regular expression equivalent to perl's '\b' tag to match word

[Gary Yu]

Hi all,

I'm trying to find a regular expression tag to match a single word, just like '\b' in perl, so that I can match, say, "good" not "goods".

BTW, grep -w seemed not exists on 11.0

thanks,
Gary

[Brian Bergstrand · Accepted solution]

I believe gnu grep can handle PRE's \b. vi and ex can use \ to match a word. But for std. grep you'll have to make up your matching rules.

HTH.

[Gary Yu]

thanks Brian,

I also have the impression that \ should be the one, but I tried it with "sed" and "grep", not good....

[Mark Grant]

As far as I know, the \b matches a word bounday in perl so I though maybe you could use[^a-Z]

I might, however be misinterpreting what you mean here.

[Graham Cameron_1]

Not so easy without perl.
You can look for whitespace...
--
#echo goods|grep -E good[[:space:]]
--
and you can look for end of line...
--
#echo goods|grep -E good$
--
and you can combine ...
--
#echo "1goods\n2good\n3good "|grep -E -e good$ -e good[[:space:]]
2good
3good
--
but it starts to get messy.
-- Graham

[Gary Yu]

like Graham said, here's the messy solution:

egrep '[[:space:]]good[[:space:].]|^good[[:space:].]|[[:space:]]good$' some_file

note the dot inside the [], it stands for period, just in case the word appears at the end of a sentence.

but that still doesn't apply to "sed", since it's extended regular expression, I still can not substitute "good" to "VERY GOOD" with "sed"

thanks,
Gary

[Mark Grant]

Still a touch confused here so excuse me if this doesn't make sense but I think, at least for the grep case, you can do

egrep '[^a-Z]good[^a-Z]' goodfile

I tried this with said but can't get it to work perhaps some better sed maestro can.

[Gary Yu]

Hi Mark,
---
your solution seems doesn't cover the case while the word is at the very begining or very end of a line (^word or word$)
---
and BTW, seemed that the range should be [A-z], not [a-Z](I got such error "egrep: Invalid range within [] expression." )
---
thanks,
Gary

[Mark Grant]

Ok then,

This one really does work :)

cat goodfile | perl -n -e 's/\bgood\b/VERY good/;print;'