How to define 'tab' delimiter with 'cut' in Bash?

Question

Here is an example of using cut to break input into fields using a space delimiter, and obtaining the second field:

cut -f2 -d' '

How can the delimiter be defined as a tab, instead of a space?

Using cut with unprintable delimiters (stackoverflow) – Brent Bradburn Nov 21 '14 at 20:14 — Brent Bradburn, Nov 21 '14 at 20:14
Tab is a default delimiter. cut -f 2 – Melah Apr 19 '17 at 19:47 — Melah, Apr 19 '17 at 19:47

score 511 · Accepted Answer · edited Oct 17 '21 at 15:47

511

Two ways:

Press Ctrl+V and then Tab to use "verbatim" quoted insert.

cut -f2 -d'   ' infile

or write it like this to use ANSI-C quoting:

cut -f2 -d$'\t' infile

The $'...' form of quotes isn't part of the POSIX shell language (not yet), but works at least in ksh, mksh, zsh and Busybox in addition to Bash.

edited Oct 17 '21 at 15:47

ilkkachu

138,973

answered Mar 30 '12 at 07:42

Birei

8,124

12

Neither works in Windows (e.g., Git Bash) obviously. Use @Mikel's answer there (default delimiter is tab, simply omit -d argument). – Ahmed Fasih Nov 04 '14 at 14:40
1

@AhmedFasih, both work in Cygwin, though (tested) – Walter Tross Jun 17 '15 at 17:16
7

What does it mean to have a $“dollarsign-prefixed string” in a script? – micrub Jul 30 '16 at 23:11
4

@poige, that comes from ksh93, not bash. That's supported by ksh93, zsh, bash, mksh and FreeBSD sh at least (it might make it to the next major reversion of the POSIX standard specification for sh). – Stéphane Chazelas Oct 23 '17 at 11:42
4

@micurb That's ANSI-C quoting: https://www.gnu.org/software/bash/manual/html_node/ANSI_002dC-Quoting.html – wisbucky Dec 08 '17 at 21:51
you can also use Ctrl-q instead of Ctrl-v – elig Sep 11 '18 at 00:57
1

@AhmedFasih the 2nd option (-d $'\t' ) works for me in git bash on windows. – Tomas Svitil Feb 26 '20 at 07:20
Does not work in zsh

qmeeus

Mar 08 '22 at 13:15

This should not be the accepted answer, see: https://unix.stackexchange.com/a/35387/391735 – JBoy Oct 27 '22 at 12:45

Mikel · Answer 2 · 2023-08-15T14:40:06.783

316

Tab is the default.

See the cut man page.

-d delim
         Use delim as the field delimiter character instead of the tab
         character.

So you can just write

cut -f 2

edited Aug 15 '23 at 14:40

answered Mar 30 '12 at 13:56

Mikel

57,299
15
134
153

4

But it is probably always safer to mention such flags explicitly, for both readability and portability. I can imagine some people designing a cut for Windows would not follow the complete standard. – willeM_ Van Onsem Apr 19 '15 at 12:32
5

This should be accepted answer: simplest code, simplest explanation. ya nailed it: https://www.youtube.com/watch?v=dpNTHl7y45Y – neuronet Jun 16 '16 at 21:28
5

@WillemVanOnsem, if someone writes a version of cut for Windows and doesn't follow the POSIX specification for it, there is no reason to assume that any POSIX script will work with that system. Stick to POSIX-specified features. Don't try to allow for hypothetical future non-compliant implementations; that's not what "portability" means. – Wildcard Mar 26 '19 at 20:37
@WillemVanOnsem Additionally, cut -f 2 is way more readable than cut -d' ' -f 2 or cut -d$'\t' -f 2 – Mouradif Jun 21 '19 at 10:10

score 19 · Answer 3 · edited May 21 '19 at 20:08

19

awk -F '\t' '{ print $2 }' inputfile

This extracts the second tab-delimited field of each line of input from inputfile.

edited May 21 '19 at 20:08

Kusalananda

333,661

answered Mar 30 '12 at 11:29

Iľja

429
2
3

2

The OP was specifically asking about cut. Moreover, cut(1) is waaay faster than awk(1). – Tripp Kinetics May 14 '20 at 19:10

score 10 · Answer 4 · answered Nov 21 '14 at 20:34

10

More generically, without requiring any invisible characters: Use tr to convert the delimiters to a format that can be specified more easily to cut.

$ echo -e "a\tb\tc" |tr '\t' ' ' |cut -d' ' -f2
b

tr is a simple, but powerful, character matching and replacement tool.

answered Nov 21 '14 at 20:34

Brent Bradburn

1,274

2

But what if the input is abc(space)def(tab)ghi? Your answer will yield def, but it should yield ghi. Similarly, if the input is ABC(tab)DEF(space)GHI, your answer will yield DEF, but it should yield DEF(space)GHI. – G-Man Says 'Reinstate Monica' Mar 26 '19 at 20:26
1

@G-Man: The space delimiter was only an example. Use whatever delimiter is appropriate for your data -- a comma for example. echo -e "abc\tdef ghi" |tr '\t' ',' |cut -d',' -f2 – Brent Bradburn Mar 26 '19 at 22:29
1

...But, yeah, if the delimiter must be a tab, then my approach won't work. – Brent Bradburn Mar 27 '19 at 00:04

Anthony Rutledge · Answer 5 · 2021-10-17T19:35:54.673

0

Alternatively, one could wrap cut in a function.

function getColumns ()
{
    local -r delimiter="${1:?}"
    local -r columns="${2:?}"
if [[ &quot;$delimiter&quot; == '\t' || &quot;$delimter&quot; == &quot;tab&quot; ]]; then
    cut &quot;--fields=${columns}&quot;
    return
fi

cut &quot;--delimiter=${delimiter}&quot; &quot;--fields=${columns}&quot; 

}

edited Oct 17 '21 at 19:35

answered Oct 17 '21 at 15:30

Anthony Rutledge

207

score -2 · Answer 6 · answered Oct 17 '21 at 19:59

-2

I use TAB and cut in these ways:

# quote the whole thing, use TAB escape
cut "-d\t" -f 2
# just quote the tab escape 
cut -d "\t" -f 2
# Use Ctrl-V to quote Ctrl-I (TAB is Ctrl-I, see man ascii)
cut -d^V^I -f 2

answered Oct 17 '21 at 19:59

waltinator

4,865

The cut utility does not understand the \t as tab. – Kusalananda Aug 15 '23 at 15:26
@Kusalananda Show the failure. By quoting , the "\t" is interpreted by the shell, which does understand backslash-escapes, and $(...), and ... What's your $SHELL or getent passwd $(id -u) | cut "-d:" -f6. Also cat --show-all datafile | head -n 10. Please [edit] your question to add whatever information you get. Do not use Add Comment. – waltinator Sep 02 '23 at 23:15
Just try it yourself in any POSIX shell: printf '%s\t%s\t%s\n' 1 2 3 | cut -d "\t" -f 2 (will not extract the 2 in the 2nd column). Just use cut without the -d option (tab is the default delimiter). – Kusalananda Sep 03 '23 at 07:56
In fact coreutils cut fails when given -d "\t" or "-d\t": “cut: the delimiter must be a single character”. – Stephen Kitt Sep 06 '23 at 10:29

How to define 'tab' delimiter with 'cut' in Bash?

6 Answers6

Linked

Related