Shell Programming and Scripting

I wanted to see if there is any duplicate of a specific command in the command search path. The following code will list all copies of "openssl" in the command search path.

Code:

find `printenv PATH | sed "s/:/ /g"`  -maxdepth 1  -name openssl

However, the above code would fail if the search path contains any directory whose name contains a space such as "my commands". The main reason for the failure is that a space is the delimiter in the list of directories for the "find" command.

The following workaround changes the delimiter from a space to a colon by modifying the global variable IFS.

Code:

IFS=":"; find `printenv PATH`  -maxdepth 1  -name openssl; IFS=$' \t\n';

However, I do not want to modify global variables. Can anyone suggest me any better methods? The method must work even if a directory name contains a space.

Thanks in advance.

blimey!
as you can see it works!

(the parentheses ensure that you don't clobber your IFS variable)

Code:

cat:$(IFS=:;echo $PATH|fmt -1| sort| uniq -c)
      1 /bin
      3 /home/mbillingham/bin
      2 /home/mbillingham/w/ntools/scripts
      2 /sbin
      1 /usr/bin
      1 /usr/games
      1 /usr/local/bin
      2 /usr/sbin

you can substitute uniq -d just to show duplicates.

If I understand correctly, you can use something like this to remove any duplicate entries from your PATH:

Code:

PATH="$(perl -e'print join ":", grep !$_{$_}++, split ":",$ENV{PATH}')"

With zsh all you need is:

Code:

typeset -U PATH

Edit: I just saw that you want to find the duplicates, so:

Code:

perl -le'print join $/, grep $_{$_}++==1, split ":",$ENV{PATH}'

or did you mean this?

Code:

( IFS=:;find  $PATH|fmt -1| sort| uniq -c )

again uniq -d may be what you want

Sorry for the confusion, but I was looking for duplicate commands (or duplicate executable files) but NOT duplicate directories.

For example, while the authentic copy of the executable file "openssl" resides in /usr/bin/, another copy of "openssl" may reside in /sbin/ by accident or by malicious intention. I would like to detect whether multiple copies of an executable file with the same name exist in different directories.

To simplify the problem, assume that I am looking for files whose name is "openssl". The search for "openssl" should be done through the directories listed in PATH (the command search path). I do not want the search to waste time in looking into directories that are not listed in PATH.

I would like the search method to meet the following two conditions. First, the search has to work even if a directory name contains a space. Second, I prefer not to modify the global variable IFS.

I showed two codes in Post #1. The first code met the second condition, but failed in the first condition.

find `printenv PATH | sed "s/:/ /g"` -maxdepth 1 -name openssl

The second code met the first condition, but failed in the second condition.

IFS=":"; find `printenv PATH` -maxdepth 1 -name openssl; IFS=$' \t\n';

I am not sure if there is any elegant method to accomplish this task without modifying IFS. However, if anyone can think of such a method, let me know.

Thanks in advance.

Hi,

If you're using bash:

Code:

type -a name

will list all commands named "name" in your PATH along with any identically named built-ins, functions, and aliases.

Alister

I believe that this small program, written by Randal L. Schwartz, could solve your problem.