How can I remove dummy information from a file named results.txt with lines like this?
Lines inside file are like this:
_my0001_split00000000.txt:Total Dynamic Power = 0.0000 mW
_my0001_split00000050.txt:Total Dynamic Power = 117.5261 uW (100%)
... and they should change to a tab-separated format like this:
0001 00000000 0.0000 mW
0001 00000050 117.5261 uW
How about a using sed instead of awk?
sed -r 's/^_my([0-9]+)_split([0-9]+)\.txt:[^=]*=\s*([0-9.]+) *(\S+).*/\1\t\2\t\3 \4/' /path/to/file
If you have GNU awk, then you can indeed specify fixed fieldwidths e.g.
gawk -vFIELDWIDTHS="3 4 6 8 30 8 1 3" -vOFS="\t" '{print $2,$4,$6,$8}' results.txt
0001 00000000 0.0000 mW
0001 00000050 117.5261 uW
However you could also consider a regular expression based solution - for example, in perl you could print the first three numeric values plus power units as:
perl -lne 'print join "\t", (/\d*\.?\d+|\b.W\b/g)[0..3]' results.txt
0001 00000000 0.0000 mW
0001 00000050 117.5261 uW
You can do it in awk. For example:
$ awk -vOFS="\t" '{
sub(/.txt:.*= */," ");
sub(/^.../,"",$1);
sub(/_split/, "\t",$1);
print $1,$2,$3
}' file
0001 00000000 0.0000 mW
0001 00000050 117.5261 uW
It really isn't the best tool for the job though. Awk excels when you want to extract fields unchanged, it can get cumbersome when you need to edit them.
I would instead use something like:
$ perl -lne '
@m=(/\D+(\d+)\D+(\d+).*=\s*([0-9.]+)\s+(\S+).*$/);
print join "\t", @m' file
0001 00000000 0.0000 mW
0001 00000050 117.5261 uW
sed 's/_my//;s/_split/\i/;s/\.txt:Total Dynamic Power *= */\i/' /path/to/input
