Moving columns with sed or awk?

btm · 09-27-2007, 10:11 AM

I have a text file that I have parsed with commas using sed. There are also spaces between the columns (making them fixed width) which I strip out in the end to make it a true csv file for exportation to a database.

Question:
I need to rearrange some of the columns. I.E. column 4 needs to be after column 1, etc.

ChrisScott · 09-27-2007, 10:41 AM

You could use SED for that (lots of alternatives as well, I expect!):

Code:

sed 's@\(.*\),\(.*\),\(.*\)@\2,\1,\3@g' file

ie,

1,4,7
2,3,4

becomes,

4,1,7
3,2,4

ChrisScott · 09-27-2007, 11:14 AM

You could use a sed script to reajust your columns to:

Code:

cat <<EOF>sedscript
s@\(.*\),\(.*\),\(.*\)@,\2,\1,\3@g
s@ *,@,@g
: width
s@,\([^,]\{0,10\}\),@,\1 ,@g
t width
s@^,@@g
EOF
sed -f sedscript file

the 10 in - s@,$[^,]\{0,10\}$,@,\1 ,@g - is the width the column will become.

pwc101 · 09-27-2007, 11:50 AM

You can use awk:

Code:

awk 'BEGIN {OFS = ","}; {print $1,$4,$2,$3}' your.file > your_output.file

The {OFS = ","} sets the output field separator as a comma, allowing for your csv file creation. The $1,$4 bit reorders to columns according to their numerical value.

btm · 09-27-2007, 03:03 PM

Thanks, all. ALL of your solutions work perfectly for me.