wireshark/make-manuf

#!/usr/bin/perl -w
#
# $Id$
#
# Make-manuf - Creates a file containing ethernet OUIs and their
# company IDs.  It merges the databases at
# http://standards.ieee.org/regauth/oui/index.shtml and
# http://www.cavebear.com/CaveBear/Ethernet/
# with entries in our template file.
#
# The script reads the comments at the top of "manuf.tmpl" and writes
# them to "manuf".  It then joins the manufacturer listing in "manuf.tmpl"
# with the listing in "oui.txt", with the entries in "manuf.tmpl" taking
# precedence.

# LWP is part of the standard Perl module libwww 

eval "require LWP::UserAgent;";
if( $@ ) {
  die "LWP isn't installed. It is part of the standard Perl\n" .
	" module libwww.  Bailing.\n";
}
$agent    = LWP::UserAgent->new;

$template = "manuf.tmpl";
$wkatmpl  = "wka.tmpl";
$outfile  = "manuf";
$inheader = 1;
$ieee_url = "http://standards.ieee.org/regauth/oui/oui_public.txt";
$cb_url   = "http://www.cavebear.com/CaveBear/Ethernet/Ethernet.txt";
%oui_list = ();
$hp       = "[0-9a-fA-F]{2}";
$oui_re   = "$hp:$hp:$hp";
$cb_re    = "$hp$hp$hp";
$ieee_re  = "$hp-$hp-$hp";

$tmpl_added   = 0;
$cb_added     = 0;
$cb_skipped   = 0;
$ieee_added   = 0;
$ieee_skipped = 0;

sub shorten
{
  my $origmanuf = shift; 
  my $manuf = " " . $origmanuf . " ";
  # Remove any punctuation
  $manuf =~ tr/',.()/    /;
  # & isn't needed when Standalone
  $manuf =~ s/ \& / /g;
  # Remove any "the", "inc", "plc" ...
  $manuf =~ s/\s(the|inc|incorporated|plc||systems|corp|corporation|s\/a|a\/s|ab|ag|kg|gmbh|co|company|limited|ltd)(?= )//gi;
  # Convert to consistent case
  $manuf =~ s/(\w+)/\u\L$1/g;
  # Remove all spaces
  $manuf =~ s/\s+//g;
  # Truncate all names to a reasonable length, say 10 characters.
  $manuf = substr($manuf, 0, 10);

  if ($manuf =~ /\Q$origmanuf\E/i) {
    return $manuf;
  } else {
    return sprintf("%-22s # %s", $manuf, $origmanuf);
  }
}

# Write out the header and populate the OUI list with our entries.

open (TMPL, "< $template") || 
  die "Couldn't open template file for reading ($template)\n";

while ($line = <TMPL>) {
  chomp($line);
  if ($line !~ /^$oui_re\s+\S/ && $inheader) {
    $header .= "$line\n";
  } elsif (($oui, $manuf) = ($line =~ /^($oui_re)\s+(\S.*)$/)) {
    $inheader = 0;
    # Ensure OUI is all upper-case
    $oui =~ tr/a-f/A-F/;
    # $oui_list{$oui} = &shorten($manuf);
    $oui_list{$oui} = $manuf;
    $tmpl_added++;
  }
}

# Add IEEE entries for OUIs not yet known.

print "Fetching $ieee_url.\n";
$request  = HTTP::Request->new(GET => $ieee_url);
$result   = $agent->request($request);

if (!$result->is_success) {
  die ("Error fetching $ieee_url: " . $result->status_line . "\n");
}
$ieee_list = $result->content;

foreach $line (split(/\n/, $ieee_list)) {
  if (($oui, $manuf) = ($line =~ /^($ieee_re)\s+\(hex\)\s+(\S.*)$/)) {
    $oui =~ tr /-/:/;  # The IEEE bytes are separated by dashes.
    # Ensure OUI is all upper-case
    $oui =~ tr/a-f/A-F/;
    if (exists $oui_list{$oui}) {
      printf "$oui - Skipping IEEE \"$manuf\" in favor of \"$oui_list{$oui}\"\n";
      $ieee_skipped++;
    } else {
      $oui_list{$oui} = &shorten($manuf);
      $ieee_added++;
    }
  }
}

# Add CaveBear entries for OUIs not yet known.

print "Fetching $cb_url.\n";
$request  = HTTP::Request->new(GET => $cb_url);
$result   = $agent->request($request);

if (!$result->is_success) {
  die ("Error fetching $cb_url: " . $result->status_line . "\n");
}
$cb_list = $result->content;

foreach $line (split(/\n/, $cb_list)) {
  if (($oui, $manuf) = ($line =~ /^($cb_re)\s+(\S.*)$/)) {
    ($h1, $h2, $h3) = ($oui =~ /($hp)($hp)($hp)/);  # The CaveBear bytes have no separators
    $oui = "$h1:$h2:$h3";
    # Ensure OUI is all upper-case
    $oui =~ tr/a-f/A-F/;
    if (exists $oui_list{$oui}) {
      # printf "$oui - Skipping CaveBear \"$manuf\" in favor of \"$oui_list{$oui}\"\n";
      $cb_skipped++;
    } else {
      printf "$oui - adding \"$manuf\" from CaveBear\n";
      $oui_list{$oui} = &shorten($manuf);
      $cb_added++;
    }
  }
}

# Write output file

open (OUT, "> $outfile") ||
  die "Couldn't open output file for writing ($outfile)\n";

print(OUT "# This file was generated by running ./make-manuf.\n");
print(OUT "# Don't change it directly, change manuf.tmpl and wka.tmpl instead.\n#\n");
print(OUT "$header");

foreach $oui (sort(keys %oui_list)) {
  print(OUT "$oui\t$oui_list{$oui}\n");
}

# Write out a blank line separating the OUIs from the well-known
# addresses, and then read the well-known address template file
# and write it to the manuf file.

open (WKATMPL, "< $wkatmpl") || 
  die "Couldn't open well-known address template file for reading ($wkatmpl)\n";

# XXX - it'd be nice to get this from the Cavebear file, but inferring
# the address mask from entries in that file involves some work.
#
print(OUT "\n");
while ($line = <WKATMPL>) {
  chomp($line);
  print(OUT "$line\n");
}

$total_added = $tmpl_added + $cb_added + $ieee_added;
print <<"Fin"
Original entries : $tmpl_added
IEEE added       : $ieee_added
CaveBear added   : $cb_added
Total            : $total_added

IEEE skipped     : $ieee_skipped
CaveBear skipped : $cb_skipped
Fin