[SOLVED] awk

dazdaz · 02-21-2013, 06:13 AM

How do I print from field 3 to the end of the line, when there may be more than 5 fields.

Therefore printing $3, $4, $5 is not a realistic method, because there might be more than 5 fields.

This is as far as I can get..

Code:

echo 66 77 88 99 | awk '{print $3}'

druuna · 02-21-2013, 06:23 AM

Have a look at this:

Code:

 echo 1 2 3 4 5 | awk '{ for (i=3; i<=NF; i++) print $i }'

NF is special in awk, it holds the number of fields for the line that is worked on.

dazdaz · 02-21-2013, 06:28 AM

Thanks, that's very useful to know.

How can I print the output without awk adding a new CR/LF each time.

druuna · 02-21-2013, 06:31 AM

Here is one way;

Code:

echo 1 2 3 4 5 | awk '{ for (i=3; i<=NF; i++) printf("%s ",$i) }END{ print"" }'

The END{ print"" } part is needed to move the cursor to the next line (try running it without that part).

danielbmartin · 02-21-2013, 07:10 AM

Quote:

Originally Posted by dazdaz

How do I print from field 3 to the end of the line, when there may be more than 5 fields.

I like to use cut for this purpose.

InFile ...

Code:

one two three four five 
one two three four five six seven 
one two three four
one two three four five six seven eight
one two three four five six

This code ...

Code:

cut -d" " -f3- $InFile > $OutFile

... produces this result ...

Code:

three four five 
three four five six seven 
three four
three four five six seven eight
three four five six

Daniel B. Martin

danielbmartin · 02-21-2013, 10:56 AM

This is a revision of the awk posted by druuna.

InFile ...

Code:

one two three four five 
one two three four five six seven 
one two three four
one two three four five six seven eight
one two three four five six

This code ...

Code:

awk '{for (i=3;i<=NF;i++) printf("%s ",$i)} {print ""}' $InFile > $OutFile

... produced this result ...

Code:

three four five 
three four five six seven 
three four 
three four five six seven eight 
three four five six

Daniel B. Martin

David the H. · 02-24-2013, 04:14 PM

I second daniel's cut suggestion, with the caveat that it can only be used for simple column extractions. You'll still need to use awk for any fancier line filtering, or if the delimiters are mixed/uneven.

(As an aside, I've often said to myself that awk really needs to add a few functions for manipulating ranges of fields and array elements. It would make this kind of thing so much easier.)

For another option, how about using a sed regex to remove the columns from the front that you don't want?

Code:

sed -r 's/^([^[:space:]]+[[:space:]]+){2}//' infile.txt

danielbmartin · 02-24-2013, 09:11 PM

InFile ...

Code:

one two three four five 
one two three four five six seven 
one two three four
one two three four five six seven eight
one two three four five six

This code ...

Code:

awk '{$1=$2="";sub(/  /,"")}1' $InFile >$OutFile

... produced this result ...

Code:

three four five 
three four five six seven 
three four 
three four five six seven eight 
three four five six

As a matter of personal style I like to write code which does not use an explicit loop. Technical intuition suggests that "loopless" code runs faster but I have no evidence to support this hypothesis.

Daniel B. Martin

grail · 02-24-2013, 09:49 PM

Another alternative:

Code:

ruby -ane 'puts $F[2..-1].join(" ")' file