Trailing (only) cr/lf Removal

[tbohon]

Scripting on an AIX box with Korn shell, I need to be able to truncate just the trailing cr/lf pair on the very last record of a multi-hundred record file. I've looked and don't even see the characters but file recipient (it's ftp-d from me to him) insists that the blank line is causing problems with their pickup/processing routine.

Assuming there is a cr/lf there (all of the other records have one) how do I remove just the one on the last record?

Tnx.

Tom

"My mind is like a steel whatchamacallit ...

 

[PHV]

Something like this ?
sed '$d' /path/to/input > output

Hope This Helps, PH.
FAQ219-2884
FAQ181-2886

 

[tbohon]

Not sure exactly what this is doing - can you elaborate a bit? I don't use sed all that often and, every time I do, I have to dig back into the books.

Tnx.

"My mind is like a steel whatchamacallit ...

 

[PHV]

Tip: man sed

 

[feherke]

Hi

Note that PHV's code will remove the last line, whatever it contains. To remove the last line only if is empty :

sed '${/^$/d}' /path/to/input > output

Tested with GNU [tt]sed[/tt]. I am sure it does not work with AIX's [tt]sed[/tt], but use it as an idea.

Feherke.

http://rootshell.be/~feherke/

 

[tbohon]

Feherke:

Thanks, I'll give that a shot. PHV's example seemed to work but the recipient - who is writing the parsing program there from scratch, apparently - said it's still causing problems on his end. Since the file is Unix standard format - lf at end of each record, cr/lf at end of the file - I'm not sure I can get rid of just that last cr. He did ask that I change the file so that all records end in a cr/lf pair - that he can apparently program for - but I'm trying to get this done under a deadline and don't want to go back through all the testing steps again.

Appreciate the assist, both of you.

Tom

"My mind is like a steel whatchamacallit ...

 

[PHV]

it's still causing problems on his end
Which problem ?

Hope This Helps, PH.
FAQ219-2884
FAQ181-2886

 

[feherke]

Hi

Hmm...

sed '${/^[red]\r[/red]$/d}' /path/to/input > output

Feherke.

http://rootshell.be/~feherke/

 

[bigoldbulldog]

Tom,

Show us the bytes to see what's actually there. Why not give a hexdump/tail of your file, e.g:

$ od -x /path/to/file | tail -n 4

Cheers,
ND

 

[p5wizard]

IMHO this is something for the reception side to take care of (or to live with). On FTP'ing a textfile from unix to windows, end of line markers from unix (LF) are translated to windows (CR+LF).

Even if you would remove the last LF of a textfile and then FTP it to windows, the last line still gets a CR+LF added again. E.g. windows editor shows this as an empty line after the last text line. Normal behaviour if you ask me.


HTH,

p5wizard

 

[kornShellScripter]

Although not an exact answer for your specific problem, I'm having to handle similar issues with the files input to me.

So thought I'd just mention the things I am dealing with at the moment in case it gives you any further ideas:

I'm writing a Unix korn shell script to analyse text files. So I make use of sed/grep/nawk. They all expect the file to be in Unix format - ie the end of the lines need to end with \n (or "lf" or "\012" however you want to put it).

I receive files with \r\n or sometimes \r\f (carriage return, form feed) which messes things up for me.

So now as I read in the files, I just do this:
cat $FILE_NAME | tr '\015' '\012' | <etc etc>
which would double space my file if read by a human, but fixes my problems - "tr '\015' '\012'" translates all carriage-returns to line-feeds.
Sometimes I'll do "tr -d '\015'" which means "delete all carriage-returns"

I found od -ta to be a useful od command when analysing my file.