ksh questions

[Stuart Abramson_2]

Hpux-admins:

I have two "variable strings" in a ksh script:

OLD_MPS="/u1001 /u1101 /u1102 /u1201 . . . /uXYYY"
NEW_MPS="/u2001 /u2101 /u2102 /u2201 . . . /uZQQQ"

I want to go through a mounted directory, and, respectively, for every occurrence of:
/u1001 replace it with /u2001
/u1101 replace it with /u2101

I don't want to replace "/u1" with "/u2". I actually want to replace the 1st occurrence in the "OLD_" list with the 1st occurrence in the "NEW_" list, the 2nd with the 2nd, the 3rd with the 3rd, etc.

My problem is not the outside loop, but the inside:

for OMP in $OLD_MPS
do
echo $OMP
# match with $NMP in $NEW_MPS

Can I turn my original variable strings into arrays? Then I could match
OLD_MPS[0] with NEW_MPS[0], etc. How would I do that?

Stuart

[curt larson_1 · Accepted solution]

set [+/-]A name args
-A will cause name to be unset prior to the assignment

set -A oldMps $OLD_MPS
is how to do it

[curt larson_1]

hopefully this won't be an issue,
but i believe that your limited to 1K array elements, i.e subscipts 0-1023

[Wouter Jagers]

My 'top of the head' way to do this would be the following. A little counter should solve your problem. It might not be optimal, but hey, we're just renaming directories right ?

(note that I took the slashes out of OLD_MPS and NEW_MPS)

----------
OLD_MPS="u1001 u1101 u1102 u1201"
NEW_MPS="u2001 u2101 u2102 u2201"
TARGETDIR="."
count=1

for OMP in $OLD_MPS
do
NMP=`echo $NEW_MPS | cut -d " " -f$count`
find $TARGETDIR -type d -name $OMP -exec mv {} $NMP \; 2>/dev/null
echo $OMP to $NMP: done.
count=`expr $count + 1`
done
----------

hope it helps
regards
Wouter

[john korterman]

Hi Stuart,
this is a simple example of reading $1 into an array:

#!/usr/bin/sh
typeset -i POS=0 i=1
while [ $POS -lt ${#1} ]
do
POS=$POS+1
KARAK=`echo ${1} | cut -c $POS`
STRENG[$POS]="$KARAK"
done

while [ $i -le $POS ]
do
echo "${STRENG[$i]}"
let i="$i+1"
done


but you of course need to take care of the "correct" number of spaces in order to e to make a sensible comparison; and use plenty of qoutes, e.g.:
# ./make_array.sh "/u1001 /u1101 /u1102 /u1201"

regards,
John K.

[curt larson_1]

Done put quotes around the array values, as in:
set -A oldMps "$OLD_MPS"

you'll only get one array element

if you more the 1K values, you could try something like this

x=1
for OMP in $OLD_MPS
do
echo $OMP
echo $NEW_MPS | awk '{print $'$x';}'
x=$(( $x + 1 ))
done

but i think awk is limited to about 3K fields

[Elmar P. Kolkman]

Using a counter can also be used otherwise...

idx=1
fld1=$(echo $OLD_MPS | cut -f$idx)
while [ -n $fld1 ]
do
if [ -d $fld1 ]
then
fld2=$(echo $NEW_MPS | cut -f$idx)
mv $fld1 $fld2
fi
idx=$(expr $idx + 1)
fld1=$(echo $OLD_MPS | cut -f$idx)
done

And there is another way too, by using not OLD_MPS as the loops input, but the content of your directory... But that is too much work to do out of head.

[curt larson_1]

another way

#put each field on a seperate line
#create a coprocess
print "$NEW_MPS | tr "[:space:]" "\012*" |&

print "$OLD_MPS | tr "[:space:]" "\012*" |
while read old
do
read -p new
print "$old $new"
done

[Stuart Abramson_2]

Thanks Curt:

Use "set -A" to turn 2 lists of corresponding values into two correspoinding arrays:

OLD_MPS=/u1001 /u1101 /u1102 /u1201
NEW_MPS=/u2001 /u2101 /u2102 /u2201

set -A OLD_ARRAY $OLD_MPS
set -A NEW_ARRAY $NEW_MPS
CNT=${#OLD_ARRAY[@]}
#
(( I=0 ))
while (( I < CNT ))
do
echo ${OLD_ARRAY[$I]} ${NEW_ARRAY[$I]}
(( I=I+1 ))
done