tools/make-manuf

   1 #!/usr/bin/perl -w
   2 #
   3 # $Id$
   4 #
   5 # Make-manuf - Creates a file containing ethernet OUIs and their
   6 # company IDs.  It merges the databases at IEEE and
   7 # http://www.cavebear.com/archive/cavebear/Ethernet/Ethernet.txt
   8 # with entries in our template file.
   9 #
  10 # The script reads the comments at the top of "manuf.tmpl" and writes
  11 # them to "manuf".  It then joins the manufacturer listing in "manuf.tmpl"
  12 # with the listing in "oui.txt", "iab.txt", etc, with the entries in
  13 # "manuf.tmpl" taking precedence.
  14
  15 # LWP is part of the standard Perl module libwww
  16
  17 # As of April 2012 the IEEE content is mostly UTF-8 encoded although some
  18 # of the entries feature sequences listed at
  19 # http://www.i18nqa.com/debug/utf8-debug.html
  20
  21 use Encode;
  22 use open ':encoding(utf8)';
  23
  24 eval "require LWP::UserAgent;";
  25 if( $@ ) {
  26   die "LWP isn't installed. It is part of the standard Perl\n" .
  27         " module libwww.  Bailing.\n";
  28 }
  29
  30 $revision = '$Revision$';
  31 if ($revision !~ /[0-9]/ ) { $revision = "unknown"; }
  32
  33 $agent    = LWP::UserAgent->new;
  34 $agent->env_proxy;
  35 $agent->agent("Wireshark make-manuf/$revision");
  36
  37 $template = "manuf.tmpl";
  38 $wkatmpl  = "wka.tmpl";
  39 $outfile  = "manuf";
  40 $inheader = 1;
  41 $oui_url  = "http://standards.ieee.org/develop/regauth/oui/oui.txt";
  42 $iab_url  = "http://standards.ieee.org/develop/regauth/iab/iab.txt";
  43 $oui36_url  = "http://standards.ieee.org/develop/regauth/oui36/oui36.txt";
  44 $cb_url   = "http://www.cavebear.com/archive/cavebear/Ethernet/Ethernet.txt";
  45 %oui_list = ();
  46 $hp       = "[0-9a-fA-F]{2}";
  47 $oui_re   = "$hp:$hp:$hp";
  48 $cb_re    = "$hp$hp$hp";
  49 $ieee_re  = "$hp-$hp-$hp";
  50
  51 $tmpl_added   = 0;
  52 $cb_added     = 0;
  53 $cb_skipped   = 0;
  54 $oui_added   = 0;
  55 $oui_skipped = 0;
  56 $iab_added   = 0;
  57 $iab_skipped = 0;
  58 $oui36_added   = 0;
  59 $oui36_skipped = 0;
  60
  61 sub shorten
  62 {
  63   my $origmanuf = shift;
  64   my $manuf = " " . $origmanuf . " ";
  65   # Remove any punctuation
  66   $manuf =~ tr/',.()/    /;
  67   # & isn't needed when Standalone
  68   $manuf =~ s/ \& / /g;
  69   # Remove any "the", "inc", "plc" ...
  70   $manuf =~ s/\s(the|inc|incorporated|plc||systems|corp|corporation|s\/a|a\/s|ab|ag|kg|gmbh|co|company|limited|ltd|holding|spa)(?= )//gi;
  71   # Convert to consistent case
  72   $manuf =~ s/(\w+)/\u\L$1/g;
  73   # Remove all spaces
  74   $manuf =~ s/\s+//g;
  75   # Truncate all names to a reasonable length, say, 8 characters.
  76   # If the string contains UTF-8, this may be substantially more than 8 bytes.
  77   $manuf = substr($manuf, 0, 8);
  78
  79   if ($manuf =~ /\Q$origmanuf\E/i) {
  80     return $manuf;
  81   } else {
  82     return sprintf("%-22s # %s", $manuf, $origmanuf);
  83   }
  84 }
  85
  86 sub fetch
  87 {
  88   my $url = shift;
  89   print "Fetching $url.\n";
  90   $request  = HTTP::Request->new(GET => $url);
  91   $result   = $agent->request($request);
  92
  93   if (!$result->is_success) {
  94     die ("Error fetching $url: " . $result->status_line . "\n");
  95   }
  96   return decode("utf-8", $result->content);
  97 }
  98
  99 # Write out the header and populate the OUI list with our entries.
 100
 101 open (TMPL, "< $template") ||
 102   die "Couldn't open template file for reading ($template)\n";
 103
 104 while ($line = <TMPL>) {
 105   chomp($line);
 106   if ($line !~ /^$oui_re\s+\S/ && $inheader) {
 107     $header .= "$line\n";
 108   } elsif (($oui, $manuf) = ($line =~ /^($oui_re)\s+(\S.*)$/)) {
 109     $inheader = 0;
 110     # Ensure OUI is all upper-case
 111     $oui =~ tr/a-f/A-F/;
 112     # $oui_list{$oui} = &shorten($manuf);
 113     $oui_list{$oui} = $manuf;
 114     $tmpl_added++;
 115   }
 116 }
 117
 118 # Add IEEE entries for IABs
 119
 120 $ieee_list = fetch($iab_url);
 121
 122 foreach $line (split(/\n/, $ieee_list)) {
 123   # determine the OUI used for IAB (currently only 00-50-C2)
 124   if (($iab_tmp, $manuf) = ($line =~ /^\s*($ieee_re)\s+\(hex\)\s+(\S.*)$/)) {
 125       $iab_base = $iab_tmp;
 126   }
 127   # determine next two bytes
 128   if (($iab4, $iab5, $manuf) = ($line =~ /^\s*($hp)($hp)$hp-$hp$hp$hp\s+\(base\s16\)\s+(\S.*)$/)) {
 129     $iab = "$iab_base:$iab4:$iab5:00/36";
 130     $iab =~ tr /-/:/; # The IEEE bytes are separated by dashes.
 131     # Ensure IAB is all upper-case
 132     $iab =~ tr/a-f/A-F/;
 133     if (exists $oui_list{$iab}) {
 134       printf "$iab - Skipping IEEE \"$manuf\" in favor of \"$oui_list{$iab}\"\n";
 135       $iab_skipped++;
 136     } else {
 137       $oui_list{$iab} = &shorten($manuf);
 138       $iab_added++;
 139     }
 140   }
 141 }
 142
 143 # Add IEEE entries for OUI-36
 144
 145 $ieee_list = fetch($oui36_url);
 146
 147 foreach $line (split(/\n/, $ieee_list)) {
 148   # determine the OUI used for OUI-36 (currently only 00-1B-C5)
 149   if (($oui36_tmp, $manuf) = ($line =~ /^\s*($ieee_re)\s+\(hex\)\s+(\S.*)$/)) {
 150       $oui36_base = $oui36_tmp;
 151   }
 152   # determine next two bytes
 153   if (($oui36_4, $oui36_5, $manuf) = ($line =~ /^\s*($hp)($hp)$hp-$hp$hp$hp\s+\(base\s16\)\s+(\S.*)$/)) {
 154     $oui36 = "$oui36_base:$oui36_4:$oui36_5:00/36";
 155     $oui36 =~ tr /-/:/; # The IEEE bytes are separated by dashes.
 156     # Ensure OUI-36 is all upper-case
 157     $oui36 =~ tr/a-f/A-F/;
 158     if (exists $oui_list{$oui36}) {
 159       printf "$oui36 - Skipping IEEE \"$manuf\" in favor of \"$oui_list{$oui36}\"\n";
 160       $oui36_skipped++;
 161     } else {
 162       $oui_list{$oui36} = &shorten($manuf);
 163       $oui36_added++;
 164     }
 165   }
 166 }
 167
 168 # Add IEEE entries for OUIs not yet known.
 169
 170 $ieee_list = fetch($oui_url);
 171
 172 foreach $line (split(/\n/, $ieee_list)) {
 173   if (($oui, $manuf) = ($line =~ /^\s*($ieee_re)\s+\(hex\)\s+(\S.*)$/)) {
 174     $oui =~ tr /-/:/;  # The IEEE bytes are separated by dashes.
 175     # Ensure OUI is all upper-case
 176     $oui =~ tr/a-f/A-F/;
 177     if (exists $oui_list{$oui}) {
 178       printf "$oui - Skipping IEEE \"$manuf\" in favor of \"$oui_list{$oui}\"\n";
 179       $oui_skipped++;
 180     } else {
 181       $oui_list{$oui} = &shorten($manuf);
 182       $oui_added++;
 183     }
 184   }
 185 }
 186
 187 # Add CaveBear entries for OUIs not yet known.
 188
 189 $cb_list = fetch($cb_url);
 190
 191 foreach $line (split(/\n/, $cb_list)) {
 192   if (($oui, $manuf) = ($line =~ /^($cb_re)\s+(\S.*)$/)) {
 193     ($h1, $h2, $h3) = ($oui =~ /($hp)($hp)($hp)/);  # The CaveBear bytes have no separators
 194     $oui = "$h1:$h2:$h3";
 195     # Ensure OUI is all upper-case
 196     $oui =~ tr/a-f/A-F/;
 197     if (exists $oui_list{$oui}) {
 198       # printf "$oui - Skipping CaveBear \"$manuf\" in favor of \"$oui_list{$oui}\"\n";
 199       $cb_skipped++;
 200     } else {
 201       printf "$oui - adding \"$manuf\" from CaveBear\n";
 202       $oui_list{$oui} = &shorten($manuf);
 203       $cb_added++;
 204     }
 205   }
 206 }
 207
 208 # Write output file
 209
 210 open (OUT, "> $outfile") ||
 211   die "Couldn't open output file for writing ($outfile)\n";
 212
 213 print(OUT "# This file was generated by running ./make-manuf.\n");
 214 print(OUT "# Don't change it directly, change manuf.tmpl and wka.tmpl instead.\n#\n");
 215 print(OUT "$header");
 216
 217 foreach $oui (sort(keys %oui_list)) {
 218   print(OUT "$oui\t$oui_list{$oui}\n");
 219 }
 220
 221 # Write out a blank line separating the OUIs from the well-known
 222 # addresses, and then read the well-known address template file
 223 # and write it to the manuf file.
 224
 225 open (WKATMPL, "< $wkatmpl") ||
 226   die "Couldn't open well-known address template file for reading ($wkatmpl)\n";
 227
 228 # XXX - it'd be nice to get this from the Cavebear file, but inferring
 229 # the address mask from entries in that file involves some work.
 230 #
 231 print(OUT "\n");
 232 while ($line = <WKATMPL>) {
 233   chomp($line);
 234   print(OUT "$line\n");
 235 }
 236
 237 $total_added = $tmpl_added + $cb_added + $oui_added + $iab_added;
 238 print <<"Fin"
 239 Original entries : $tmpl_added
 240 IEEE OUI added   : $oui_added
 241 IEEE IAB added   : $iab_added
 242 IEEE OUI36 added : $oui36_added
 243 CaveBear added   : $cb_added
 244 Total            : $total_added
 245
 246 IEEE OUI skipped : $oui_skipped
 247 IEEE IAB skipped : $iab_skipped
 248 IEEE OUI36 skipd : $oui36_skipped
 249 CaveBear skipped : $cb_skipped
 250 Fin