I am trying to make a tricky substitution with sed: I have a file 'test.txt' with the following records: address: 192.168.120.1 other: 192.168.120.15 list: 2here: 192.168.120.1;192.168.120.2 files: list: 2here: //192.168.120.1/files I would like to change all the 192.168.120.1 occurences with 129.181.231.23 For this, I use the following command: sed -e 's/192\.186\.120\.1/129\.181\.231\.23/g' test.txt > newtest.txt But the problem is that the second line is then replaced by this one: other: 129.181.231.235 No problem, but if I use the following command: sed -e 's/192\.186\.120\.1[^0-9]/129\.181\.231\.23/g' test.txt > newtest.txt The ';' disapears in the third line and the '/' disapears too in the last line! What I would like to do is checking the character that follows a string without replacing or deleting it. I have no more ideas. Is it possible to do such a thing with only one sed command? If not, any ideas? Thanks !

sed -e 's/\(.*\)\(192\.168\.120\.1\)\([^0-9].*\|$\)/\1129\.181\.231\.23\3/g'

It works great Thanks a lot! Thanks !