I am not 100% sure if every sed understands this form of character-encoding. Here is a solution which may look a bit longer but will work with every sed:

Code:

sed 's/\([0-9]*[^0-9,][^0-9,]*[0-9]*\),/\'\1\',/g
     s/\([0-9]*[^0-9,][^0-9,]*[0-9]*\)$/\'\1\',/' /path/to/file

I get the notion from your own script that you want to exclude the first line (header) from the processing. If so, modify the above like this:

Code:

sed '1 !{ s/\([0-9]*[^0-9,][^0-9,]*[0-9]*\),/\'\1\',/g
          s/\([0-9]*[^0-9,][^0-9,]*[0-9]*\)$/\'\1\',/
        }' /path/to/file

Also notice that you may want to not repeat the operation if there are already single quotes in place, i.e. you may want to transform

Code:

foo,bar,999,,       ==> 'foo','bar',999,,

but probably not want this:

Code:

'foo','bar',999,,       ==> ''foo'',''bar'',999,,

and instead leave it unchanged. If this is the case you should change the sed-script this way:

Code:

sed '1 !{ s/\'*\([0-9]*[^0-9,][^0-9,]*[0-9]*\)\'*,/\'\1\',/g
          s/\'*\([0-9]*[^0-9,][^0-9,]*[0-9]*\)\'*$/\'\1\',/
        }' /path/to/file

You may want to trim trailing blanks first, otherwise you might get this (blanks encoded as "<b>"):

Code:

foo,bar,999,,<b><b>     ==> 'foo','bar',999,,'<b><b>'

which is probably not what you want. Therefore a last modification (replace "<b>" and "<t>" with literal blanks/tabs) :

Code:

sed 's/[<b><t>]*$//
     1 !{ s/\'*\([0-9]*[^0-9,][^0-9,]*[0-9]*\)\'*,/\'\1\',/g
          s/\'*\([0-9]*[^0-9,][^0-9,]*[0-9]*\)\'*$/\'\1\',/
        }' /path/to/file

I hope this helps.

bakunin