Iterating over two lists in parallel in /bin/sh
Question
I have two lists of equal length, with no spaces in the individual items:
list1="a b c"
list2="1 2 3"
I want to iterate over these two lists in parallel, pairing a with 1, b with 2, etc.:
a 1
b 2
c 3
I'm attempting to support modern portable Bourne shell, so Bash/ksh arrays aren't available. Shelling out to awk would be acceptable in a pinch, but I'd rather keep this in pure sh if possible.
Thank you for any pointers you can provide!
Solution
This is a bit hacky but does the job:
#!/bin/sh
list1="1 2 3"
list2="a b c"
while [ -n "$list1" ]
do
head1=`echo "$list1" | cut -d ' ' -f 1`
list1=`echo "$list1" | sed 's/[^ ]* *\(.*\)$/\1/'`
head2=`echo "$list2" | cut -d ' ' -f 1`
list2=`echo "$list2" | sed 's/[^ ]* *\(.*\)$/\1/'`
echo $head1 $head2
done
OTHER TIPS
Probably not portable (look at all those bash-isms!), but it is easy to read and someone else might find it useful...
list1="a b c"
list2="1 2 3"
array1=($list1)
array2=($list2)
count=${#array1[@]}
for i in `seq 1 $count`
do
echo ${array1[$i-1]} ${array2[$i-1]}
done
This should be a fairly clean solution, but unless you use bash
's process substition, it requires the use of temporary files. I don't know if that's better or worse than invoking cut
and sed
over every iteration.
#!/bin/sh
list1="1 2 3"
list2="a b c"
echo $list1 | sed 's/ /\n/g' > /tmp/a.$$
echo $list2 | sed 's/ /\n/g' > /tmp/b.$$
paste /tmp/a.$$ /tmp/b.$$ | while read item1 item2; do
echo $item1 - $item2
done
rm /tmp/a.$$
rm /tmp/b.$$
This should be portable and also works with more than two lists:
#!/bin/sh
x="1 2 3 4 5"
y="a b c d e"
z="A B C D E"
while
read current_x x <<EOF
$x
EOF
read current_y y <<EOF
$y
EOF
read current_z z <<EOF
$z
EOF
[ -n "$current_x" ]
do
echo "x=$current_x y=$current_y z=$current_z"
done
Using positional paramers works, too. Please note, the list elements may not start with "-". Otherwise "set" will fail.
#!/bin/sh
x="1 2 3 4 5"
y="a b c d e"
z="A B C D E"
while
[ -n "$x" ]
do
set $x
current_x=$1
shift
x="$*"
set $y
current_y=$1
shift
y="$*"
set $z
current_z=$1
shift
z="$*"
echo "x=$current_x y=$current_y z=$current_z"
done
NEVERMIND, SAW "BOURNE" and thought "BOURNE AGAIN". Leaving this here because it might be useful for someone, but clearly not the answer to the question asked, sorry!
--
This has some shortcomings (it doesn't gracefully handle lists that are different sizes), but it works for the example you gave:
#!/bin/bash
list1="a b c"
list2="1 2 3"
c=0
for i in $list1
do
l1[$c]=$i
c=$(($c+1))
done
c=0
for i in $list2
do
echo ${l1[$c]} $i
c=$(($c+1))
done
There are more graceful ways using common unix tools like awk and cut, but the above is a pure-bash implementation as requested
Commenting on the accepted answer, it didn't work for me in either linux or Solaris, the problem was the \S character class shortcut in the regexp for sed. I replaced it with [^ ] and it worked:
#!/bin/sh
list1="1 2 3"
list2="a b c"
while [ -n "$list1" ]
do
head1=`echo "$list1" | cut -d ' ' -f 1`
list1=`echo "$list1" | sed 's/[^ ]* *\(.*\)$/\1/'`
head2=`echo "$list2" | cut -d ' ' -f 1`
list2=`echo "$list2" | sed 's/[^ ]* *\(.*\)$/\1/'`
echo $head1 $head2
done
Solution not using arrays:
list1="aaa1 aaa2 aaa3"
list2="bbb1 bbb2 bbb3"
tmpfile1=$( mktemp /tmp/list.XXXXXXXXXX ) || exit 1
tmpfile2=$( mktemp /tmp/list.XXXXXXXXXX ) || exit 1
echo $list1 | tr ' ' '\n' > $tmpfile1
echo $list2 | tr ' ' '\n' > $tmpfile2
paste $tmpfile1 $tmpfile2
rm --force $tmpfile1 $tmpfile2
As a one liner:
list2="1 2 3";
list1="a b c";
for i in $list1; do
x=`expr index "$list2" " "`;
[ $x -eq 0 ] && j=$list2 || j=${list2:0:$x};
list2=${list2:$x};
echo "$i $j";
done
$ list1="1 2 3"
$ list2="a b c"
$ echo "$list1 $list2" | awk '{n=NF/2; for (i=1;i<=n;i++) print $i,$(n+i) }'
1 a
2 b
3 c
I had been working on a sed-based answer when the first solutions started showing up here. But upon further investigation, it turned out that the items in the list were separated by newlines, not spaces, which allowed me to go with a solution based on head and tail:
original_revs="$(cd original && git rev-parse --all)" &&
working_revs="$(cd working && git rev-parse --all)" &&
while test -n "$original_revs"; do
original_commit="$(echo "$original_revs" | head -n 1)" &&
working_commit="$(echo "$working_revs" | head -n 1)" &&
original_revs="$(echo "$original_revs" | tail -n +2)" &&
working_revs="$(echo "$working_revs" | tail -n +2)" &&
...
done
I'm posting this just in case somebody encounters this variant of the problem, but I'm awarding the accepted answer based on the problem as posted.