John,Doe,Dallas,TX,75051
Jane,Doe,Dallas,TX,75052

Can anyone provide an example, using something like awk or sed, of how to
convert delimited records to fixed length records?
Where the desired output is a fixed-length record with:

First name as 15 characters, blank padded on the right
Last name as 20 characters, blank padded on the right
City as 20 characters, blank padded on the right
State as 2 characters
ZIP code as 5 characters

Something like this:

John            Doe                Dallas                TX75051
Jane            Doe                Dallas                TX75052

Thanks in advance,

David

  :: Make a demo text file
  echo John,Doe,Dallas,TX,75051> tmp$$$.txt
  echo Jane,Doe,Dallas,TX,75052>>tmp$$$.txt
  ::
  :: Do it
  gawk -F, '{printf"%%-15s%%-20s%%-20s%%-2s%%-5s\n",$1,$2,$3,$4,$5}' tmp$$$.txt
  ::
  :: Clean up
  for %%f in (tmp$$$.*) do if exist %%f del %%f

   All the best, Timo

--
Prof. Timo Salmi ftp & http://garbo.uwasa.fi/ archives 193.166.120.5
Department of Accounting and Business Finance  ; University of Vaasa

Useful batch files and tricks ftp://garbo.uwasa.fi/pc/link/tsbat.zip

--

Linux solution for data management and processing.

Other than that, a fine post.

I wasn't going to post, but since I took the time to flame your post, I
figure I might as well give an on-topic, AWK solution:

BEGIN   { nf = split("15 20 20 2 5",widths);FS = "," }
{
for (i=1; i<=nf; i++) printf("%-*s",widths[i],$i)
print ""

        cat food | ...
with:
        < food ...

(which is ugly [compared to the usual paradigm of doing the redirections at
the end of the line], but does preserve that "input on the left" feel that
some people seem to think important)

However, it seems you can't do this with a while loop:

        < food while read foo

didn't parse in any of the sh-ish shells I tried.  So, you have to do:

        while read foo
        do ...
        done < food

   All the best, Timo

--
Prof. Timo Salmi ftp & http://garbo.uwasa.fi/ archives 193.166.120.5
Department of Accounting and Business Finance  ; University of Vaasa

Useful batch files and tricks ftp://garbo.uwasa.fi/pc/link/tsbat.zip

According to Kenny, "[the one-liner] doesn't work with original (e.g.,
Solaris) awk."

I can't test either

    awk -F ',' '{printf "%-15s%-20s%-20s%2s%5s\n", $1, $2, $3, $4, $5}'

or

    BEGIN   { nf = split("15 20 20 2 5",widths);FS = "," }
    {
        for (i=1; i<=nf; i++) printf("%-*s",widths[i],$i)
        print ""
    }

on Solaris but that particular awk is apparently so flakey (Search Google
Groups for "Sun", "Solaris") that I believe anything bad anyone says about
it.  Also, because it is so bad, I see no reason anyone would want to
specifically make sure their code was compatible with that version* (so I
wonder why Kenny brought it up?).

*Except for antique programming in an environment that doesn't have a better
awk (Solaris ships with another working awk, but it's not in the path), or
the ability to install Gnu Awk.

    - Dan

Still a one liner and more efficient than the 5 liner if anyone
bothered to check it out.

$ cat sed.scr

# first, add more than enough spaces in front of every comma
s/,/                         ,/g
# the line below sets width of second-from-right field
s/\(.* ,..\) *,/\1/
s/\(.* ,...................\) *,/\1/
s/\(.* ,...................\) *,/\1/
# the line below sets width of left-most field
s/\(...............\) *,/\1/

$ sed -f sed.scr data

John           Doe                Dallas             TX75051
Jane           Doe                Dallas             TX75052

I've avoided the \{ extension so this will work with all versions of sed.

Note: any spaces in your comma-separated data will mess with my script,
so if any data can contain spaces replace each space in the sed script
with say, an underscore if no underscores occur in the data, and append
a command to as a final step resubstitute those underscores with spaces:
s/_/ /g

--
John Savage            (for email, replace "ks" with "k" and delete "n")