Oliver Jones wrote:
To continue the trend set by James Clark... Here is my choice for useful command-line tool of the day.
fold
Ooh, my favourite is "join(1)"[1] with a bit of join(1), cut(1), grep(1) you can do basic rdbms queries on text files. The thing to remember about join is it requires both input files to be sorted on the key. an example join command: w | awk '{print $1}' | sort | join -j 1 -t: -o 2.5 - <(getent passwd | sort) to break this down w | List the currently logged in users and pipe its output to: awk '{print $1}' | awk then for each line of the file prints the first column (getting rid of the other random junk w has) and pipes its output to: sort | Both import files must be sorted, so we sort the input. yay. This gets piped to the join. The join also reads from a named pipe (thats what the <() is about) consisting of: getent passwd | sort List everyone in the passwd database and sort it (everyone is in ldap here, so the usual sort /etc/passwd doesn't work) The join command then joins on field 1 on both files (-j 1), considering fields to be separated by :'s (-t :), and outputting column 5 from file 2 (-o 2.5), and reading from standard in (our post processed w pipeline) and from the other passwd file parsing pipeline. So the net result of this command is that it lists the gecos information (real name etc) of all the users currently logged in, a bit long winded, but useful on multiuser machines where you don't directly know who that random person is that's running ./memhog. ---- [1]: Which nobody uses.