How to keep from duplicating path variable in csh

StackOverflow https://stackoverflow.com/questions/135754

Question

It is typical to have something like this in your cshrc file for setting the path:

set path = ( . $otherpath $path )

but, the path gets duplicated when you source your cshrc file multiple times, how do you prevent the duplication?

EDIT: This is one unclean way of doing it:

set localpaths = ( . $otherpaths )
echo ${path} | egrep -i "$localpaths" >& /dev/null
if ($status != 0) then
    set path = ( . $otherpaths $path )
endif
Was it helpful?

Solution

you can use the following Perl script to prune paths of duplicates.

#!/usr/bin/perl
#
# ^^ ensure this is pointing to the correct location.
#
# Title:    SLimPath
# Author:   David "Shoe Lace" Pyke <eselle@users.sourceforge.net >
#   :   Tim Nelson 
# Purpose: To create a slim version of my envirnoment path so as to eliminate
#       duplicate entries and ensure that the "." path was last.
# Date Created: April 1st 1999
# Revision History:
#   01/04/99: initial tests.. didn't wok verywell at all
#       : retreived path throught '$ENV' call
#   07/04/99: After an email from Tim Nelson <wayland@ne.com.au> got it to
#         work.
#       : used 'push' to add to array
#       : used 'join' to create a delimited string from a list/array.
#   16/02/00: fixed cmd-line options to look/work better
#   25/02/00: made verbosity level-oriented
#
#

use Getopt::Std;

sub printlevel;

$initial_str = "";
$debug_mode = "";
$delim_chr = ":";
$opt_v = 1;

getopts("v:hd:l:e:s:");

OPTS: {
    $opt_h && do {
print "\n$0 [-v level] [-d level] [-l delim] ( -e varname | -s strname | -h )";
print "\nWhere:";
print "\n   -h  This help";
print "\n   -d  Debug level";
print "\n   -l  Delimiter (between path vars)";
print "\n   -e  Specify environment variable (NB: don't include \$ sign)";
print "\n   -s  String (ie. $0 -s \$PATH:/looser/bin/)";
print "\n   -v  Verbosity (0 = quiet, 1 = normal, 2 = verbose)";
print "\n";
        exit;
    };
    $opt_d && do {
        printlevel 1, "You selected debug level $opt_d\n";
        $debug_mode = $opt_d;
    };
    $opt_l && do {
        printlevel 1, "You are going to delimit the string with \"$opt_l\"\n";
        $delim_chr = $opt_l;
    };
    $opt_e && do {
        if($opt_s) { die "Cannot specify BOTH env var and string\n"; }
        printlevel 1, "Using Environment variable \"$opt_e\"\n";
        $initial_str = $ENV{$opt_e};
    };
    $opt_s && do {
        printlevel 1, "Using String \"$opt_s\"\n";
        $initial_str = $opt_s;
    };
}

if( ($#ARGV != 1) and !$opt_e and !$opt_s){
    die "Nothing to work with -- try $0 -h\n";
}

$what = shift @ARGV;
# Split path using the delimiter
@dirs = split(/$delim_chr/, $initial_str);

$dest;
@newpath = ();
LOOP: foreach (@dirs){
    # Ensure the directory exists and is a directory
    if(! -e ) { printlevel 1, "$_ does not exist\n"; next; }
    # If the directory is ., set $dot and go around again
    if($_ eq '.') { $dot = 1; next; }

#   if ($_ ne `realpath $_`){
#           printlevel 2, "$_ becomes ".`realpath $_`."\n";
#   }
    undef $dest;
    #$_=Stdlib::realpath($_,$dest);
    # Check for duplicates and dot path
    foreach $adir (@newpath) { if($_ eq $adir) { 
        printlevel 2, "Duplicate: $_\n";
        next LOOP; 
    }}

    push @newpath, $_;
}

# Join creates a string from a list/array delimited by the first expression
print join($delim_chr, @newpath) . ($dot ? $delim_chr.".\n" : "\n");

printlevel 1, "Thank you for using $0\n";
exit;

sub printlevel {
    my($level, $string) = @_;

    if($opt_v >= $level) {
        print STDERR $string;
    }
}

i hope thats useful.

OTHER TIPS

Im surprised no one used the tr ":" "\n" | grep -x techique to search if a given folder already exists in $PATH. Any reason not to?

In 1 line:

if ! $(echo "$PATH" | tr ":" "\n" | grep -qx "$dir") ; then PATH=$PATH:$dir ; fi

Here is a function ive made myself to add several folders at once to $PATH (use "aaa:bbb:ccc" notation as argument), checking each one for duplicates before adding:

append_path()
{
    local SAVED_IFS="$IFS"
    local dir
    IFS=:
    for dir in $1 ; do
        if ! $( echo "$PATH" | tr ":" "\n" | grep -qx "$dir" ) ; then
            PATH=$PATH:$dir
        fi
    done
    IFS="$SAVED_IFS"
}

It can be called in a script like this:

append_path "/test:$HOME/bin:/example/my dir/space is not an issue"

It has the following advantages:

  • No bashisms or any shell-specific syntax. It run perfectly with !#/bin/sh (ive tested with dash)
  • Multiple folders can be added at once
  • No sorting, preserves folder order
  • Deals perfectly with spaces in folder names
  • A single test works no matter if $folder is at begginning, end, middle, or is the only folder in $PATH (thus avoiding testing x:*, *:x, :x:, x, as many of the solutions here implicitly do)
  • Works (and preserve) if $PATH begins or ends with ":", or has "::" in it (meaning current folder)
  • No awk or sed needed.
  • EPA friendly ;) Original IFS value is preserved, and all other variables are local to the function scope.

Hope that helps!

ok, not in csh, but this is how I append $HOME/bin to my path in bash...

case $PATH in
    *:$HOME/bin | *:$HOME/bin:* ) ;;
    *) export PATH=$PATH:$HOME/bin
esac

season to taste...

I've been using the following (Bourne/Korn/POSIX/Bash) script for most of a decade:

:   "@(#)$Id: clnpath.sh,v 1.6 1999/06/08 23:34:07 jleffler Exp $"
#
#   Print minimal version of $PATH, possibly removing some items

case $# in
0)  chop=""; path=${PATH:?};;
1)  chop=""; path=$1;;
2)  chop=$2; path=$1;;
*)  echo "Usage: `basename $0 .sh` [$PATH [remove:list]]" >&2
    exit 1;;
esac

# Beware of the quotes in the assignment to chop!
echo "$path" |
${AWK:-awk} -F: '#
BEGIN   {   # Sort out which path components to omit
            chop="'"$chop"'";
            if (chop != "") nr = split(chop, remove); else nr = 0;
            for (i = 1; i <= nr; i++)
                omit[remove[i]] = 1;
        }
{
    for (i = 1; i <= NF; i++)
    {
        x=$i;
        if (x == "") x = ".";
        if (omit[x] == 0 && path[x]++ == 0)
        {
            output = output pad x;
            pad = ":";
        }
    }
    print output;
}'

In Korn shell, I use:

export PATH=$(clnpath /new/bin:/other/bin:$PATH /old/bin:/extra/bin)

This leaves me with PATH containing the new and other bin directories at the front, plus one copy of each directory name in the main path value, except that the old and extra bin directories have bin removed.

You would have to adapt this to C shell (sorry - but I'm a great believer in the truths enunciated at C Shell Programming Considered Harmful). Primarily, you won't have to fiddle with the colon separator, so life is actually easier.

Well, if you don't care what order your paths are in, you could do something like:

set path=(`echo $path | tr ' ' '\n' | sort | uniq | tr '\n' ' '`)

That will sort your paths and remove any extra paths that are the same. If you have . in your path, you may want to remove it with a grep -v and re-add it at the end.

Here is a long one-liner without sorting:
set path = ( echo $path | tr ' ' '\n' | perl -e 'while (<>) { print $_ unless $s{$_}++; }' | tr '\n' ' ')

dr_peper,

I usually prefer to stick to scripting capabilities of the shell I am living in. Makes it more portable. So, I liked your solution using csh scripting. I just extended it to work on per dir in the localdirs to make it work for myself.

foreach dir ( $localdirs )
    echo ${path} | egrep -i "$dir" >& /dev/null
    if ($status != 0) then
        set path = ( $dir $path )
    endif
end

Using sed(1) to remove duplicates. 

$ PATH=$(echo $PATH | sed -e 's/$/:/;s/^/:/;s/:/::/g;:a;s#\(:[^:]\{1,\}:\)\(.*\)\1#\1\2#g;ta;s/::*/:/g;s/^://;s/:$//;')

This will remove the duplicates after the first instance, which may or may not be what you want, e.g.:

$ NEWPATH=/bin:/usr/bin:/bin:/usr/local/bin:/usr/local/bin:/bin
$ echo $NEWPATH | sed -e 's/$/:/; s/^/:/; s/:/::/g; :a; s#\(:[^:]\{1,\}:\)\(.*\)\1#\1\2#g; t a; s/::*/:/g; s/^://; s/:$//;'
/bin:/usr/bin:/usr/local/bin
$

Enjoy!

Here's what I use - perhaps someone else will find it useful:

#!/bin/csh
#  ABSTRACT
#    /bin/csh function-like aliases for manipulating environment
#    variables containing paths.
#
#  BUGS
#    - These *MUST* be single line aliases to avoid parsing problems apparently related
#      to if-then-else
#    - Aliases currently perform tests in inefficient in order to avoid parsing problems
#    - Extremely fragile - use bash instead!!
#
#  AUTHOR
#    J. P. Abelanet - 11/11/10

#  Function-like alias to add a path to the front of an environment variable
#    containing colon (':') delimited paths, without path duplication
#
#  Usage: prepend_path ENVVARIABLE /path/to/prepend
alias prepend_path \
  'set arg2="\!:2";  if ($?\!:1 == 0) setenv \!:1 "$arg2";  if ($?\!:1 && $\!:1 !~ {,*:}"$arg2"{:*,}) setenv \!:1 "$arg2":"$\!:1";'

#  Function-like alias to add a path to the back of any environment variable 
#    containing colon (':') delimited paths, without path duplication
#
#  Usage: append_path ENVVARIABLE /path/to/append
alias append_path \
  'set arg2="\!:2";  if ($?\!:1 == 0) setenv \!:1 "$arg2";  if ($?\!:1 && $\!:1 !~ {,*:}"$arg2"{:*,}) setenv \!:1 "$\!:1":"$arg2";'

I always set my path from scratch in .cshrc. That is I start off with a basic path, something like:

set path = (. ~/bin /bin /usr/bin /usr/ucb /usr/bin/X11)

(depending on the system).

And then do:

set path = ($otherPath $path)

to add more stuff

I have the same need as the original question. Building on your previous answers, I have used in Korn/POSIX/Bash:

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

I had difficulties to translate it directly in csh (csh escape rules are insane). I have used (as suggested by dr_pepper):

set path = ( `echo $otherpath $path | tr ' ' '\n' | perl -ne 'print $_ unless $h{$_}++' | tr '\n' ' '`)

Do you have ideas to simplify it more (reduce the number of pipes) ?

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top