Regex to replace special characters

[soumya ravindranath]

Hi,

I need to replace special characters when they occur in certain places. The following method works well except for the first occurence of the pattern. I am trying hard to debug, but am getting nowhere. Could anyone please see what is going wrong here ?

The special characters should be replaced only when they occur before the '|' sign. But in the result as you see, in the first occurence, the replacement is made to both sides of '|' and in the rest of the text it is done correctly.

My search text sample -

[first name�|firstname�] this is my page, please go through it and send me feedback. [cchrp|Chris Harp] my page is open to all
[G�nterL�her|G�nter L�her] Replace the sp.characters here please!

Current result after replacement - ( I am trying to escape HTML for correct display here, hence the extra whitespaces )
[first name&s zlig;|firstname&s zlig;] this is my page, please go through it and send me feedback. [cchrp|Chris Harp] my page is open to all [G&u uml ;nterL&o uml ;her|G�nter L�her] Replace the sp.characters here please!

public static void testRegex(CharSequence text) {

Pattern pat = Pattern.compile("(�|�|�|�|�|�|�)(?=.*?\\|)");
Matcher mat = pat.matcher(text);
System.out.println("text :" + text);
StringBuffer buf = new StringBuffer();

while (mat.find()) {        
String groupStr = mat.group();            
            switch(groupStr.charAt(0)) {
            
            case '�' : mat.appendReplacement(buf, "ü");               
               break;
            case '�' : mat.appendReplacement(buf, "Ü");               
           break;
            case '�' : mat.appendReplacement(buf, "ä");               
                    break;
            case '�' : mat.appendReplacement(buf, "Ä");               
                    break;
            case '�' : mat.appendReplacement(buf, "ö");               
                    break;
            case '�' : mat.appendReplacement(buf, "Ö");               
                    break;
            case '�' : mat.appendReplacement(buf, "ß");               
           break;
        //default  : break;            
            }                                    
    }
mat.appendTail(buf);
System.out.println("buffer : " + buf.toString());
}
}

thank you,
Soumya
[ April 27, 2006: Message edited by: soumya ravindranath ]

[Henry Wong]

The special characters should be replaced only when they occur before the '|' sign. But in the result as you see, in the first occurence, the replacement is made to both sides of '|' and in the rest of the text it is done correctly.

I believe this problem occurs when there are multiple "|" per line. The reason it is matching is because while it is after the "|" locally, it is still *before* the next one.

Henry

[soumya ravindranath]

Hi,

that makes perfect sense! I am trying to get this greedy matching to change to non-greedy ( though I expected .*? to make a non-greedy match in the first place ), but to no avail. Any suggestion how I can improve this regex to behave the way I require ?
thanks,
Soumya.

[Alan Moore]

Try this:

Pattern pat = Pattern.compile("([�������])(?=[^|\\]]*\\|)");

BTW, regex questions really belong in the Java in General: Intermediate forum.

[soumya ravindranath]

Thank you very much, that works for my example.
Yes, I was about to post it in Java forum, but then seeing another regex question here, changed my mind...

Further, I don't still quite get the regex writing well. Could you please explain what exactly the second group in the pattern does ?

thanks in advance,
soumya
[ April 28, 2006: Message edited by: soumya ravindranath ]

[Alan Moore]

Sure. If you aren't familiar with negated character classes, read this. The character class ensures that the lookahead doesn't go beyond the end of the current set of brackets to look for a pipe. Come to think of it, you don't even need to put the pipe in the character class. But if there are any special characters that aren't inside brackets, and you don't want to replace them, you should include the left bracket in the character class as well:

Pattern pat = Pattern.compile("([�������])(?=[^\\[\\]]*\\|)");

[soumya ravindranath]

thanks Alan!