Multiple substitution 5

[dickiebird]

I've wrestled with this for long enough !
I have a text file with variable number of 0 (zero) on the beginning of most lines. I need to replace these leading zeros with same no. of spaces.
So far, I've got this :
sed 's/^0+/ /g' infile > outfile

which nearly does it ( It replaces 00000 or 000 etc with one space only.)
How would I replace all leading strings of zeros with a string of same number of spaces ?
TIA
;-)
Dickie Bird
db@dickiebird.freeserve.co.uk

 

[bigoldbulldog]

Two ways:

sed -e :a -e 's/^\(0*\)0/\1 /;ta' infile > outfile

or

awk 'BEGIN{
ZEROS="^0+"
}
{
if( match( $0, ZEROS )){
sub( ZERO, "", $0 )
for( i = 0; i < RLENGTH; i++ )
printf( &quot; &quot; )
}
print
}' infile > outfile

The awk choice could be re-written many ways - maybe someone can find a good one-liner on it.
Cheers,
ND

bigoldbulldog@hotmail.com

 

[marsd]

Why not use awk?
awk ' {
if ($0 ~ /^0/) {
gsub(/0/,&quot; &quot;,$0)
}
print
} infile > outfile

 

[marsd]

OOps...sorry bout that,that kills all zeroes..
Big old bulldogs looks like it works..

 

[dickiebird]

Cheers BigOldBulldog - the sed version is just perfect !
I'll now spend an hour or two looking through sed&awk
to figure out why it works !
TAL ;-)
Dickie Bird
db@dickiebird.freeserve.co.uk

 

[jamisar]

bigoldbulldog BRILLANT a star vox clamantis in deserto.

 

[toolkit]

Or how about the perl one-liner:

perl -p -e &quot;s/^(0+)/' ' x length(\$1)/e&quot; input > output

Cheers, Neil

 

[dickiebird]

Perl would be nice - except this particular server is on AIX 3.2.5 - pre-perl !
;-) Dickie Bird
db@dickiebird.freeserve.co.uk

 

[jamisar]

dickiebird:
sed -e :a -e 's/^\(0*\)0/\1-/;ta'
he recursively replace
(000)0 to 000-
(00)0- to 00--
... and so on

:a mark a label
^\(0*\) look for string of zeros at line begin
0 followed by a zero
\1 print the founded string
- followed by a - (in the example was a space)
; concatenate commands (no spaces!!)
ta if substitution was made restart at label, on the same input line

simply performant sed, once again: compliment bigoldbugdog !

toolkit:
i cannot get your version working,
my perl5 complains about \$1
if i un-mask it, he miscompute
the length($1) == length(whole-line)
what's wrong ?
PS: i hope you agree, bigoldbugdog's one-liner is more elegant.

marsd:
how can you propose awk, after reading bigoldbugdog's one-liner ?
vox clamantis in deserto.

 

[iribach]

F O R M I D A B L E bigoldbulldog !!
jamais je n'oublirai ce 'statement', merci, mon etoile.

 

[dickiebird]

Thanks for the explanation, jamisar.
Perhaps marsd was formulating his solution at the same time as BigOldBulldog, and hadn't seen the one-line sed solution ?
Anyway - I've been spoilt today by everyone - Thanks again all!
;-)
Dickie Bird
db@dickiebird.freeserve.co.uk

 

[jamisar]

sure dickiebird you are rigth,
sorry marsd vox clamantis in deserto.

 

[marsd]

Didn't see it till too late...I've been hiding ever
since
That is pretty to look at isn't it.

 

[toolkit]

jamisar,
the \$1 escapes it for the shell. sed is for evil geniuses only ;-)

 

[toolkit]

from the shell, not for the shell (-: And bigoldbulldog's solutions is by far the most elegant I