Delete Multiple Column Headers

rduncan10 · 08-21-2007, 08:01 AM

Hi,

Our ERP system on SUSE can print reports to a printer or an ASCII file. I am writing a script to strip away all the page header/footer content from the ASCII file, leaving just the data, so that it can be imported into another program.

I've got most of this stuff cleaned out, but I'm left with the column headers, which are repeated at the top of each page.

I am tryting to find out if there is some sed command for deleting all but the first occurence of a line.

I can't use uniq because there might be duplicate lines within the data that I don't want to delete.

I can do it like this:

Code:

hdr=$(head -1 file)
echo $hdr > newfile
cat file | sed '/$hdr/d' >> newfile
or 
cat file | grep -v $hdr >> newfile

This works fine, but I was wondering if there might be some sed trick to this that might work better.

Thanks.

ghostdog74 · 08-21-2007, 08:40 AM

show how your file looks like...

rduncan10 · 08-21-2007, 01:02 PM

Code:

#!/bin/sh                            
                                     
cat $1 | grep "|" > $1.tmp             # get rid of all lines that don't have | character
                                       # (all header/and footer lines)
hdr=$(head -1 $1.tmp)                  # store first line (column header)
echo $hdr > $1                         # add column header to new file.
cat $1.tmp | grep -v "$hdr" >> $1      # Add remaining lines to new file, except those matching
                                       # header
rm $1.tmp

This works fine, but I just thought there might be some sed technique to do this.

Thanks.

radoulov · 08-21-2007, 01:24 PM

Quote:

Originally Posted by rduncan10

Code:

sed '2,${/^I/d;}' infile