eg/unuc

   1 #!/usr/bin/perl
   2
   3 print STDERR "Loading proper nouns...\n";
   4 open(DICT,"/usr/dict/words") || die "Can't find /usr/dict/words: $!\n";
   5 while (<DICT>) {
   6     if (/^[A-Z]/) {
   7         chop;
   8         ($lower = $_) =~ y/A-Z/a-z/;
   9         $proper{$lower} = $_;
  10     }
  11 }
  12 close DICT;
  13 print STDERR "Loading exceptions...\n";
  14
  15 $prog = <<'EOT';
  16 while (<>) {
  17     next if /[a-z]/;
  18     y/A-Z/a-z/;
  19     s/(\w+)/$proper{$1} ? $proper{$1} : $1/eg;
  20     s/^(\s*)([a-z])/$1 . (($tmp = $2) =~ y:a-z:A-Z:,$tmp)/e;
  21     s/([-.?!]["']?(\n\s*|  \s*)["']?)([a-z])/$1 . (($tmp = $3) =~ y:a-z:A-Z:,$tmp)/eg;
  22     s/\b([b-df-hj-np-tv-xz]+)\b/(($tmp = $1) =~ y:a-z:A-Z:,$tmp)/eg;
  23     s/([a-z])'([SDT])\b/$1 . "'" . (($tmp = $2) =~ y:A-Z:a-z:,$tmp)/eg;
  24 EOT
  25 while (<DATA>) {
  26     chop;
  27     next if /^$/;
  28     next if /^#/;
  29     if (! /;$/) {
  30         $foo = $_;
  31         $foo =~ y/A-Z/a-z/;
  32         print STDERR "Dup $_\n" if $proper{$foo};
  33         $foo =~ s/([^\w ])/\\$1/g;
  34         $foo =~ s/ /(\\s+)/g;
  35         $foo = "\\b" . $foo if $foo =~ /^\w/;   # XXX till patch 9
  36         $foo .= "\\b" if $foo =~ /\w$/;
  37         $i = 0;
  38         ($bar = $_) =~ s/ /'$' . ++$i/eg;
  39         $_ = "s/$foo/$bar/gi;";
  40     }
  41     $prog .= '    ' . $_ . "\n";
  42 }
  43 $prog .= "}\ncontinue {\n    print;\n}\n";
  44
  45 $/ = '';
  46 #print $prog;
  47 eval $prog; die $@ if $@;
  48 __END__
  49 A.M.
  50 Air Force
  51 Air Force Base
  52 Air Force Station
  53 American
  54 Apr.
  55 Ariane
  56 Aug.
  57 August
  58 Bureau of Labor Statistics
  59 CIT
  60 Caltech
  61 Cape Canaveral
  62 Challenger
  63 China
  64 Corporation
  65 Crippen
  66 Daily News in Brief
  67 Daniel Quayle
  68 Dec.
  69 Discovery
  70 Edwards
  71 Endeavour
  72 Feb.
  73 Ford Aerospace
  74 Fri.
  75 General Dynamics
  76 George Bush
  77 Headline News
  78 HOTOL
  79 I
  80 II
  81 III
  82 IV
  83 IX
  84 Institute of Technology
  85 JPL
  86 Jan.
  87 Jul.
  88 Jun.
  89 Kennedy Space Center
  90 LDEF
  91 Long Duration Exposure Facility
  92 Long March
  93 Mar.
  94 March
  95 Martin
  96 Martin Marietta
  97 Mercury
  98 Mon.
  99 in May
 100 s/\bmay (\d)/May $1/g;
 101 s/\boffice of (\w)/'Office of ' . (($tmp = $1) =~ y:a-z:A-Z:,$tmp)/eg;
 102 National Science Foundation
 103 NASA Select
 104 New Mexico
 105 Nov.
 106 OMB
 107 Oct.
 108 Office of Management and Budget
 109 President
 110 President Bush
 111 Richard Truly
 112 Rocketdyne
 113 Russian
 114 Russians
 115 Sat.
 116 Sep.
 117 Soviet
 118 Soviet Union
 119 Soviets
 120 Space Shuttle
 121 Sun.
 122 Thu.
 123 Tue.
 124 U.S.
 125 Union of Soviet Socialist Republics
 126 United States
 127 VI
 128 VII
 129 VIII
 130 Vice President
 131 Vice President Quayle
 132 Wed.
 133 White Sands
 134 Kaman Aerospace
 135 Aerospace Daily
 136 Aviation Week
 137 Space Technology
 138 Washington Post
 139 Los Angeles Times
 140 New York Times
 141 Aerospace Industries Association
 142 president of
 143 Johnson Space Center
 144 Space Services
 145 Inc.
 146 Co.
 147 Hughes Aircraft
 148 Company
 149 Orbital Sciences
 150 Swedish Space
 151 Arnauld
 152 Nicogosian
 153 Magellan
 154 Galileo
 155 Mir
 156 Jet Propulsion Laboratory
 157 University
 158 Department of Defense
 159 Orbital Science
 160 OMS
 161 United Press International
 162 United Press
 163 UPI
 164 Associated Press
 165 AP
 166 Cable News Network
 167 Cape York
 168 Zenit
 169 SYNCOM
 170 Eastern
 171 Western
 172 Test Range
 173 Jcsat
 174 Japanese Satellite Communications
 175 Defence Ministry
 176 Defense Ministry
 177 Skynet
 178 Fixed Service Structure
 179 Launch Processing System
 180 Asiasat
 181 Launch Control Center
 182 Earth
 183 CNES
 184 Glavkosmos
 185 Pacific
 186 Atlantic