Pipe symbol | in AWK field delimiter

Question

I have a file foo that has the following data:

A<|>B<|>C<|>D
1<|>2<|>3<|>4

I want to prope

Answer 1

Awk reads your separator as a regex, "< or >". You have to escape the pipe character (twice, seeing that dynamic regexps such as the field separator are scanned twice): "<\\|>".

You can specify the field separator also as a parameter:

awk -F '<\\|>' '{out=""; for(i=1;i<=NF;i++){out=out" "$i}; print out}' <<< 'A<|>B<|>C<|>D'
 A B C D

Depending on your version of awk, you might get away with just single escaping. For me, mawk 1.3.3 works with both -F '<\|>' and -F '<\\|>', and gawk 4.0.1 requires -F '<\\|>'. I'm not fully sure which way POSIX awk goes, but running gawk in --posix mode requires the double escapes, too.

Answer 2

The pipe is a special character in a regex, so you need to escape it with a backslash. But this backslash is also a special character for the string literal, so it needs to be escaped again. So you end up with the following:

awk -F '<\\|>' '{$1=$1}1'

awk 'BEGIN {FS="<\\|>"} {$1=$1}1' 

The reason for this syntax is explained quite well here: http://www.gnu.org/software/gawk/manual/gawk.html#Computed-Regexps. In short, the expression is parsed twice.

Answer 3

Btw, also sed can be used here:

sed 's/<|>/ /g' file

If you want to "include" the head command:

sed -n '1,10s/<|>/ /gp' file