cut: can we set multiple spaces as the delimiter?

Question

I have text like this:

word1 word2   word3  word4

There may be more than one space between a pair of words and I want to get some columns of words from each line . When I use cat file | cut -d ' ' -f1,2,4 it seems that some fields are space which is not what I expected. I know awk can achieve this. The question is can we do this with cut only? i.e., can we set multiple spaces as the delimiter in cut, yet the number varies?

Answer 1

No, you cannot. If you want to be able to use more than one character (or even a regex) for the delimiter then use awk instead.

Answer 2

As others have stated, cut can't do it alone (and awk is the best choice, because it's the only tool required). If you still want to use cut, you can combine it with tr, however:

tr -s ' ' <<<"word1 word2   word3  word4" | cut -d ' ' -f1,2,4

tr -s ' ' folds each span of multiple spaces into one space each.

Answer 3

use awk, bro: awk '{print $1, $2, $4;}' file

Thank you @fedorqui for the sugesstion

Answer 4

Also, you can use the positional parameters

line="word1 word2   word3  word4"
set -- $line                       # no quotes here!
echo $1 $2 $4

word1 word2 word4

Answer 5

Assuming you have a reasonable IFS (e.g. IFS=$' \n\t' ) Use word-splitting before passing to cut

$ var="word1 word2   word3  word4"; echo $var
word1 word2 word3 word4
$ var="word1 word2   word3  word4"; echo $var | cut -d ' ' -f1,2,4
word1 word2 word4

So for you

$ var=$(cat "file"); echo $var | cut -d ' ' -f1,2,4

Answer 6

No, I'm sorry, delimiter in cut is always only one character. But you can use read instead

while read col1 col2 col3 col4
do
    echo "$col1 $col2 $col4"
done < file

Answer 7

Since BASH's default Internal Field Separator is whitespace, you can avoid using cut or any other external tools by using the BASH read builtin command.

while read f1 f2 _ f4; do echo "$f1 $f2 $f4"; done < file

Here, the _ simply acts as a junk variable for the third field.