#!/usr/bin/perl

# Trim an http transcript, removing things you don't want to play back.
# Copyright 2002-2007 Patrick Killelea under the GPL.
# You may freely copy and redistribute this file.
# Version: Tue Jan  8 15:11:48 PST 2008
# Please write p@patrick.net with improvements.
# Documentation below.

use bytes; # don't allow unicode crap

# Read whole file in single gulp.
undef $/;
$_ = <>;

@units = split(/(.*>>>>>>>>.*\n|.*<<<<<<<<.*\n)/);

# shift to ignore everything before 1st request marker.
shift @units;

# Tack delimiter back on to unit.
for ($i = 0; $units[$i]; $i += 2) {
    push @tmp, $units[$i] . $units[$i+1];
}

for (@units = @tmp) { s|.*<<<<<<<<.*?^(HTTP.*?)\r.*(<title>.*</title>).*|<<<<<<<<\n$1\n$2\n\n|smi }
for (@tmp = @units) { s|.*<<<<<<<<.*?^(HTTP.*?)\r.*|<<<<<<<<\n$1\n\n|smi }

# Remove the thread number and timestamp from request.
for (@units = @tmp) { s|.*>>>>>>>>.*?$|>>>>>>>>|smi }

print @units;

__END__

Usage: trim out.http > trimmed.http

The replay command works much better with trimmed http traffic.