Shell Programming and Scripting

Hi,

I want to insert the text 'Unknown' in 2 specific columns in a csv file (actually | separated) if the column is blank. Its always the same columns.

I have tried using sed: sed "s/||/|Unknown|/g"

but there are occasion where other fields are blank and they need to be left blank. This command inserts Unknowns in all blank fields.

Any help much appreciated
Thanks
Kieran

checking columns 3 and 5

Code:

awk -F='|' '$3=="" {$3="Unknown"}; $5=="" {$5="Unknown"};1' OFS='|' myFile

The "g" at the end replaces ALL occurrences, therefore you had all those false positives. You can put a number there to replace only the n-th occurrence of the match. "1" is implied if you omit the number, therefore

Code:

sed "s/||/|Unknown|/"

will only change the first occurrence. As your file is probably record-oriented this might work.

If it doesn't work you will have to extend your regexp to match only the one field you want to change. A field looks like: "any number (including 0 if the field is empty) of non-'|' followed by a '|'". The regexp for a single field is:

Code:

/[^|]*|/

Now repeat this n-1 times and add the n-th field you are looking for empty. We will have to group with "\(...\)" and use the regexp multiplier "\{...\}". After this we will need an additional grouping because we need to preserve the beginning of the line unchanged.

For instance, if n=6:

Code:

sed '/^\(\([^|]*|\)\{5\}\)||/\1|Unknown|/'

I hope this helps.

bakunin

Code:

sed 's/[|] *[|]/|Unknown|/g'

I'm curious. Why do you believe that your suggestion will behave differently from what was already tried?

Regards,
Alister