regex Pattern class and spaces

[Rajagopal Manohar]

Hi,

I was using the regex package to format data read from a excel spread sheet. I wanted to collapse a set of continous white spaces to a single white space. so I used the Pattern "\\s{2,)" and replaced it with " ".

But I found that it worked only partially. Trying to debug I realised that the some of the data used some "no break space" in unicode with int value of the char being 160, which was missing in the \s pattern.

So I had to do some thing clumsy like
char space = 160;
pattern = "[\\s" + space + "]{2,1}";

Now what I cannot understand is why does the \s pattern class not include this space (char 160). And how do I know that tommorow if I try to read data from another file system in another platform I will not encounter a new space char. Does it not make my code platform dependant (otherwise \s should have handled all possible white spaces)

just a thought, I am sure there is a better explanation

regards,
Rajagopal

[Stefan Wagner]

Ascii(160) isn't allways a kind of space.
On Dos it is � AFAIK.

[Rajagopal Manohar]

Originally posted by Stefan Wagner:
Ascii(160) isn't allways a kind of space.
On Dos it is � AFAIK.

Does that mean that when I see a " " on screen on one platform save it and read it in another platform then I will see a "�". i'snt that a strange behaviour.

does java not promise platform independence? is there no way to guarantee
a common interpretation on all platforms

ps: forgive my ignorance but I thought in java every thing was converted to unicode. apparently i am wrong
[ May 16, 2005: Message edited by: Rajagopal Manohar ]

[Alan Moore]

Yes, Java uses Unicode internally, so ASCII 160 will always be a non-breaking space as far as Java is concerned. To match it, just use the Unicode escape for the character:

  pattern = "[\\s\\u00A0]{2,}";

If you're normalizing the whitespace, shouldn't you also be converting single linefeeds, tabs, NBSP's, etc. into space characters?

  pattern = "[\\s\\u00A0]{2,}|[[\\s\\u00A0]&&[^ ]]";

That is, any two or more consecutive whitespace characters, or any single whitespace character that isn't a space (ASCII 32).

[Rajagopal Manohar]

If you're normalizing the whitespace, shouldn't you also be converting single linefeeds, tabs, NBSP's, etc. into space characters?

I guess yes.
Thanks
Rajagopal