egrep search for whitespace

Question

I'm trying to use egrep with a regex pattern to match whitespace.

I've used RegEx with Perl and C# before and they both support the pattern \s to search for whitespace. egrep (or at least the version I'm using) does not seem to support this pattern.

In a few articles online I've come across a shorthand [[:space:]], but this does not seem to work. Any help is appreciated.

Using: SunOS 5.10

Jon 'links in bio' Ericson · Accepted Answer · 2009-01-16T17:33:10.420

25

I see the same issue on SunOS 5.10. /usr/bin/egrep does not support extended regular expressions.

Try using /usr/xpg4/bin/egrep:

$ echo 'this line has whitespace
thislinedoesnthave' | /usr/xpg4/bin/egrep '[[:space:]]'
this line has whitespace

Another option might be to just use perl:

$ echo 'this line has whitespace
thislinedoesnthave' | perl -ne 'chomp;print "$_\n" if /[[:space:]]/'
this line has whitespace

edited Jan 16 '09 at 17:33

answered Jan 16 '09 at 00:30

Jon 'links in bio' Ericson

20,880
12
98
148

I don't understand why you say you have the same issue, look like it works with egrep .. ? – Aif Jan 16 '09 at 10:15
The default egrep does not support advanced character sets like [[:space:]]. You need to either change your PATH or call out the absolute path as I did above. – Jon 'links in bio' Ericson Jan 16 '09 at 17:08
Which version of egrep was it? – mkierc Dec 20 '15 at 13:41

score 14 · Answer 2 · edited Jan 16 '09 at 00:55

14

If you're using 'degraded' versions of grep (I quote the term because most UNIX'es I work on still use the original REs, not those fancy ones with "\s" or "[[:space:]]" :-), you can just revert to the lowest form of RE.

For example, if :space: is defined as spaces and tabs, just use:

egrep '[ ^I]' file

That ^I is an actual tab character, not the two characters ^ and I.

This is assuming :space: is defined as tabs and spaces, otherwise adjust the choices within the [] characters.

The advantage of using degraded REs is that they should work on all platforms (at least for ASCII; Unicode or non-English languages may have different rules but I rarely find a need).

edited Jan 16 '09 at 00:55

Jonathan Leffler

730,956
141
904
1,278

answered Jan 16 '09 at 00:52

paxdiablo

854,327
234
1,573
1,953

A nice solution. I like the idea of working to the lowest common denominator – thecoshman Jun 13 '12 at 10:18
[GNU grep 2.5.4 on CMD](http://gnuwin32.sourceforge.net/packages/grep.htm) is happy with `[ \t]+` to match one or more whitespace chars. – handle Aug 09 '16 at 13:09

score 3 · Answer 3 · answered Sep 10 '12 at 03:36

If you are using bash, then syntax to put a tab in a line is

$'foo\tbar'

I was recently working with sed to do some fixups on a tab-delimited file. Part of the file was:

sed -E -e $'s/\t--QUOTE--/\t"/g'

That argument is parsed by bash, and sed sees a regex with literal tabs.

score 0 · Answer 4 · answered Jan 15 '09 at 23:51

0

Maybe you should protect the pattern with quotes (if bash, or anything equivalent for the shell you are using).

[ and ] may have special meaning for the shell.

answered Jan 15 '09 at 23:51

Giacomo

11,087
5
25
25

Aif · Answer 5 · 2009-01-17T11:48:11.460

-3

$ cat > file
this line has whitespace
thislinedoesnthave
$ egrep [[:space:]] file 
this line has whitespace

Works under debian.

For Solaris, isn't there an "eselect" like (see gentoo) or alternatives file to set default your egrep version?

Have you tried grep -E, because if the egrep that is on your path is not the good one, maybe grep is.

edited Jan 17 '09 at 11:48

answered Jan 15 '09 at 23:47

Aif

11,015
1
30
44

You might get some credit if you explained where 'here' was. It presumably wasn't Solaris 10. Or, if it was Solaris 10, then it probably wasn't /usr/bin/egrep that you used. – Jonathan Leffler Jan 16 '09 at 00:57

egrep search for whitespace

5 Answers5