I fairly pretty uncomfortable running any of the solutions presented here: I know enough about C/C++ to know that it's pretty easy to write code that can be pretty destructive to an instance of an operating system. The last thing I want to do is bring down the production servers that I'm maintaining that are important enough to make activity logs with a program designed to make the logs readable...
I ended up cobbling something together with Python. Thanks to the suggestions from Micah Cowan and Will Stevenson, I had source code to look at while cobbling... It's not perfect, by any means. I lose the meaning of anything that happens from inside a text editor. I should probably get in the habit of cat'ing the files I change, before and after the changes made, so that the changes are included in the maintenance logs... ... but it's good enough that I can usually read most of what happens, and the original isn't destroyed in the process, so I always have that, too... -----Original Message----- From: Micah Cowan [mailto:mi...@cowan.name] Sent: Friday, January 30, 2009 3:24 AM To: Jae Norment Cc: screen-users@gnu.org Subject: Re: using screen logs -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Jae Norment wrote: > I use screen to log sessions where I patch Debian servers. Screen > captures ANSI control codes (like positioning and color changes), which > is probably appropriate, however, I want a version of the logs without > those codes so that I can open it in a text editor. I've been told that > a little perl could help, been directed to 'script', both of which don't > seem to do the trick that I want. For the perl solution, I need to know > too much about the ANSI codes that I want to replace. Script just > dumps... there doesn't seem to be a good way to capture without the > codes... > > Screen is a widely used application, and my use to generate logs of my > sessions has got to be a fairly popular function of screen... so what is > everyone else using to make the logs human readable? ( Ideally, I'd > like something that will reformat the log either to a file or to a pipe. > ) I think you can use my Teseq program for this. I was going to give it a "strip" option, but haven't yet. In the meantime, you could do: teseq -DLE screenlog.0 | reseq - - You can get teseq from ftp.gnu.org/gnu/teseq/. This has the advantage of stripping out entire escape sequences, where stripping out only control characters would leave the printable majority from sequences like \E[31;1m. However, it would still leave non-escape-sequence control characters in (the main one you'd need to worry about is probably backspace), so you might want to additionally pipe it through that sed script that was recommended (after teseq/reseq). - -- Micah J. Cowan Programmer, musician, typesetting enthusiast, gamer. GNU Maintainer: wget, screen, teseq http://micah.cowan.name/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkmCxy8ACgkQ7M8hyUobTrFgZQCeLFOlnmJ8KgG7sKvAy5LSIOsR AnYAmgMXaR+gR06zHF5TyFORiB+FmFDQ =KdlU -----END PGP SIGNATURE----- _______________________________________________ screen-users mailing list screen-users@gnu.org http://lists.gnu.org/mailman/listinfo/screen-users
