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
 255                 //Quickhack
 256                 //Because of '?-Bug in Postgresql-JDBC-Driver
 257                 StringBuffer temp = new StringBuffer();
 258                 for(int i=0;i<s.length();i++){
 259                         if(s.charAt(i)=='\''){
 260                                 temp.append("&acute;");
 261                         } else {
 262                                 temp.append(s.charAt(i));
 263                         }
 264                 }
 265                 String s2 = temp.toString();
 266                 //Quickhack end
 267
 268                 s2 = quote(s2, '\"');
 269                 return s2;
 270         }
 271
 272         /**
 273          * schließt <code>s</code> in <code>'</code> ein und setzt Backslashes vor
 274          * "gefährliche" Zeichen innerhalb des Strings
 275          *
 276          * @param s String, der gequoted werden soll
 277          * @param quoteChar zu quotendes Zeichen
 278          * @return gequoteter String
 279          */
 280         public static String quote(String s, char quoteChar)
 281         {
 282                 StringBuffer buf = new StringBuffer(s.length());
 283                 int pos = 0;
 284                 while (pos < s.length()) {
 285                         int i = s.indexOf(quoteChar, pos);
 286                         if (i < 0) i = s.length();
 287                         buf.append(s.substring(pos, i));
 288                         pos = i;
 289                         if (pos < s.length()) {
 290                                 buf.append('\\');
 291                                 buf.append(quoteChar);
 292                                 pos++;
 293                         }
 294                 }
 295                 return buf.toString();
 296         }
 297
 298         /**
 299          * replaces dangerous characters in <code>s</code>
 300          *
 301          */
 302
 303         public static String unquote(String s)
 304         {
 305                 char quoteChar='\'';
 306                 StringBuffer buf = new StringBuffer(s.length());
 307                 int pos = 0;
 308                 String searchString = "\\"+quoteChar;
 309                 while (pos < s.length()) {
 310                         int i = s.indexOf(searchString, pos);
 311                         if (i < 0) i = s.length();
 312                         buf.append(s.substring(pos, i));
 313                         pos = i+1;
 314                 }
 315                 return buf.toString();
 316         }
 317
 318         /**
 319          * Wandelet String in byte[] um.
 320          * @param s
 321          * @return byte[] des String
 322          */
 323
 324         public static byte[] stringToBytes(String s) {
 325                 String crlf = System.getProperty("line.separator");
 326                 if (!crlf.equals("\n"))
 327                         s = replace(s, "\n", crlf);
 328                 // byte[] buf = new byte[s.length()];
 329                 byte[] buf = s.getBytes();
 330                 return buf;
 331         }
 332
 333                 /**
 334          * Ersetzt in String <code>s</code> das <code>pattern</code> durch <code>substitute</code>
 335          * @param s
 336          * @param pattern
 337          * @param substitute
 338          * @return String mit den Ersetzungen
 339          */
 340         public static String replace(String s, String pattern, String substitute) {
 341                 int i = 0, pLen = pattern.length(), sLen = substitute.length();
 342                 StringBuffer buf = new StringBuffer(s.length());
 343                 while (true) {
 344                         int j = s.indexOf(pattern, i);
 345                         if (j < 0) {
 346                                 buf.append(s.substring(i));
 347                                 break;
 348                         } else {
 349                                 buf.append(s.substring(i, j));
 350                                 buf.append(substitute);
 351                                 i = j+pLen;
 352                         }
 353                 }
 354                 return buf.toString();
 355         }
 356
 357         /**
 358          * Ersetzt in String <code>s</code> das Regexp <code>pattern</code> durch <code>substitute</code>
 359          * @param s
 360          * @param pattern
 361          * @param substitute
 362          * @return String mit den Ersetzungen
 363          */
 364         public static String regexpReplace(String haystack, String pattern, String substitute) {
 365                 try {
 366                         RE regex = new RE(pattern);
 367                         return regex.substituteAll(haystack,substitute);
 368                 } catch(REException ex){
 369                         return null;
 370                 }
 371         }
 372
 373
 374
 375
 376         /**
 377          * Fügt einen Separator an den Pfad an
 378          * @param path
 379          * @return Pfad mit Separator am Ende
 380          */
 381         public static final String addSeparator (String path) {
 382                 return  path.length() == 0 || path.endsWith(File.separator) ? path : path
 383                                 + File.separatorChar;
 384         }
 385
 386         /**
 387          * Fügt ein <code>/</code> ans ende des Strings and
 388          * @param path
 389          * @return Pfad mit <code>/</code> am Ende
 390          */
 391         public static final String addSlash (String path) {
 392                 return  path.length() == 0 || path.endsWith("/") ? path : path + '/';
 393         }
 394
 395         /**
 396          * Löscht <code>/</code> am Ende des Strings, falls vorhanden
 397          * @param path
 398          * @return String ohne <code>/</code> am Ende
 399          */
 400         public static final String removeSlash (String path) {
 401                 return  path.length() > 1 && path.endsWith("/") ? path.substring(0, path.length()
 402                                 - 1) : path;
 403         }
 404
 405         /**
 406          * Checks to see if the path is absolute by looking for a leading file
 407          * separater
 408          * @param path
 409          * @return
 410          */
 411         public static boolean isAbsolutePath (String path) {
 412                 return  path.startsWith(File.separator);
 413         }
 414
 415         /**
 416          * Löscht Slash am Anfang des Strings
 417          * @param path
 418          * @return
 419          */
 420         public static String removeFirstSlash (String path) {
 421                 return  path.startsWith("/") ? path.substring(1) : path;
 422         }
 423
 424         /**
 425          * formatiert eine Zahl (0-99) zweistellig (z.B. 5 -> 05)
 426          * @return zwistellige Zahl
 427          */
 428         public static String pad2 (int number) {
 429                 return  number < 10 ? "0" + number : String.valueOf(number);
 430         }
 431
 432         /**
 433          * formatiert eine Zahl (0-999) dreistellig (z.B. 7 -> 007)
 434          *
 435          * @return 3-stellige Zahl
 436          */
 437         public static String pad3 (int number) {
 438                 return  number < 10 ? "00" + number : number < 100 ? "0" + number : String.valueOf(number);
 439         }
 440
 441         /**
 442          * Konvertiert Unix-Linefeeds in Win-Linefeeds
 443          * @param s
 444          * @return Konvertierter String
 445          */
 446         public static String unixLineFeedsToWin(String s) {
 447                 int i = -1;
 448                 while (true) {
 449                         i = s.indexOf('\n', i+1);
 450                         if (i < 0) break;
 451                         if ((i == 0 || s.charAt(i-1) != '\r') &&
 452                                 (i == s.length()-1 || s.charAt(i+1) != '\r')) {
 453                                 s = s.substring(0, i)+'\r'+s.substring(i);
 454                                 i++;
 455                         }
 456                 }
 457                 return s;
 458         }
 459
 460
 461         /**
 462          * verwandelt einen String in eine gültige Url, konvertiert Sonderzeichen
 463          * und Spaces werden zu Underscores
 464          *
 465          * @return gültige Url
 466          */
 467         public static String convert2url(String s) {
 468                 s = toLowerCase(s);
 469                 StringBuffer buf = new StringBuffer();
 470                 for(int i = 0; i < s.length(); i++ ) {
 471                                 switch( s.charAt( i ) ) {
 472                                 case 'ö':
 473                         buf.append( "oe" ); break;
 474                                 case 'ä':
 475                         buf.append( "ae" ); break;
 476                                 case 'ü':
 477                         buf.append( "ue" ); break;
 478                                 case 'ã':
 479                         buf.append( "a" ); break;
 480                                 case '´':
 481                                 case '.':
 482                         buf.append( "_" ); break;
 483                                 case ' ':
 484                         if( buf.charAt( buf.length() - 1 ) != '_' ) {
 485                                         buf.append( "_" );
 486                         }
 487                         break;
 488                                 default:
 489                         buf.append( s.charAt( i ) );
 490                                 }
 491                 }
 492                 return buf.toString();
 493         }
 494
 495         /**
 496          * wandelt Sonderzeichen in Quotes um
 497          *
 498          * @return Kovertierter String
 499          */
 500         public static String encodeHtml(String s) {
 501                 StringBuffer buf = new StringBuffer();
 502                 for(int i=0;i < s.length(); i++ ) {
 503
 504                         /** @todo looks inefficient, to ask for index of every char, in
 505                          *  case of failure it runs to the end.*/
 506                         if (s.charAt(i)=='&') {
 507                                 // convert html to xml-parsable representation
 508                                 if( s.indexOf( "&ouml;", i ) == i ) {
 509                                         buf.append( "&#246;" ); i += 5;
 510                                         continue;
 511                                 }
 512                                 if( s.indexOf( "&auml;", i ) == i ) {
 513                                         buf.append( "&#228;" ); i += 5;
 514                                         continue;
 515                                 }
 516                                 if( s.indexOf( "&uuml;", i ) == i ) {
 517                                         buf.append( "&#252;" ); i += 5;
 518                                         continue;
 519                                 }
 520                                 if( s.indexOf( "&Ouml;", i ) == i ) {
 521                                         buf.append( "&#214;" ); i += 5;
 522                                         continue;
 523                                 }
 524                                 if( s.indexOf( "&Auml;", i ) == i ) {
 525                                         buf.append( "&#196;" ); i += 5;
 526                                         continue;
 527                                 }
 528                                 if( s.indexOf( "&Uuml;", i ) == i ) {
 529                                         buf.append( "&#220;" ); i += 5;
 530                                         continue;
 531                                 }
 532                                 if( s.indexOf( "&szlig;", i ) == i ) {
 533                                         buf.append( "&#223;" ); i += 6;
 534                                         continue;
 535                                 }
 536
 537                                 /** @todo should only escape outside of tags */
 538
 539                                 if( s.indexOf( "&quot;", i ) == i ) {
 540                                         buf.append( "&#223;" ); i += 5;
 541                                         continue;
 542                                 }
 543                                 if( s.indexOf( "&ndash;", i ) == i ) {
 544                                         buf.append( "&#8211;" ); i += 6;
 545                                         continue;
 546                                 }
 547                                 if( s.indexOf( "&mdash;", i ) == i ) {
 548                                         buf.append( "&#8212;" ); i += 6;
 549                                         continue;
 550                                 }
 551                                 if( s.indexOf( "&ldquo;", i ) == i ) {
 552                                         buf.append( "&#8220;" ); i += 6;
 553                                         continue;
 554                                 }
 555                                 if( s.indexOf( "&rdquo;", i ) == i ) {
 556                                         buf.append( "&#8221;" ); i += 6;
 557                                         continue;
 558                                 }
 559                                 if( s.indexOf( "&bdquo;", i ) == i ) {
 560                                         buf.append( "&#8222;" ); i += 6;
 561                                         continue;
 562                                 }
 563
 564                                 //looks pretty stupid
 565                                 if( s.indexOf( "&lt;", i ) == i ) {
 566                                         buf.append( "&lt;" ); i += 3;
 567                                         continue;
 568                                 }
 569                                 if( s.indexOf( "&gt;", i ) == i ) {
 570                                         buf.append( "&gt;" ); i += 3;
 571                                         continue;
 572                                 }
 573                                 if( s.indexOf( "&acute;", i ) == i ) {
 574                                         buf.append( "&acute;" ); i += 6;
 575                                         continue;
 576                                 }
 577                                 if( s.indexOf( "&nbsp;", i ) == i ) {
 578                                         buf.append( "&nbsp;" ); i += 5;
 579                                         continue;
 580                                 }
 581                                 //has to be the last
 582                                 //if( s.indexOf( "&", i ) == i ) {
 583                                 //  buf.append( "&#38;" ); i += 0;
 584                                 //  continue;
 585                                 //}
 586                         }
 587                         // convert umlauts an other special charakters
 588                         switch( s.charAt(i) ) {
 589                                 case 'ö': buf.append( "&#246;" ); break;
 590                                 case 'ä': buf.append( "&#228;" ); break;
 591                                 case 'ü': buf.append( "&#252;" ); break;
 592                                 case 'Ö': buf.append( "&#214;" ); break;
 593                                 case 'Ä': buf.append( "&#196;" ); break;
 594                                 case 'Ü': buf.append( "&#220;" ); break;
 595                                 case 'ß': buf.append( "&#223;" ); break;
 596                                 case 'é': buf.append( "&#233;" ); break;
 597                                 case 'è': buf.append( "&#232;" ); break;
 598                                 case 'á': buf.append( "&#225;" ); break;
 599                                 case 'à': buf.append( "&#224;" ); break;
 600                                 case 'â': buf.append( "&#226;" ); break;
 601                                 case 'ã': buf.append( "&#227;" ); break;
 602                                 case '¬': buf.append( "&#172;" ); break;
 603                                 case '¹': buf.append( "&#185;" ); break;
 604                                 case '²': buf.append( "&#178;" ); break;
 605                                 case '³': buf.append( "&#179;" ); break;
 606                                 case '¼': buf.append( "&#188;" ); break;
 607                                 case '½': buf.append( "&#189;" ); break;
 608                                 case '¾': buf.append( "&#190;" ); break;
 609                                 case '¶': buf.append( "&#182;" ); break;
 610                                 case 'æ': buf.append( "&#230;" ); break;
 611                                 case 'ð': buf.append( "&#240;" ); break;
 612                                 case '|': buf.append( "&#166;" ); break;
 613                                 case '·': buf.append( "&#183;" ); break;
 614                                 case '°': buf.append( "&#176;" ); break;
 615                                 case '§': buf.append( "&#167;" ); break;
 616                                 case 'ø': buf.append( "&#248;" ); break;
 617                                 case 'ç': buf.append( "&#231;" ); break;
 618                                 case '¤': buf.append( "&#164;" ); break;
 619                                 case 'ª': buf.append( "&#170;" ); break;
 620                                 case 'Ç': buf.append( "&#199;" ); break;
 621                                 case 'Ã': buf.append( "&#195;" ); break;
 622                                 case 'Â': buf.append( "&#194;" ); break;
 623                                 case 'Æ': buf.append( "&#198;" ); break;
 624                                 case '©': buf.append( "&#169;" ); break;
 625                                 case '®': buf.append( "&#174;" ); break;
 626                                 case '¥': buf.append( "&#165;" ); break;
 627                                 case 'Þ': buf.append( "&#254;" ); break;
 628                                 case '¯': buf.append( "&#175;" ); break;
 629                                 case 'Ð': buf.append( "&#208;" ); break;
 630                                 case 'º': buf.append( "&#186;" ); break;
 631                                 case '¡': buf.append( "&#161;" ); break;
 632                                 case '£': buf.append( "&#163;" ); break;
 633                                 case '±': buf.append( "&#177;" ); break;
 634                                 case '¿': buf.append( "&#191;" ); break;
 635                                 case 'Ø': buf.append( "&#216;" ); break;
 636                                 case 'Á': buf.append( "&#192;" ); break;
 637                                 case 'À': buf.append( "&#193;" ); break;
 638                                 case 'É': buf.append( "&#200;" ); break;
 639                                 case 'È': buf.append( "&#201;" ); break;
 640                                 case 'ù': buf.append( "&#250;" ); break;
 641                                 case 'ñ': buf.append( "&#241;" ); break;
 642                                 case 'Ñ': buf.append( "&#209;" ); break;
 643                                 case 'µ': buf.append( "&#181;" ); break;
 644                                 case 'Í': buf.append( "&#204;" ); break;
 645                                 case 'Ì': buf.append( "&#205;" ); break;
 646                                 case 'í': buf.append( "&#236;" ); break;
 647                                 case 'ì': buf.append( "&#237;" ); break;
 648                                 case 'î': buf.append( "&#238;" ); break;
 649                                 case 'Î': buf.append( "&#206;" ); break;
 650                                 case 'ó': buf.append( "&#243;" ); break;
 651                                 case 'Ó': buf.append( "&#210;" ); break;
 652                                 case 'ò': buf.append( "&#206;" ); break;
 653                                 case 'Ò': buf.append( "&#211;" ); break;
 654                                 case 'ô': buf.append( "&#244;" ); break;
 655                                 case 'Ô': buf.append( "&#212;" ); break;
 656                                 case 'õ': buf.append( "&#245;" ); break;
 657                                 case 'Õ': buf.append( "&#213;" ); break;
 658                                 case 'ý': buf.append( "&#253;" ); break;
 659                                 case 'Ý': buf.append( "&#221;" ); break;
 660                                 case 'û': buf.append( "&#251;" ); break;
 661                                 case 'Û': buf.append( "&#219;" ); break;
 662                                 case 'ú': buf.append( "&#249;" ); break;
 663                                 case 'Ú': buf.append( "&#217;" ); break;
 664                                 case 'Ù': buf.append( "&#218;" ); break;
 665                                 case 'Ê': buf.append( "&#202;" ); break;
 666                                 case 'ê': buf.append( "&#234;" ); break;
 667                                 case 'å': buf.append( "&#229;" ); break;
 668                                 case 'Å': buf.append( "&#197;" ); break;
 669                                 case 'ë': buf.append( "&#235;" ); break;
 670                                 case 'Ë': buf.append( "&#203;" ); break;
 671                                 case 'ÿ': buf.append( "&#255;" ); break;
 672                                 case 'ï': buf.append( "&#239;" ); break;
 673                                 case 'Ï': buf.append( "&#207;" ); break;
 674                                 case '«': buf.append( "&#171;" ); break;
 675                                 case '»': buf.append( "&#187;" ); break;
 676                                 case '\'': buf.append( "&acute;" ); break;
 677                                 case '\"': buf.append( "&quot;" ); break;
 678                                 //case '\u8211': buf.append( "&#8211;" ); break;
 679                                 //case '\u8212': buf.append( "&#8212;" ); break;
 680                                 //case '\u8220': buf.append( "&#8220;" ); break;
 681                                 //case '\u8221': buf.append( "&#8221;" ); break;
 682                                 //case '\u8222': buf.append( "&#8222;" ); break;
 683                                 //case '\"': buf.append( "&#34;" ); break;
 684                                 default: buf.append( s.charAt(i) );
 685                         }
 686
 687                 }
 688                 return buf.toString();
 689         }
 690
 691
 692         public static String decodeHTMLinTags(String s){
 693                 StringBuffer buffer = new StringBuffer();
 694                 boolean start = false;
 695                 boolean stop = false;
 696                 int startIndex = 0;
 697                 int stopIndex = 0;
 698                 int temp = 0;
 699
 700                 for(int i=0;i<s.length();i++){
 701                         if(s.charAt(i)=='<'){
 702                                 start = true;
 703                                 startIndex = i;
 704                         } else if(s.charAt(i)=='>'){
 705                                 stop = true;
 706                                 stopIndex = i;
 707
 708                                 if(start && stop){
 709                                         buffer.append(s.substring(temp,startIndex));
 710                                         buffer.append(replaceQuot(s.substring(startIndex,stopIndex+1)));
 711                                         i= temp= stopIndex+1;
 712                                         start= stop= false;
 713                                 }
 714                         }
 715                 }
 716                 if(stopIndex>0){
 717                         buffer.append(s.substring(stopIndex+1));
 718                         return buffer.toString();
 719                 } else {
 720                         return s;
 721                 }
 722         }
 723
 724         public static String replaceQuot(String s) {
 725                 StringBuffer buffer = new StringBuffer();
 726                 for(int j = 0; j < s.length();j++){
 727                         if(s.charAt(j)=='&'){
 728                                 if(s.indexOf( "&quot;",j) == j) {
 729                                         buffer.append( "\"" );
 730                                         j += 5;
 731                                 }//if
 732                         } else {
 733                                 buffer.append(s.charAt(j));
 734                         }//else
 735                 }//for
 736                 return buffer.toString();
 737         }
 738
 739         /** wandelt Quotes in Sonderzeichen um
 740          */
 741         /**
 742         public static String decodeHtml(String s) {
 743                 StringBuffer buf = new StringBuffer();
 744                 for(int i=0;i < s.length(); i++ ) {
 745                         if( s.indexOf( "&ouml;", i ) == i ) {
 746                                 buf.append( "ö" ); i += 5;
 747                                 continue;
 748                         }
 749                         if( s.indexOf( "&auml;", i ) == i ) {
 750                                 buf.append( "ä" ); i += 5;
 751                                 continue;
 752                         }
 753                         if( s.indexOf( "&uuml;", i ) == i ) {
 754                                 buf.append( "ü" ); i += 5;
 755                                 continue;
 756                         }
 757                         if( s.indexOf( "&Ouml;", i ) == i ) {
 758                                 buf.append( "Ö" ); i += 5;
 759                                 continue;
 760                         }
 761                         if( s.indexOf( "&Auml;", i ) == i ) {
 762                                 buf.append( "Ä" ); i += 5;
 763                                 continue;
 764                         }
 765                         if( s.indexOf( "&Uuml;", i ) == i ) {
 766                                 buf.append( "Ü" ); i += 5;
 767                                 continue;
 768                         }
 769                         if( s.indexOf( "&szlig;", i ) == i ) {
 770                                 buf.append( "ß" ); i += 6;
 771                                 continue;
 772                         }
 773                         if( s.indexOf( "&quot;", i ) == i ) {
 774                                 buf.append( "\"" ); i += 5;
 775                                 continue;
 776                         }
 777                         buf.append( s.charAt(i) );
 778                 }
 779                 return buf.toString();
 780         }
 781          */
 782
 783         /**
 784          * schnellere Variante der String.toLowerCase()-Routine
 785          *
 786          * @return String in Kleinbuchsten
 787          */
 788         public static String toLowerCase(String s) {
 789                 int l = s.length();
 790                 char[] a = new char[l];
 791                 for (int i = 0; i < l; i++)
 792                         a[i] = Character.toLowerCase(s.charAt(i));
 793                 return new String(a);
 794         }
 795
 796                 /**
 797          * Findet <code>element</code> im String-Array <code>array</code>
 798          * @param array
 799          * @param element
 800          * @return Fundstelle als int oder -1
 801          */
 802         public static int indexOf(String[] array, String element) {
 803                 if (array != null)
 804                         for (int i = 0; i < array.length; i++)
 805                                 if (array[i].equals(element))
 806                                         return i;
 807                 return -1;
 808         }
 809
 810         /**
 811          * Testet auf Vorkommen von <code>element</code> in <code>array</code>
 812          * @param array String-Array
 813          * @param element
 814          * @return true wenn <code>element</code> vorkommt, sonst false
 815          */
 816         public static boolean contains(String[] array, String element) {
 817                 return indexOf(array, element) >= 0;
 818         }
 819
 820                 /**
 821          * Ermittelt CRC-Prüfsumme von String <code>s</code>
 822          * @param s
 823          * @return CRC-Prüfsumme
 824          */
 825         public static int getCRC(String s) {
 826                 int h = 0;
 827                 char val[] = s.toCharArray();
 828                 int len = val.length;
 829
 830                 for (int i = 0 ; i < len; i++) {
 831                         h &= 0x7fffffff;
 832                         h = (((h >> 30) | (h << 1)) ^ (val[i]+i));
 833                 }
 834
 835                 return (h << 8) | (len & 0xff);
 836         }
 837
 838                 /**
 839          * Liefert Default-Wert def zurück, wenn String <code>s</code>
 840          * kein Integer ist.
 841          *
 842          * @param s
 843          * @param def
 844          * @return geparster int aus s oder def
 845          */
 846         public static int parseInt(String s, int def) {
 847                 if (s == null) return def;
 848                 try {
 849                         return Integer.parseInt(s);
 850                 } catch (NumberFormatException e) {
 851                         return def;
 852                 }
 853         }
 854
 855         /**
 856          * Liefert Defaultwert def zurück, wenn s nicht zu einem float geparsed werden kann.
 857          * @param s
 858          * @param def
 859          * @return geparster float oder def
 860          */
 861         public static float parseFloat(String s, float def) {
 862                 if (s == null) return def;
 863                 try {
 864                         return new Float(s).floatValue();
 865                 } catch (NumberFormatException e) {
 866                         return def;
 867                 }
 868         }
 869
 870                 /**
 871          * Findet Ende eines Satzes in String <code>text</code>
 872          * @param text
 873          * @param startIndex
 874          * @return index des Satzendes, oder -1
 875          */
 876         public static int findEndOfSentence(String text, int startIndex) {
 877                  while (true) {
 878                          int i = text.indexOf('.', startIndex);
 879                          if (i < 0) return -1;
 880                          if (i > 0 && !Character.isDigit(text.charAt(i-1)) &&
 881                                         (i+1 >= text.length()
 882                                         || text.charAt(i+1) == ' '
 883                                         || text.charAt(i+1) == '\n'
 884                                         || text.charAt(i+1) == '\t'))
 885                                         return i+1;
 886                          startIndex = i+1;
 887                  }
 888         }
 889
 890                 /**
 891          * Findet Wortende in String <code>text</code> ab <code>startIndex</code>
 892          * @param text
 893          * @param startIndex
 894          * @return Index des Wortendes, oder -1
 895          */
 896         public static int findEndOfWord(String text, int startIndex) {
 897                 int i = text.indexOf(' ', startIndex),
 898                         j = text.indexOf('\n', startIndex);
 899                 if (i < 0) i = text.length();
 900                 if (j < 0) j = text.length();
 901                 return Math.min(i, j);
 902         }
 903
 904
 905         /**
 906          *  convertNewline2P ist eine regex-routine zum umwandeln von 2 oder mehr newlines (\n)
 907          *  in den html-tag <p>
 908          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 909          */
 910         public static String convertNewline2P(String haystack) {
 911                         return re_brbr2p.substituteAll(haystack,"\n</p><p>");
 912         }
 913
 914         /**
 915          *  convertNewline2Break ist eine regex-routine zum umwandeln von 1 newline (\n)
 916          *  in den html-tag <br>
 917          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 918          */
 919         public static String convertNewline2Break(String haystack) {
 920                 return re_newline2br.substituteAll(haystack,"$0<br>");
 921         }
 922
 923         /**
 924          *  createMailLinks wandelt text im email-adressenformat
 925          *  in einen klickbaren link um
 926          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 927          */
 928         public static String createMailLinks(String haystack) {
 929                         return re_mail.substituteAll(haystack,"<a href=\"mailto:$0\">$0</a>");
 930         }
 931
 932
 933         /**
 934          *  createMailLinks wandelt text im email-adressenformat
 935          *  in einen klickbaren link um
 936          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 937          */
 938         public static String createMailLinks(String haystack, String imageRoot, String mailImage) {
 939                 return re_mail.substituteAll(haystack,"<img src=\""+imageRoot+"/"+mailImage+"\" border=\"0\"/>&#160;<a href=\"mailto:$0\">$0</a>");
 940         }
 941
 942
 943         /**
 944          *  createURLLinks wandelt text im url-format
 945          *  in einen klickbaren link um
 946          *  nur sinnvoll, wenn text nicht im html-format eingegeben
 947          */
 948         public static String createURLLinks(String haystack) {
 949                 return re_url.substituteAll(haystack,"<a href=\"$0\">$0</a>");
 950         }
 951
 952         /**
 953          * this routine takes text in url format and makes
 954          * a clickaeble "<href>" link removing any "illegal" html tags
 955          * @param haystack, the url
 956          * @param title, the href link text
 957          * @param imagRoot, the place to find icons
 958          * @param extImage, the url of the icon to show next to the link
 959          * @return a String containing the url
 960          */
 961         public static String createURLLinks(String haystack, String title, String imageRoot,String extImage) {
 962                 if (title == null) {
 963                         return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/>&#160;<a href=\"$0\">$0</a>");
 964                 } else {
 965                         title = removeHTMLTags(title);
 966                         return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/>&#160;<a href=\"$0\">"+title+"</a>");
 967                 }
 968         }
 969
 970         /**
 971          * this routine takes text in url format and makes
 972          * a clickaeble "<href>" link removing any "illegal" html tags
 973          * @param haystack, the url
 974          * @param imageRoot, the place to find icons
 975          * @param extImage, the url of the icon to show next to the link
 976          * @param intImage, unused
 977          * @return a String containing the url
 978          */
 979         public static String createURLLinks(String haystack, String title, String imageRoot,String extImage,String intImage) {
 980                 return createURLLinks(haystack, title, imageRoot, extImage);
 981         }
 982
 983         /**
 984          * this routine takes text in url format and makes
 985          * an image link removing any "illegal" html tags
 986          * @param haystack, the url
 987          * @param title, the image alt text, can be null
 988          * @param height, height of the image
 989          * @param width, width of the image
 990          * @return a String containing the url
 991          */
 992         public static String createIMGLinks(String haystack, String title,
 993                                         String height,String width) {
 994         String wh="";
 995         if ( (height != null) && (width != null) )
 996         {
 997             wh = "width=\""+width+"\" height=\""+height+"\"";
 998         }
 999                 if (title != null) {
1000                         title = removeHTMLTags(title);
1001                         return re_url.substituteAll(haystack,
1002                                         "<img hspace=\"10\" vspace=\"6\" "+
1003                                         "align=\"left\" src=\"$0\" "+wh+
1004                                         " alt=\""+title+"\"/>&#160;<br><i>"+
1005                                         title+"</i>");
1006                 } else {
1007                         return re_url.substituteAll(haystack,
1008                                         "<img hspace=\"10\" vspace=\"6\" "+
1009                                         "align=\"left\" src=\"$0\" "+wh+
1010                                         " alt=\"\"/>&#160;");
1011                 }
1012         }
1013
1014
1015          /**
1016          *  deleteForbiddenTags
1017          *  this method deletes all <script>, <body> and <head>-tags
1018          */
1019         public static final String deleteForbiddenTags(String haystack) {
1020                 try {
1021                         RE regex = new RE("<[ \t\r\n](.*?)script(.*?)/script(.*?)>",RE.REG_ICASE);
1022                         haystack = regex.substituteAll(haystack,"");
1023                         regex = new RE("<head>(.*?)</head>");
1024                         haystack = regex.substituteAll(haystack,"");
1025                         regex = new RE("<[ \t\r\n/]*body(.*?)>");
1026                         haystack = regex.substituteAll(haystack,"");
1027                         return haystack;
1028                 } catch(REException ex){
1029                         return null;
1030                 }
1031         }
1032
1033         /**
1034          * this method deletes all html tags
1035          */
1036         public static final String removeHTMLTags(String haystack){
1037                         return re_tags.substituteAll(haystack,"");
1038         }
1039
1040
1041         /**
1042          * this method deletes all but the approved tags html tags
1043          * it also deletes approved tags which contain malicious-looking attributes and doesn't work at all
1044          */
1045         public static String approveHTMLTags(String haystack){
1046                 try {
1047                         String approvedTags="a|img|h1|h2|h3|h4|h5|h6|br|b|i|strong|p";
1048                         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";
1049                         String approvedProtocols="rtsp|http|ftp|https|freenet|mailto";
1050
1051                         // kill all the bad tags that have attributes
1052                         String s = "<\\s*/?\\s*(?!(("+approvedTags+")\\s))\\w+\\s[^>]*>";
1053                         RE regex = new RE(s,RE.REG_ICASE);
1054                         haystack = regex.substituteAll(haystack,"");
1055
1056                         // kill all the bad tags that are attributeless
1057                         regex = new RE("<\\s*/?\\s*(?!(("+approvedTags+")\\s*>))\\w+\\s*>",RE.REG_ICASE);
1058                         haystack = regex.substituteAll(haystack,"");
1059
1060                         // kill all the tags which have a javascript attribute like onLoad
1061                         regex = new RE("<[^>]*("+badAttributes+")[^>]*>",RE.REG_ICASE);
1062                         haystack = regex.substituteAll(haystack,"");
1063
1064                         // kill all the tags which include a url to an unacceptable protocol
1065                         regex = new RE("<\\s*a\\s+[^>]*href=(?!(\'|\")?("+approvedProtocols+"))[^>]*>",RE.REG_ICASE);
1066                         haystack = regex.substituteAll(haystack,"");
1067
1068                         return haystack;
1069                 } catch(REException ex){
1070                         ex.printStackTrace();
1071                         return null;
1072                 }
1073         }
1074
1075
1076         /**
1077          *  createHTML ruft alle regex-methoden zum unwandeln eines nicht
1078          *  htmlcodierten string auf und returnt einen htmlcodierten String
1079          */
1080         public static String createHTML(String content){
1081                 content=convertNewline2Break(content);
1082                 content=convertNewline2P(content);
1083                 content=createMailLinks(content);
1084                 content=createURLLinks(content);
1085                 return content;
1086         }
1087
1088
1089         /**
1090          *  createHTML ruft alle regex-methoden zum unwandeln eines nicht
1091          *  htmlcodierten string auf und returnt einen htmlcodierten String
1092          */
1093         public static String createHTML(String content,String producerDocRoot,String mailImage,String extImage,String intImage){
1094                 content=convertNewline2Break(content);
1095                 content=convertNewline2P(content);
1096                 content=createMailLinks(content,producerDocRoot,mailImage);
1097                 content=createURLLinks(content,null,producerDocRoot,extImage,intImage);
1098                 return content;
1099         }
1100
1101 }
1102