2 * Copyright (C) 2001, 2002 The Mir-coders group
4 * This file is part of Mir.
6 * Mir is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
11 * Mir is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License
17 * along with Mir; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 * In addition, as a special exception, The Mir-coders gives permission to link
21 * the code of this program with any library licensed under the Apache Software License,
22 * The Sun (tm) Java Advanced Imaging library (JAI), The Sun JIMI library
23 * (or with modified versions of the above that use the same license as the above),
24 * and distribute linked combinations including the two. You must obey the
25 * GNU General Public License in all respects for all of the code used other than
26 * the above mentioned libraries. If you modify this file, you may extend this
27 * exception to your version of the file, but you are not obligated to do so.
28 * If you do not wish to do so, delete this exception statement from your version.
33 import gnu.regexp.REException;
36 import java.text.NumberFormat;
37 import java.util.Calendar;
38 import java.util.GregorianCalendar;
39 import java.util.StringTokenizer;
42 * Statische Hilfsmethoden zur Stringbehandlung
44 * @version $Id: StringUtil.java,v 1.33 2003/04/21 12:42:52 idfx Exp $
45 * @author mir-coders group
48 public final class StringUtil {
50 private static RE re_newline2br, re_brbr2p, re_mail, re_url, re_tags,
51 re_tables, re_forbiddenTags;
53 private StringUtil() { } // this avoids contruction
58 re_newline2br = new RE("(\r?\n){1}");
59 re_brbr2p = new RE("(<br>\r?\n<br>){1,}");
60 re_mail = new RE("([a-zA-Z0-9_.-]+)@([a-zA-Z0-9_-]+)\\.([a-zA-Z0-9_.-]+)");
61 re_url = new RE("((https://)|(http://)|(ftp://)){1}([a-zA-Z0-9_-]+).([a-zA-Z0-9_.:-]+)/?([^ \t\r\n<>\\)\\]]+[^ \t\r\n.,<>\\)\\]])");
62 re_tags = new RE("<[^>]*>",RE.REG_ICASE);
63 re_tables = new RE("<[ \t\r\n/]*(table|td|tr)[ \t\r\n]*>",RE.REG_ICASE);
64 re_forbiddenTags = new RE("<[ \t\r\n/]*(body|head|script)[ \t\r\n]*>",RE.REG_ICASE);
66 catch (REException e){
67 System.err.println("FATAL: StringUtil: could not precompile REGEX: "+e.toString());
72 * Formats a number with the specified minimum and maximum number of digits.
74 public static synchronized String zeroPaddingNumber(long value, int minDigits,
77 NumberFormat numberFormat = NumberFormat.getInstance();
78 numberFormat.setMinimumIntegerDigits(minDigits);
79 numberFormat.setMaximumIntegerDigits(maxDigits);
80 return numberFormat.format(value);
84 * Wandelt Datum in einen 8-ziffrigen String um (yyyymmdd)
86 * @return 8-ziffriger String (yyyymmdd)
89 public static final String date2webdbDate (GregorianCalendar theDate) {
90 StringBuffer webdbDate = new StringBuffer();
91 webdbDate.append(String.valueOf(theDate.get(Calendar.YEAR)));
92 webdbDate.append(pad2(theDate.get(Calendar.MONTH) + 1));
93 webdbDate.append(pad2(theDate.get(Calendar.DATE)));
94 return webdbDate.toString();
98 * Wandelt Calendar in einen 12-ziffrigen String um (yyyymmddhhmm)
100 * @return 12-ziffriger String (yyyymmdd)
103 public static final String date2webdbDateTime (GregorianCalendar theDate) {
104 StringBuffer webdbDate = new StringBuffer();
105 webdbDate.append(String.valueOf(theDate.get(Calendar.YEAR)));
106 webdbDate.append(pad2(theDate.get(Calendar.MONTH) + 1));
107 webdbDate.append(pad2(theDate.get(Calendar.DATE)));
108 webdbDate.append(pad2(theDate.get(Calendar.HOUR)));
109 webdbDate.append(pad2(theDate.get(Calendar.MINUTE)));
110 return webdbDate.toString();
114 * Return a http://www.w3.org/TR/NOTE-datetime formatted date (yyyy-mm-ddThh:mm:ssTZ)
116 * @return w3approved datetime
119 public static final String date2w3DateTime (GregorianCalendar theDate) {
120 StringBuffer webdbDate = new StringBuffer();
121 webdbDate.append(String.valueOf(theDate.get(Calendar.YEAR)));
122 webdbDate.append("-");
123 webdbDate.append(pad2(theDate.get(Calendar.MONTH) + 1));
124 webdbDate.append("-");
125 webdbDate.append(pad2(theDate.get(Calendar.DATE)));
126 webdbDate.append("T");
127 webdbDate.append(pad2(theDate.get(Calendar.HOUR)));
128 webdbDate.append(":");
129 webdbDate.append(pad2(theDate.get(Calendar.MINUTE)));
130 webdbDate.append(":");
131 webdbDate.append(pad2(theDate.get(Calendar.SECOND)));
132 //assumes you are an hour-multiple away from UTC....
133 int offset=(theDate.get(Calendar.ZONE_OFFSET)/(60*60*1000));
135 webdbDate.append("-");
138 webdbDate.append("+");
140 webdbDate.append(pad2(Math.abs(offset)));
141 webdbDate.append(":00");
142 return webdbDate.toString();
146 * wandelt Calendar in dd.mm.yyyy / hh.mm um
148 * @return String mit (dd.mm.yyyy / hh.mm um)
150 public static String date2readableDateTime (GregorianCalendar theDate) {
151 String readable = "";
153 readable += pad2(theDate.get(Calendar.DATE));
154 readable += "." + pad2(theDate.get(Calendar.MONTH) + 1);
155 readable += "." + String.valueOf(theDate.get(Calendar.YEAR));
156 hour = theDate.get(Calendar.HOUR);
157 if (theDate.get(Calendar.AM_PM) == Calendar.PM)
159 readable += " / " + pad2(hour);
160 readable += ":" + pad2(theDate.get(Calendar.MINUTE));
165 * deleteForbiddenTags
166 * this method deletes all <script>, <body> and <head>-tags
168 public static final String deleteForbiddenTags(String haystack) {
169 return re_forbiddenTags.substituteAll(haystack,"");
173 * deleteHTMLTableTags
174 * this method deletes all <table>, <tr> and <td>-tags
176 public static final String deleteHTMLTableTags(String haystack) {
177 return re_tables.substituteAll(haystack,"");
181 * wandelt eine Datum in einen 8-buchstabigen String, der durch <code>/</code>
185 * @return String mit <code>/yyyy/mm/dd</code>
187 public static final String webdbDate2path (String webdbDate) {
188 StringBuffer path = new StringBuffer();
189 path.append("/").append(webdbDate.substring(0, 4));
190 path.append("/").append(webdbDate.substring(4, 6));
193 //path.append("/").append(webdbDate.substring(6, 8));
194 return path.toString();
198 * converts string from format: yyyy-mm-dd__hh:mm:ss.d
199 * to dd.mm.yyyy hh:mm
201 public static String dateToReadableDate(String date) {
202 StringBuffer returnDate = new StringBuffer();
205 returnDate.append(date.substring(8,10)).append('.');
206 returnDate.append(date.substring(5,7)).append('.');
207 returnDate.append(date.substring(0,4)).append(' ');
208 returnDate.append(date.substring(11,16));
210 return returnDate.toString();
214 * converts string from format: yyyy-mm-dd__hh:mm:ss.d
217 public static String dateToYear (String date) {
218 StringBuffer returnDate = new StringBuffer();
221 returnDate.append(date.substring(0,4));
223 return returnDate.toString();
227 * converts string from format: yyyy-mm-dd__hh:mm:ss.d
230 public static String dateToMonth (String date) {
231 StringBuffer returnDate = new StringBuffer();
233 if (!date.substring(5,6).equalsIgnoreCase("0")) returnDate.append(date.substring(5,7));
234 else returnDate.append(date.substring(6,7));
236 return returnDate.toString();
240 * converts string from format: yyyy-mm-dd__hh:mm:ss.d
243 public static String dateToDayOfMonth (String date) {
244 StringBuffer returnDate = new StringBuffer();
246 if (!date.substring(8,9).equalsIgnoreCase("0")) returnDate.append(date.substring(8,10));
247 else returnDate.append(date.substring(9,10));
249 return returnDate.toString();
253 * converts string from format: yyyy-mm-dd__hh:mm:ss.d
256 public static String dateToTime (String date) {
257 StringBuffer returnDate = new StringBuffer();
259 returnDate.append(date.substring(11,16));
261 return returnDate.toString();
265 * Splits the provided CSV text into a list. stolen wholesale from
266 * from Jakarta Turbine StrinUtils.java -mh
268 * @param text The CSV list of values to split apart.
269 * @param separator The separator character.
270 * @return The list of values.
272 public static String[] split(String text, String separator)
274 StringTokenizer st = new StringTokenizer(text, separator);
275 String[] values = new String[st.countTokens()];
277 while (st.hasMoreTokens())
279 values[pos++] = st.nextToken();
285 * Joins the elements of the provided array into a single string
286 * containing a list of CSV elements. Stolen wholesale from Jakarta
287 * Turbine StringUtils.java. -mh
289 * @param list The list of values to join together.
290 * @param separator The separator character.
291 * @return The CSV text.
293 public static String join(String[] list, String separator)
295 StringBuffer csv = new StringBuffer();
296 for (int i = 0; i < list.length; i++)
300 csv.append(separator);
304 return csv.toString();
308 * Wandelet String in byte[] um.
310 * @return byte[] des String
313 public static byte[] stringToBytes(String s) {
314 String crlf = System.getProperty("line.separator");
315 if (!crlf.equals("\n"))
316 s = replace(s, "\n", crlf);
317 // byte[] buf = new byte[s.length()];
318 byte[] buf = s.getBytes();
323 * Ersetzt in String <code>s</code> das <code>pattern</code> durch <code>substitute</code>
327 * @return String mit den Ersetzungen
329 public static String replace(String s, String pattern, String substitute) {
330 int i = 0, pLen = pattern.length(), sLen = substitute.length();
331 StringBuffer buf = new StringBuffer(s.length());
333 int j = s.indexOf(pattern, i);
335 buf.append(s.substring(i));
338 buf.append(s.substring(i, j));
339 buf.append(substitute);
343 return buf.toString();
347 * Ersetzt in String <code>s</code> das Regexp <code>pattern</code> durch <code>substitute</code>
351 * @return String mit den Ersetzungen
353 public static String regexpReplace(String haystack, String pattern, String substitute) {
355 RE regex = new RE(pattern);
356 return regex.substituteAll(haystack,substitute);
357 } catch(REException ex){
366 * F?gt einen Separator an den Pfad an
368 * @return Pfad mit Separator am Ende
370 public static final String addSeparator (String path) {
371 return path.length() == 0 || path.endsWith(File.separator) ? path : path
372 + File.separatorChar;
376 * F?gt ein <code>/</code> ans ende des Strings and
378 * @return Pfad mit <code>/</code> am Ende
380 public static final String addSlash (String path) {
381 return path.length() == 0 || path.endsWith("/") ? path : path + '/';
385 * L?scht <code>/</code> am Ende des Strings, falls vorhanden
387 * @return String ohne <code>/</code> am Ende
389 public static final String removeSlash (String path) {
390 return path.length() > 1 && path.endsWith("/") ? path.substring(0, path.length()
395 * Checks to see if the path is absolute by looking for a leading file
400 public static boolean isAbsolutePath (String path) {
401 return path.startsWith(File.separator);
405 * L?scht Slash am Anfang des Strings
409 public static String removeFirstSlash (String path) {
410 return path.startsWith("/") ? path.substring(1) : path;
414 * formatiert eine Zahl (0-99) zweistellig (z.B. 5 -> 05)
415 * @return zwistellige Zahl
417 public static String pad2 (int number) {
418 return number < 10 ? "0" + number : String.valueOf(number);
422 * formatiert eine Zahl (0-999) dreistellig (z.B. 7 -> 007)
424 * @return 3-stellige Zahl
426 public static String pad3 (int number) {
427 return number < 10 ? "00" + number : number < 100 ? "0" + number : String.valueOf(number);
431 * Konvertiert Unix-Linefeeds in Win-Linefeeds
433 * @return Konvertierter String
435 public static String unixLineFeedsToWin(String s) {
438 i = s.indexOf('\n', i+1);
440 if ((i == 0 || s.charAt(i-1) != '\r') &&
441 (i == s.length()-1 || s.charAt(i+1) != '\r')) {
442 s = s.substring(0, i)+'\r'+s.substring(i);
450 * schnellere Variante der String.toLowerCase()-Routine
452 * @return String in Kleinbuchsten
454 public static String toLowerCase(String s) {
456 char[] a = new char[l];
457 for (int i = 0; i < l; i++)
458 a[i] = Character.toLowerCase(s.charAt(i));
459 return new String(a);
463 * Findet <code>element</code> im String-Array <code>array</code>
466 * @return Fundstelle als int oder -1
468 public static int indexOf(String[] array, String element) {
470 for (int i = 0; i < array.length; i++)
471 if (array[i].equals(element))
477 * Testet auf Vorkommen von <code>element</code> in <code>array</code>
478 * @param array String-Array
480 * @return true wenn <code>element</code> vorkommt, sonst false
482 public static boolean contains(String[] array, String element) {
483 return indexOf(array, element) >= 0;
487 * Ermittelt CRC-Pr?fsumme von String <code>s</code>
489 * @return CRC-Pr?fsumme
491 public static int getCRC(String s) {
493 char val[] = s.toCharArray();
494 int len = val.length;
496 for (int i = 0 ; i < len; i++) {
498 h = (((h >> 30) | (h << 1)) ^ (val[i]+i));
501 return (h << 8) | (len & 0xff);
505 * Liefert Default-Wert def zur?ck, wenn String <code>s</code>
510 * @return geparster int aus s oder def
512 public static int parseInt(String s, int def) {
513 if (s == null) return def;
515 return Integer.parseInt(s);
516 } catch (NumberFormatException e) {
522 * Liefert Defaultwert def zur?ck, wenn s nicht zu einem float geparsed werden kann.
525 * @return geparster float oder def
527 public static float parseFloat(String s, float def) {
528 if (s == null) return def;
530 return new Float(s).floatValue();
531 } catch (NumberFormatException e) {
537 * Findet Ende eines Satzes in String <code>text</code>
540 * @return index des Satzendes, oder -1
542 public static int findEndOfSentence(String text, int startIndex) {
544 int i = text.indexOf('.', startIndex);
545 if (i < 0) return -1;
546 if (i > 0 && !Character.isDigit(text.charAt(i-1)) &&
547 (i+1 >= text.length()
548 || text.charAt(i+1) == ' '
549 || text.charAt(i+1) == '\n'
550 || text.charAt(i+1) == '\t'))
557 * Findet Wortende in String <code>text</code> ab <code>startIndex</code>
560 * @return Index des Wortendes, oder -1
562 public static int findEndOfWord(String text, int startIndex) {
563 int i = text.indexOf(' ', startIndex),
564 j = text.indexOf('\n', startIndex);
565 if (i < 0) i = text.length();
566 if (j < 0) j = text.length();
567 return Math.min(i, j);
572 * convertNewline2P ist eine regex-routine zum umwandeln von 2 oder mehr newlines (\n)
573 * in den html-tag <p>
574 * nur sinnvoll, wenn text nicht im html-format eingegeben
576 public static String convertNewline2P(String haystack) {
577 return re_brbr2p.substituteAll(haystack,"\n</p><p>");
581 * convertNewline2Break ist eine regex-routine zum umwandeln von 1 newline (\n)
582 * in den html-tag <br>
583 * nur sinnvoll, wenn text nicht im html-format eingegeben
585 public static String convertNewline2Break(String haystack) {
586 return re_newline2br.substituteAll(haystack,"$0<br />");
590 * createMailLinks wandelt text im email-adressenformat
591 * in einen klickbaren link um
592 * nur sinnvoll, wenn text nicht im html-format eingegeben
594 public static String createMailLinks(String haystack) {
595 return re_mail.substituteAll(haystack,"<a href=\"mailto:$0\">$0</a>");
600 * createMailLinks wandelt text im email-adressenformat
601 * in einen klickbaren link um
602 * nur sinnvoll, wenn text nicht im html-format eingegeben
604 public static String createMailLinks(String haystack, String imageRoot, String mailImage) {
605 return re_mail.substituteAll(haystack,"<img src=\""+imageRoot+"/"+mailImage+"\" border=\"0\"/> <a href=\"mailto:$0\">$0</a>");
610 * createURLLinks wandelt text im url-format
611 * in einen klickbaren link um
612 * nur sinnvoll, wenn text nicht im html-format eingegeben
614 public static String createURLLinks(String haystack) {
615 return re_url.substituteAll(haystack,"<a href=\"$0\">$0</a>");
619 * this routine takes text in url format and makes
620 * a clickaeble "<href>" link removing any "illegal" html tags
621 * @param haystack, the url
622 * @param title, the href link text
623 * @param imagRoot, the place to find icons
624 * @param extImage, the url of the icon to show next to the link
625 * @return a String containing the url
627 public static String createURLLinks(String haystack, String title, String imageRoot,String extImage) {
629 return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/> <a href=\"$0\">$0</a>");
631 title = removeHTMLTags(title);
632 return re_url.substituteAll(haystack,"<img src=\""+imageRoot+"/"+extImage+"\" border=\"0\"/> <a href=\"$0\">"+title+"</a>");
637 * this routine takes text in url format and makes
638 * a clickaeble "<href>" link removing any "illegal" html tags
639 * @param haystack, the url
640 * @param imageRoot, the place to find icons
641 * @param extImage, the url of the icon to show next to the link
642 * @param intImage, unused
643 * @return a String containing the url
645 public static String createURLLinks(String haystack, String title, String imageRoot,String extImage,String intImage) {
646 return createURLLinks(haystack, title, imageRoot, extImage);
650 * this method deletes all html tags
652 public static final String removeHTMLTags(String haystack){
653 return re_tags.substituteAll(haystack,"");
657 * this method deletes all but the approved tags html tags
658 * it also deletes approved tags which contain malicious-looking attributes and doesn't work at all
660 public static String approveHTMLTags(String haystack){
662 String approvedTags="a|img|h1|h2|h3|h4|h5|h6|br|b|i|strong|p";
663 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";
664 String approvedProtocols="rtsp|http|ftp|https|freenet|mailto";
666 // kill all the bad tags that have attributes
667 String s = "<\\s*/?\\s*(?!(("+approvedTags+")\\s))\\w+\\s[^>]*>";
668 RE regex = new RE(s,RE.REG_ICASE);
669 haystack = regex.substituteAll(haystack,"");
671 // kill all the bad tags that are attributeless
672 regex = new RE("<\\s*/?\\s*(?!(("+approvedTags+")\\s*>))\\w+\\s*>",RE.REG_ICASE);
673 haystack = regex.substituteAll(haystack,"");
675 // kill all the tags which have a javascript attribute like onLoad
676 regex = new RE("<[^>]*("+badAttributes+")[^>]*>",RE.REG_ICASE);
677 haystack = regex.substituteAll(haystack,"");
679 // kill all the tags which include a url to an unacceptable protocol
680 regex = new RE("<\\s*a\\s+[^>]*href=(?!(\'|\")?("+approvedProtocols+"))[^>]*>",RE.REG_ICASE);
681 haystack = regex.substituteAll(haystack,"");
684 } catch(REException ex){
685 ex.printStackTrace();
692 * createHTML ruft alle regex-methoden zum unwandeln eines nicht
693 * htmlcodierten string auf und returnt einen htmlcodierten String
695 public static String createHTML(String content){
696 content=convertNewline2Break(content);
697 content=convertNewline2P(content);
698 content=createMailLinks(content);
699 content=createURLLinks(content);
705 * createHTML ruft alle regex-methoden zum unwandeln eines nicht
706 * htmlcodierten string auf und returnt einen htmlcodierten String
708 public static String createHTML(String content,String producerDocRoot,String mailImage,String extImage,String intImage){
709 content=convertNewline2Break(content);
710 content=convertNewline2P(content);
711 content=createMailLinks(content,producerDocRoot,mailImage);
712 content=createURLLinks(content,null,producerDocRoot,extImage,intImage);