[SOLVED] sed - substituting multiple occurrances of whitespace with a ':'

simplified · 12-04-2009, 02:38 PM

Hi All

I'm having a few problems removing multiple occurrance of whitespace with a ':'.

Here's the string:

Code:

     8          contact_name                    jsmith

As you can see, there's a lot of whitespace. What I'd like to do is remove the whitespace and substitute a ':' in it's place, so it looks like this:

Code:

8:contact_name:cralph

So, in short I'd like to:

Remove the whitespace at the front
Substitute the remainding whitespace with a ':'

I've tried:

Code:

cat userlist | sed -e 's/[ ]/:/'

...and:

Code:

cat userlist | sed -e 's/[ ]/:/g'

...but can't quite get it right. Any ideas? This is driving me crazy!

TIA, Simplified

anon237 · 12-04-2009, 02:43 PM

Hi,

Is this what you're looking for:

sed -e 's/[[:space:]][[:space:]]*/:/g' -e 's/^://' infile

Replaces all sorts of whitespaces (and removes the leading one).

Hope this helps.

simplified · 12-04-2009, 04:06 PM

Druuna - you are the man!! Thanks a million

pixellany · 12-04-2009, 05:15 PM

Consider also:
[[:blank:]]
(removes tabs also)

anon237 · 12-04-2009, 05:26 PM

Hi,

For clarity:

[[:blank:]] -> spaces and tabs only (!)
[[:space:]] -> all (!) whitespace chars.

vonbiber · 12-04-2009, 05:38 PM

Quote:

Originally Posted by simplified

Here's the string:

Code:

     8          contact_name                    jsmith

As you can see, there's a lot of whitespace. What I'd like to do is remove the whitespace and substitute a ':' in it's place, so it looks like this:

Code:

8:contact_name:cralph

ok, you want to replace any occurence of one or more space(s) by a ':'
If it's just spaces (no tabs) this should work

Code:

cat userlist | sed 's/ \{1,\}/:/g'

If there are tabs as well as spaces:

Code:

cat userlist | sed 's/[ \t]\{1,\}/:/g'

anon237 · 12-04-2009, 05:42 PM

@vonbiber: You are forgetting the removal of the leading whitespace..... Thread was already solved as well.

pixellany · 12-04-2009, 07:45 PM

Quote:

Originally Posted by druuna

Hi,

For clarity:

[[:blank:]] -> spaces and tabs only (!)
[[:space:]] -> all (!) whitespace chars.

Ooooops!! 3rd mistake this year.....

Why do I think that is counter-intuitive?

Other than a newline, what is whitespace that is not a space or a tab?

anon237 · 12-05-2009, 04:38 AM

@pixellany: Besides the space and (horizontal) tab it also removes vertical tabs and formfeeds.

[ \t\r\n\v\f] == [[:space::]]
[ \t] == [[::blank::]]

I can understand the confusion