I need to search and replace a particular string in a file. Only the exact match of the string should be replaced.
eg: File contents : abc abcd abcdef
--> Replace only 'abc' with 'xyz', but it should not replace abcd with xyzd.

So the o/p should be: xyz abcd abcdef.

How can this be done? I tried sed 's/\<abc\>/xyz/g' , but it dint work.

Any help would be appreciated.

Thanks in advance!

echo "abc abcd abcdef" | sed -e 's/^abc/xyz/g'

I tried it. It dint work. Let me be more specific. The string need not be at the beginning of the line:

i/p: abcd abc abcdef
o/p: abcd xyz abcdef

Thanks!

echo "string" | sed -e 's/ abc / xyz /g' | sed -e 's/^abc /xyz /g' -e 's/ abc$/ xyz/g'

Your first example should work using the <> to define a "word" but only with GNU sed or GNU awk and similar enhanced SEDs and AWKs. There are SEDs and AWKs out there that have a not as broad understanding of regular expressions like the modern GNU versions.
On most Linux boxes your line should work flawless.

Afaik you will have to emulate the <word> RegExp behaviour like reddybs posted. Only thing I see is it could be written so sed is called only once, but I would go with that too, example:

$> echo "abc abcd abcde abc abcd xabc dabc"| sed '
> s/ abc / ooo /g
> s/^abc /ooo /g
> s/ abc$/ ooo/g
> '
ooo abcd abcde ooo abcd xabc dabc

Thanks zaxxon and reddybs. That worked!