Splitting comma separated values into an array

Shell Programming and Scripting
1

I'm attempting to create a KSH array out of a string like this: ",,,value1,value2,,"

I have created the array but I only get two elements, one for value1 and one for value2.

I have ended up with something like this but I don't like it:

set -A JUNK
xx=0
for i in $(print ",,,value1,value2,,"  | nawk '{gsub(/,/," \n",$0);print}')
do
    print ".$i."
    JUNK[xx]=$i
    (( xx += 1 ))
done
for i in "${JUNK[@]}"
do
    print ".$i."
done

The results:

. .
. .
. .
.value1 .
.value2 .
. .
. .

If i leave out the space character before the newline in this gsub command:

gsub(/,/," \n",$0);

...I get this output:

.value1 .
.value2 .

I really don't want the spaces at the end of each string and I nead an array of every CSV field (even the empty values).

I have tried several approaches including changing IFS and looping through the fields but this doesn't seem to work at all.

What am I missing here?

Thomas

2

Where do these csv lines come from? I think that I've seen this sort of question before.

And I think that there is an issue that arises when the data contains a comma. What happens then? I seem to remember that the field with a comma is surrounded by quotes, but I don't remember for sure.

I guess that there is (a perhaps null) field before the first comma and after the last comma. Is this correct?

3

Okay, now I am confused.

Believe it or not, my first inclination was to code something like this:


typeset IFS=,
set -A JUNK $(print ",,,value1,value2,,")

set | grep JUNK

JUNK[0]=''
JUNK[1]=''
JUNK[2]=''
JUNK[3]=value1
JUNK[4]=value2
JUNK[5]=''

I can not figure out why didn't work previously! :mad:

Ah well, I've been wrestling with it for a while and now it suddenly works.
I simply don't get it.

It looks like this case is closed.

Thomas

4

Sorry, our postings must have crossed paths. The csv values are going to be generated from a Java program that I can direct the design of at this stage. The values represent customer parameters to drive some data loads and they are positional. If there were two parameters I would be making simple assignments but since there are several, I decided to use an array.

I can have any number of csv lines to parse during a load session so it's not like picking up a one-time set of control parameters.

As I indicated in my last post, I have apparently ended up in the most logical approach working but I don't know why. I have worked through this for hours and used, what I thought, the exact approach that I documented with no luck. I stumbled on the working lines when I thought that commas could have been the issue and I replaced them with colons. When colons worked, I tried commas again. They worked too!

I don't get it.

BTW, it seems that you are correct with respect to the quotation marks but that is inconsequential. I only care if the parameter was provided or not. I can cope with quotes.

thomas