irctail - Read all IRC channels in one terminal


The general idea is to tailf multiple irssi log files into a single one, and
tailf from that into a pretty colorized file:

user@host $ for x in irclogs/*; do tailf $x | sed /^/$x/ >>/tmp/irssitail & ; done

In practice this is implemented as:

user@host $ while true; do ~/build/irctail/irctail-spawn.sh; sleep 10; done
user@host $ tailf /tmp/irssitail | ./irssi-to-base.py | ./baseformat.py | ./irc-colorize.py >>/tmp/feed

The continuous loop above allows to pick up any channels you join later on.
This could alternatively be put in a crontab.

To view the results a simple tailf will do, but screen is useful to rewrap the
text when resizing the terminal:

user@host$ screen tailf /tmp/feed

The reason for a separate /tmp/irssitail and /tmp/feed is to give room to tailf
lots of different (non IRC) things into /tmp/feed, and so only one instance of
the output colorizer is needed, instead of one per channel.

Put this (or similar) in .irssi/config:
  core = {
    log_timestamp = "%s ";
  };
  "fe-common/core" = {
    autolog = "yes";
    autolog_level = "all -joins -parts -quits";
    autolog_colors = "yes";
  };

Files:

irctail-spawn.sh
  Checks for any new log files to tailf into /tmp/irssitail
  (also prepends the channel name to each line)

irctail-reap.sh
  Kills any old (or all) tailf's created by irctail-spawn.sh
  Only run this if you have rotating logs and want to kill tailf's on
  logs that have been rotated out.

baseformat.py
  For any line in "base" format, performs right-alignment on names,
  colorizes channel names and nicks as well as in text, and prints the
  log time in human readable form.

irssi-to-base.py
  Converts some types of Irssi log line into a format that is easier
  to parse ("base" format).

irc-colorize.py
  Interprets IRC color sequences and prints equivalent bash escape
  sequences.


Improvements welcome :)


If /tmp/feed or /tmp/irssitail grow too large, just do:

user@host $ printf '' >/tmp/feed; printf '' >/tmp/irssitail


tail -F can be used with multiple files (instead of spawning one per log), but
in that case we'd need to pipe into a program to reformat "==> #python.log <=="
style lines, which is more faff than just using sed.