à': buf.append( "&#224;" ); break;
 586                                 case 'â': buf.append( "&#226;" ); break;
 587                                 case 'ã': buf.append( "&#227;" ); break;
 588                                 case '¬': buf.append( "&#172;" ); break;
 589                                 case '¹': buf.append( "&#185;" ); break;
 590                                 case '²': buf.append( "&#178;" ); break;
 591                                 case '³': buf.append( "&#179;" ); break;
 592                                 case '¼': buf.append( "&#188;" ); break;
 593                                 case '½': buf.append( "&#189;" ); break;
 594                                 case '¾': buf.append( "&#190;" ); break;
 595                                 case '¶': buf.append( "&#182;" ); break;
 596                                 case 'æ': buf.append( "&#230;" ); break;
 597                                 case 'ð': buf.append( "&#240;" ); break;
 598                                 case '|': buf.append( "&#166;" ); break;
 599                                 case '·': buf.append( "&#183;" ); break;
 600                                 case '°': buf.append( "&#176;" ); break;
 601                                 case '§': buf.append( "&#167;" ); break;
 602                                 case 'ø': buf.append( "&#248;" ); break;
 603                                 case 'ç': buf.append( "&#231;" ); break;
 604                                 case '¤': buf.append( "&#164;" ); break;
 605                                 case 'ª': buf.append( "&#170;" ); break;
 606                                 case 'Ç': buf.append( "&#199;" ); break;
 607                                 case 'Ã': buf.append( "&#195;" ); break;
 608                                 case 'Â': buf.append( "&#194;" ); break;
 609                                 case 'Æ': buf.append( "&#198;" ); break;
 610                                 case '©': buf.append( "&#169;" ); break;
 611                                 case '®': buf.append( "&#174;" ); break;
 612                                 case '¥': buf.append( "&#165;" ); break;
 613                                 case 'Þ': buf.append( "&#254;" ); break;
 614                                 case '¯': buf.append( "&#175;" ); break;
 615                                 case 'Ð': buf.append( "&#208;" ); break;
 616                                 case 'º': buf.append( "&#186;" ); break;
 617                                 case '¡': buf.append( "&#161;" ); break;
 618                                 case '£': buf.append( "&#163;" ); break;
 619                                 case '±': buf.append( "&#177;" ); break;
 620                                 case '¿': buf.append( "&#191;" ); break;
 621                                 case 'Ø': buf.append( "&#216;" ); break;
 622                                 case 'Á': buf.append( "&#192;" ); break;
 623                                 case 'À': buf.append( "&#193;" ); break;
 624                                 case 'É': buf.append( "&#200;" ); break;
 625                                 case 'È': buf.append( "&#201;" ); break;
 626                                 case 'ù': buf.append( "&#250;" ); break;
 627                                 case 'ñ': buf.append( "&#241;" ); break;
 628                                 case 'Ñ': buf.append( "&#209;" ); break;
 629                                 case 'µ': buf.append( "&#181;" ); break;
 630                                 case 'Í': buf.append( "&#204;" ); break;
 631                                 case 'Ì': buf.append( "&#205;" ); break;
 632                                 case 'í': buf.append( "&#236;" ); break;
 633                                 case 'ì': buf.append( "&#237;" ); break;
 634                                 case 'î': buf.append( "&#238;" ); break;
 635                                 case 'Î': buf.append( "&#206;" ); break;
 636                                 case 'ó': buf.append( "&#243;" ); break;
 637                                 case 'Ó': buf.append( "&#210;" ); break;
 638                                 case 'ò': buf.append( "&#206;" ); break;
 639                                 case 'Ò': buf.append( "&#211;" ); break;
 640                                 case 'ô': buf.append( "&#244;" ); break;
 641                                 case 'Ô': buf.append( "&#212;" ); break;
 642                                 case 'õ': buf.append( "&#245;" ); break;
 643                                 case 'Õ': buf.append( "&#213;" ); break;
 644                                 case 'ý': buf.append( "&#253;" ); break;
 645                                 case 'Ý': buf.append( "&#221;" ); break;
 646                                 case 'û': buf.append( "&#251;" ); break;
 647                                 case 'Û': buf.append( "&#219;" ); break;
 648                                 case 'ú': buf.append( "&#249;" ); break;
 649                                 case 'Ú': buf.append( "&#217;" ); break;
 650                                 case 'Ù': buf.append( "&#218;" ); break;
 651                                 case 'Ê': buf.append( "&#202;" ); break;
 652                                 case 'ê': buf.append( "&#234;" ); break;
 653                                 case 'å': buf.append( "&#229;" ); break;
 654                                 case 'Å': buf.append( "&#197;" ); break;
 655                                 case 'ë': buf.append( "&#235;" ); break;
 656                                 case 'Ë': buf.append( "&#203;" ); break;
 657                                 case 'ÿ': buf.append( "&#255;" ); break;
 658                                 case 'ï': buf.append( "&#239;" ); break;
 659                                 case 'Ï': buf.append( "&#207;" ); break;
 660                                 case '«': buf.append( "&#171;" ); break;
 661                                 case '»': buf.append( "&#187;" ); break;
 662                                 case '\'': buf.append( "&acute;" ); break;
 663                                 case '\"': buf.append( "&quot;" ); break;
 664                                 //case '\u8211': buf.append( "&#8211;" ); break;
 665                                 //case '\u8212': buf.append( "&#8212;" ); break;
 666                                 //case '\u8220': buf.append( "&#8220;" ); break;
 667                                 //case '\u8221': buf.append( "&#8221;" ); break;
 668                                 //case '\u8222': buf.append( "&#8222;" ); break;
 669                                 //case '\"': buf.append( "&#34;" ); break;
 670                                 default: buf.append( s.charAt(i) );
 671                         }
 672
 673                 }
 674                 return buf.toString();
 675         }
 676
 677
 678         public static String decodeHTMLinTags(String s){
 679                 StringBuffer buffer = new StringBuffer();
 680                 boolean start = false;
 681                 boolean stop = false;
 682                 int startIndex = 0;
 683                 int stopIndex = 0;
 684                 int temp = 0;
 685
 686                 for(int i=0;i<s.length();i++){
 687                         if(s.charAt(i)=='<'){
 688                                 start = true;
 689                                 startIndex = i;
 690                         } else if(s.charAt(i)=='>'){
 691                                 stop = true;
 692                                 stopIndex = i;
 693
 694                                 if(start && stop){
 695                                         buffer.append(s.substring(temp,startIndex));
 696                                         buffer.append(replaceQuot(s.substring(startIndex,stopIndex+1)));
 697                                         i= temp= stopIndex+1;
 698                                         start= stop= false;
 699                                 }
 700                         }
 701                 }
 702                 if(stopIndex>0){
 703                         buffer.append(s.substring(stopIndex+1));
 704                         return buffer.toString();
 705                 } else {
 706                         return s;
 707                 }
 708         }
 709
 710         public static String replaceQuot(String s) {
 711                 StringBuffer buffer = new StringBuffer();
 712                 for(int j = 0; j < s.length();j++){
 713                         if(s.charAt(j)=='&'){
 714                                 if(s.indexOf( "&quot;",j) == j) {
 715                                         buffer.append( "\"" );
 716                                         j += 5;
 717                                 }//if
 718                         } else {
 719                                 buffer.append(s.charAt(j));
 720                         }//else
 721                 }//for
 722                 return buffer.toString();
 723         }
 724
 725         /** wandelt Quotes in Sonderzeichen um
 726          */
 727         /**
 728         public static String decodeHtml(String s) {
 729                 StringBuffer buf = new StringBuffer();
 730                 for(int i=0;i < s.length(); i++ ) {
 731                         if( s.indexOf( "&ouml;", i ) == i ) {
 732                                 buf.append( "ö" ); i += 5;
 733                                 continue;
 734                         }
 735                         if( s.indexOf( "&auml;", i ) == i ) {
 736                                 buf.append( "ä" ); i += 5;
 737                                 continue;
 738                         }
 739                         if( s.indexOf( "&uuml;", i ) == i ) {
 740                                 buf.append( "ü" ); i += 5;
 741                                 continue;
 742                         }
 743                         if( s.indexOf( "&Ouml;", i ) == i ) {
 744                                 buf.append( "Ö" ); i += 5;
 745                                 continue;
 746                         }
 747                         if( s.indexOf( "&Auml;", i ) == i ) {
 748                                 buf.append( "Ä" ); i += 5;
 749                                 continue;
 750                         }
 751                         if( s.indexOf( "&Uuml;", i ) == i ) {
 752                                 buf.append( "Ü" ); i += 5;
 753                                 continue;
 754                         }
 755                         if( s.indexOf( "&szlig;", i ) == i ) {
 756                                 buf.append( "ß" ); i += 6;
 757                                 continue;
 758                         }
 759                         if( s.indexOf( "&quot;", i ) == i ) {
 760                                 buf.append( "\"" ); i += 5;
 761                                 continue;
 762                         }
 763                         buf.append( s.charAt(i) );
 764                 }
 765                 return buf.toString();
 766         }
 767          */
 768
 769         /**
 770          * schnellere Variante der String.toLowerCase()-Routine
 771          *
 772          * @return String in Kleinbuchsten
 773          */
 774         public static String toLowerCase(String s) {
 775                 int l = s.length();
 776                 char[] a = new char[l];
 777                 for (int i = 0; i < l; i++)
 778                         a[i] = Character.toLowerCase(s.charAt(i));
 779                 return new String(a);
 780         }
 781
 782                 /**
 783          * Findet <code>element</code> im String-Array <code>array</code>
 784          * @param array
 785          * @param element
 786          * @return Fundstelle als int oder -1
 787          */
 788         public static int indexOf(String[] array, String element) {
 789                 if (array != null)
 790                         for (int i = 0; i < array.length; i++)
 791                                 if (array[i].equals(element))
 792                                         return i;
 793                 return -1;
 794         }
 795
 796         /**
 797          * Testet auf Vorkommen von <code>element</code> in <code>array</code>
 798          * @param array String-Array
 799          * @param element
 800          * @return true wenn <code>element</code> vorkommt, sonst false
 801          */
 802         public static boolean contains(String[] array, String element) {
 803                 return indexOf(array, element) >= 0;
 804         }
 805
 806                 /**
 807          * Ermittelt CRC-Prüfsumme von String <code>s</code>
 808          * @param s
 809          * @return CRC-Prüfsumme
 810          */
 811         public static int getCRC(String s) {
 812                 int h = 0;
 813                 char val[] = s.toCharArray();
 814                 int len = val.length;
 815
 816                 for (int i = 0 ; i < len; i++) {
 817                         h &= 0x7fffffff;
 818                         h = (((h >> 30) | (h << 1)) ^ (val[i]+i));
 819                 }
 820
 821                 return (h << 8) | (len & 0xff);
 822         }
 823
 824                 /**
 825          * Liefert Default-Wert def zurück, wenn String <code>s</code>
 826          * kein Integer ist.
 827          *
 828          * @param s
 829          * @param def
 830          * @return geparster int aus s oder def
 831          */
 832         public static int parseInt(String s, int def) {
 833                 if (s == null) return def;
 834                 try {
 835                         return Integer.parseInt(s);
 836                 } catch (NumberFormatException e) {
 837                         return def;
 838                 }
 839         }
 840
 841         /**
 842          * Liefert Defaultwert def zurück, wenn s nicht zu einem float geparsed werden kann.
 843          * @param s
 844          * @param def
 845          * @return geparster float oder def
 846          */
 847         public static float parseFloat(String s, float def) {
 848                 if (s == null) return def;
 849                 try {
 850                         return new Float(s).floatValue();
 851                 } catch (NumberFormatException e) {
 852                         return def;
 853                 }
 854         }
 855
 856                 /**
 857          * Findet Ende eines Satzes in String <code>text</code>
 858          * @param text
 859          * @param startIndex
 860          * @return index des Satzendes, oder -1
 861          */
 862         public static int findEndOfSentence(String text, int startIndex) {
 863                  while (true) {
 864                          int i = text.indexOf('.', startIndex);
 865                          if (i < 0) return -1;
 866                          if (i > 0 && !Character.isDigit(text.charAt(i-1)) &&
 867                                         (i+1 >= text.length()
 868                                         || text.charAt(i+1) == ' '
 869                                         || text.charAt(i+1) == '\n'
 870                                         || text.charAt(i+1) == '\t'))
 871                                         return i+1;
 872                          startIndex = i+1;
 873                  }
 874         }
 875
 876                 /**
 877          * Findet Wortende in String <code>text</code> ab <code>startIndex</code>
 878          * @param text
 879          * @param startIndex
 880          * @return Index des Wortendes, oder -1
 881          */
 882         public static int findEndOfWord(String text, int startIndex) {
 883                 int i = text.indexOf(' ', startIndex),
 884                         j = text.indexOf('\n', startIndex);
 885                 if (i < 0) i = text.length();
 886                 if (j < 0) j = text.length();
 887                 return Math.min(i, j);
 888         }
 889
 890
 891         /**
 892          *  convertNewline2P ist eine regex-routine zum umwandeln von 2 oder mehr newlines (\n)
 893          *  in den html-tag <p>
 894          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 895          */
 896         public static String convertNewline2P(String haystack) {
 897                         return re_brbr2p.substituteAll(haystack,"\n</p><p>");
 898         }
 899
 900         /**
 901          *  convertNewline2Break ist eine regex-routine zum umwandeln von 1 newline (\n)
 902          *  in den html-tag <br>
 903          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 904          */
 905         public static String convertNewline2Break(String haystack) {
 906                 return re_newline2br.substituteAll(haystack,"$0<br />");
 907         }
 908
 909         /**
 910          *  createMailLinks wandelt text im email-adressenformat
 911          *  in einen klickbaren link um
 912          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 913          */
 914         public static String createMailLinks(String haystack) {
 915                         return re_mail.substituteAll(haystack,"<a href=\"mailto:$0\">$0</a>");
 916         }
 917
 918
 919         /**
 920          *  createMailLinks wandelt text im email-adressenformat
 921          *  in einen klickbaren link um
 922          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 923          */
 924         public static String createMailLinks(String haystack, String imageRoot, String mailImage) {
 925                 return re_mail.substituteAll(haystack,"<img src=\""+imageRoot+"/"+mailImage+"\" border=\"0\"/>&#160;<a href=\"mailto:$0\">$0</a>");
 926         }
 927
 928
 929         /**
 930          *  createURLLinks wandelt text im url-format
 931          *  in einen klickbaren link um
 932          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 933          */
 934         public static String createURLLinks(String haystack) {
 935                 return re_url.substituteAll(haystack,"<a href=\"$0\">$0</a>");
 936         }
 937
 938         /**
 939          * this routine takes text in url format and makes
 940          * a clickaeble "<href>" link removing any "illegal" html tags
 941          * @param haystack, the url
 942          * @param title, the href link text
 943          * @param imagRoot, the place to find icons
 944          * @param extImage, the url of the icon to show next to the link
 945          * @return a String containing the url
 946          */
 947         public static String createURLLinks(String haystack, String title, String imageRoot,String extImage) {
 948                 if (title == null) {
 949                         return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/>&#160;<a href=\"$0\">$0</a>");
 950                 } else {
 951                         title = removeHTMLTags(title);
 952                         return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/>&#160;<a href=\"$0\">"+title+"</a>");
 953                 }
 954         }
 955
 956         /**
 957          * this routine takes text in url format and makes
 958          * a clickaeble "<href>" link removing any "illegal" html tags
 959          * @param haystack, the url
 960          * @param imageRoot, the place to find icons
 961          * @param extImage, the url of the icon to show next to the link
 962          * @param intImage, unused
 963          * @return a String containing the url
 964          */
 965         public static String createURLLinks(String haystack, String title, String imageRoot,String extImage,String intImage) {
 966                 return createURLLinks(haystack, title, imageRoot, extImage);
 967         }
 968
 969          /**
 970          *  deleteForbiddenTags
 971          *  this method deletes all <script>, <body> and <head>-tags
 972          */
 973         public static final String deleteForbiddenTags(String haystack) {
 974                 try {
 975                         RE regex = new RE("<[ \t\r\n](.*?)script(.*?)/script(.*?)>",RE.REG_ICASE);
 976                         haystack = regex.substituteAll(haystack,"");
 977                         regex = new RE("<head>(.*?)</head>");
 978                         haystack = regex.substituteAll(haystack,"");
 979                         regex = new RE("<[ \t\r\n/]*body(.*?)>");
 980                         haystack = regex.substituteAll(haystack,"");
 981                         return haystack;
 982                 } catch(REException ex){
 983                         return null;
 984                 }
 985         }
 986
 987         /**
 988          * this method deletes all html tags
 989          */
 990         public static final String removeHTMLTags(String haystack){
 991                         return re_tags.substituteAll(haystack,"");
 992         }
 993
 994
 995         /**
 996          * this method deletes all but the approved tags html tags
 997          * it also deletes approved tags which contain malicious-looking attributes and doesn't work at all
 998          */
 999         public static String approveHTMLTags(String haystack){
1000                 try {
1001                         String approvedTags="a|img|h1|h2|h3|h4|h5|h6|br|b|i|strong|p";
1002                         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";
1003                         String approvedProtocols="rtsp|http|ftp|https|freenet|mailto";
1004
1005                         // kill all the bad tags that have attributes
1006                         String s = "<\\s*/?\\s*(?!(("+approvedTags+")\\s))\\w+\\s[^>]*>";
1007                         RE regex = new RE(s,RE.REG_ICASE);
1008                         haystack = regex.substituteAll(haystack,"");
1009
1010                         // kill all the bad tags that are attributeless
1011                         regex = new RE("<\\s*/?\\s*(?!(("+approvedTags+")\\s*>))\\w+\\s*>",RE.REG_ICASE);
1012                         haystack = regex.substituteAll(haystack,"");
1013
1014                         // kill all the tags which have a javascript attribute like onLoad
1015                         regex = new RE("<[^>]*("+badAttributes+")[^>]*>",RE.REG_ICASE);
1016                         haystack = regex.substituteAll(haystack,"");
1017
1018                         // kill all the tags which include a url to an unacceptable protocol
1019                         regex = new RE("<\\s*a\\s+[^>]*href=(?!(\'|\")?("+approvedProtocols+"))[^>]*>",RE.REG_ICASE);
1020                         haystack = regex.substituteAll(haystack,"");
1021
1022                         return haystack;
1023                 } catch(REException ex){
1024                         ex.printStackTrace();
1025                         return null;
1026                 }
1027         }
1028
1029
1030         /**
1031          *  createHTML ruft alle regex-methoden zum unwandeln eines nicht
1032          *  htmlcodierten string auf und returnt einen htmlcodierten String
1033          */
1034         public static String createHTML(String content){
1035                 content=convertNewline2Break(content);
1036                 content=convertNewline2P(content);
1037                 content=createMailLinks(content);
1038                 content=createURLLinks(content);
1039                 return content;
1040         }
1041
1042
1043         /**
1044          *  createHTML ruft alle regex-methoden zum unwandeln eines nicht
1045          *  htmlcodierten string auf und returnt einen htmlcodierten String
1046          */
1047         public static String createHTML(String content,String producerDocRoot,String mailImage,String extImage,String intImage){
1048                 content=convertNewline2Break(content);
1049                 content=convertNewline2P(content);
1050                 content=createMailLinks(content,producerDocRoot,mailImage);
1051                 content=createURLLinks(content,null,producerDocRoot,extImage,intImage);
1052                 return content;
1053         }
1054
1055 }
1056