Single white space replacement whilst ignoring multiple spaces

[eric lipede_1]

Hi
does anyone know how to replace a single white space and leave and multiple spaces alone....
so, for example:

echo "12 3 4 5 6"

should produce

12 34 56

(where the only single spaces were between 3 &4 and 5& )

awk/sed or perl is fine thks - preferable in one line - thks

[Steven Schweda]

Perhaps:

$ echo '12 3 4 5 6' | \
sed -e 's/\([^ ]\) \([^ ]\)/\1\2/g'
12 34 56

(Replace non-space1+space+non-space2 with
non-space1+non-space2 everywhere.)

[James R. Ferguson]

Hi Eric:

A bit late, but in Perl:

echo "12 3 4 5 6"|perl -ple 's/(\S)\s(\S)/$1$2/g'
12 34 56

The '\s' is a whitespace (space, tab) character whereas the '\S' is a non-whitespace character.

Regards!

...JRF...

[Hein van den Heuvel · Accepted solution]

IMHO this would be an excellent example for using the \b 'boundary' matching in perl's regulare expression:

# echo "12 3 4 5 6"|perl -ple 's/\b\s\b//g'
12 345 6

(12ss3s4s5ss6)

Please observe how this replaced ALL single-white-space instances.

JRF's solution, would replace the single space between 3 and 4, but would NOT replace the single space between 4 and 5 for the example I used, as the match engine has already dealt with '4'.

You have to decide what is 'right' for your usage. I happen to think the using \b is more true to the original problem statement.

fwiw,
Hein


# echo "12 3 4 5 6"|perl -ple 's/(\S)\s(\S)/$1$2/g'
12 34 5 6

[eric lipede_1]

Thks to all of you.

Hein, I think your solution was the most robust

[Dennis Handly]

>Hein: JRF's solution, would replace the single space between 3 and 4, but would NOT replace the single space between 4 and 5 for the example I used

It seems Steven's fails the same way. But one solution is simple, do it again:
echo '12 3 4 4 5 6' | sed -e 's/\([^ ]\) \([^ ]\)/\1\2/g' -e 's/\([^ ]\) \([^ ]\)/\1\2/g'

[eric lipede_1]

I re-opened this thread as I think that this is an interesting problem ...and of course to assign points to great answers....

[James R. Ferguson]

Hi (again) Eriic:

> I re-opened this thread as I think that this is an interesting problem ...

From a Perl perspective, I think Hein provided the most general, most direct solution. It works for the original cases as well as Dennis's sample data.

Dennis does offer an interesting solution, though. My original offering of:

# perl -ple 's/(\S)\s(\S)/$1$2/g'

...could be modified to perform the repetitive substitutions of Dennis's 'sed' by writing:

# echo "12 3 4 4 5 6"|perl -ple '1 while s/(\S)\s(\S)/$1$2/g'
12 344 56

Regards!

...JRF...

[eric lipede_1]

Hi

Ive tested it with more variations of the sample data...and files . Both remove white spaces and ignore multiple instances of them -thks!