Shell Programming and Scripting

I have a large csv file that looks like this:

The 3rd field is a unix time stamp that I want to convert to human readable.
I wrote a bash script with this code:

Code:

IFS=$','
cat $1  | while read ID user DATE text flags read; do
echo -e "$ID,$user,$(date -d @$DATE),$text,$flags,$read
done
unset IFS
exit 0

The problem is that the 4th field (text field) may or may not contain a comma, thus messing up my field separation, when it comes to outputting. The third field (when a comma is used as the FS) is ALWAYS the unix time stamp. I want to change change just the 3rd field in the file by substituting the unix time stamp for the human readable one (achieved with "date -d@DATE").

Any help much appreciated

stumpy

Are you sure you want to use pure shell for that?

Code:

perl -F, -lane'
  $F[2] = localtime $F[2] if $. > 1;
  print join ",", @F
  ' infile

or:

Code:

perl -lpe'
  s{((?:\d+,){2})(\d+)}{$1.localtime $2}e
  ' infile

Hi radoulov

Thanks very much for the help. I have no knowledge of perl, hence I went with bash. Your code was correct based on the information that I gave you.
I noticed some problems with the output on a large file that I processed with your code. It seems that there is new line characters in some of the 4th fields (text field) that mess up the output, I guess that is because a new line character acts as a Field Separator. Any thoughts on how to handle this. I am guessing that with newline characters in the text field I am sunk here?

I used your first suggestion (perl -F, -lane etc)

Hm,
could you post a sample of the problematic input?
I need it to reproduce the issue.