Regular Expression Problem

[daflores]

Hello all,

A small example of the problem I have:

String:
test   another test

In this string, I want to replace   (say by empty string) but ONLY when there is just a single occurrence. For example, this is the desired output for teh previous string:

Desired Ouput:
test   anothertest

Is there a regular expression that could be used in the replace method that could accomplish this? I have tried with several expressions with no success but I am no expert in regular expressions.

If this can not be done usign a regular expression, what would be the most efficient way of accomplishing this? Using the indexOf method?

Thanks in advance for your help.

David

[masijade]

"(?<!;)&nbsp;(?!&)"

[daflores]

Thanks for your reply masijade. I tried the expression but it also replaces the first occurrence of   even when there are several consecutive occurrences of it. For the string I gave as an example, the output using your regular expression is:

test;  another;test

Also, the expression replaces &nbsp but leaves the semicolon (;) in the output (as in the above output).

[wolfcro]

i get the correct output with that expression

test   anothertest

with this code:

String s = "test   another test";
		
String s2 = s.replaceAll("(?<!;)&nbsp;(?!&)", "");
System.out.println(s2);

[Steve11235]

You could also try

String string = "test   another test";
string = string.replaceAll("((?<!&nbsp;)(&nbsp;)(?!&nbsp;))", "");

This does a negative look behind and a negative look ahead for " ", which seems to be what you are wanting to do.

Looks like Wolfcro posted just before I did!

[daflores]

Thank you all for your replies. I initially tried masijade on a replace applet on the web and it did not work but when I saw wolfcro's reply I tried it again now directly in the code and it worked beautifully. Not sure why teh applet gave different results. Thanks for your reply too Steve, I'll give it a try too. Now if you could only shed some light on why it works I would appreciate it (does not need to be long). As I said I am no expert in regular expressions.

Thanks,
David

[Steve11235]

Wolfcro and I did the same thing. He's been beating me to answering questions all morning.

Here's a link to a reference site. Look at the Lookaround section.

Basically, look after, placed after some other construct, tells the engine that, after it finds the construct, to look immediately after the construct to see if specified text is found. Look ahead works in a similar manner, and both have negative (not found) versions.

In your case, we looked ahead and after " " to make sure that the was no " " just ahead or just after.

[daflores]

Thx a lot guys!

[masijade]

the expression replaces &nbsp but leaves the semicolon (;) in the output (as in the above output).

Then you used it before I modified it. The first time I had forgotten the semicolon (there was only about a minute were you could have loaded the wrong one, but seemingly you managed to fit into that window). That would also be the reason for it having removed the first one. If you look at the expression again, you will see a subtle difference (the semicolon after nbsp). Try it that way.

Edit: Also, that expression, as it is, will simply remove any   without a ; directly before it and/or without a & directly behind it. A single whitespace between those html codes will allow the thing to be removed, and any html code before or after   will prevent   from being removed. If you don't mind removing any additional whitespace around the   then I would do it this way, to be sure

"(?<!&nbsp;)\\s*&nbsp;\\s*(?!&nbsp;)"