example-scripts/10grep

   1 #!/usr/bin/perl -w
   2
   3 # Copyright (C) 2001 Simon Huggins
   4
   5 # This program is free software; you can redistribute it and/or modify it
   6 # under the terms of the GNU General Public License as published by the Free
   7 # Software Foundation; either version 2 of the License, or (at your option)
   8 # any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful, but
  11 # WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
  12 # or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  13 # for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc., 59
  17 # Temple Place, Suite 330, Boston, MA 02111-1307  USA
  18
  19 use strict;
  20
  21 my (@tags,%words,%blacklist);
  22
  23 if (defined $cfg{'grep_debug'} and $cfg{'grep_debug'}) {
  24         open(DEBUG, ">$cfg{'grep_debugfile'}");
  25 }
  26
  27 open(HANDLE, "<$cfg{'tagfile'}") or htagdie "Could not open $cfg{'tagfile'}: $!\n";
  28 @tags=<HANDLE>;
  29 close(HANDLE);
  30
  31 while(<DATA>) {
  32         chomp;
  33         s/\s+//g;
  34         $blacklist{$_}++;
  35 }
  36
  37 open(MSG, "<$cfg{'msgfile'}");
  38 while(<MSG>) {
  39         s/[\s\t\n]+/ /g;
  40         tr/A-Za-z0-9 //dc; # delete non-alphanumeric
  41         s/\b\d+\b//g;
  42         $_ = lc $_;
  43         my @words = split;
  44         foreach (@words) {
  45                 next if length($_)>9;
  46                 $words{$_}++ if not exists $blacklist{$_};
  47         }
  48 }
  49 close(MSG);
  50
  51 my @goodtags;
  52 my $count=0;
  53 foreach my $key (sort { $words{$b} <=> $words{$a} }
  54                 keys %words) {
  55         print DEBUG "$key occurred $words{$key} times\n" if $cfg{'grep_debug'};
  56         my @foundtags = grep { /\b$key\b/i } @tags;
  57         push @goodtags,@foundtags; # Tags with more than one matching word will get
  58                                    # pushed on more than one time
  59         print DEBUG join "\n",@foundtags if $cfg{'grep_debug'};
  60         $count++;
  61         last if $count >20;
  62 }
  63
  64 open(OUT, ">$cfg{'tmptagfile'}")
  65         or htagdie "$0: Could not open $cfg{'tmptagfile'}: $!\n";
  66 reg_deletion("$cfg{'tmptagfile'}");
  67 if (@goodtags) {
  68         print OUT $goodtags[rand(@goodtags)];
  69 } else {
  70         exit(5);
  71 }
  72
  73 END {
  74         close(OUT);
  75
  76         if ($cfg{'grep_debug'}) {
  77                 close(DEBUG);
  78         }
  79 }
  80
  81 __DATA__
  82 a
  83 about
  84 again
  85 all
  86 am
  87 an
  88 and
  89 another
  90 any
  91 apr
  92 are
  93 arent
  94 as
  95 at
  96 aug
  97 be
  98 because
  99 been
 100 before
 101 being
 102 but
 103 by
 104 can
 105 cant
 106 cat
 107 could
 108 dec
 109 did
 110 do
 111 doesnt
 112 dont
 113 down
 114 ehlo
 115 esmtp
 116 even
 117 every
 118 feb
 119 for
 120 fri
 121 from
 122 gmt
 123 go
 124 great
 125 had
 126 hadnt
 127 has
 128 have
 129 he
 130 her
 131 here
 132 hers
 133 herself
 134 him
 135 himself
 136 his
 137 how
 138 however
 139 i
 140 id
 141 if
 142 im
 143 in
 144 instead
 145 into
 146 is
 147 it
 148 its
 149 itself
 150 ive
 151 jan
 152 jul
 153 jun
 154 know
 155 like
 156 lots
 157 mar
 158 may
 159 maybe
 160 me
 161 might
 162 might
 163 mine
 164 mon
 165 more
 166 must
 167 my
 168 near
 169 need
 170 new
 171 no
 172 not
 173 nov
 174 now
 175 oct
 176 of
 177 off
 178 oh
 179 on
 180 or
 181 ought
 182 ours
 183 out
 184 over
 185 please
 186 quite
 187 received
 188 said
 189 same
 190 sat
 191 seem
 192 seemed
 193 seems
 194 sep
 195 she
 196 should
 197 should
 198 smtp
 199 so
 200 some
 201 such
 202 sun
 203 than
 204 that
 205 thats
 206 the
 207 their
 208 theirs
 209 them
 210 then
 211 there
 212 theres
 213 these
 214 they
 215 this
 216 thu
 217 to
 218 tom
 219 too
 220 tue
 221 up
 222 us
 223 very
 224 want
 225 was
 226 we
 227 wed
 228 well
 229 went
 230 were
 231 what
 232 when
 233 which
 234 while
 235 who
 236 why
 237 will
 238 with
 239 wont
 240 would
 241 would
 242 yes
 243 yet
 244 you
 245 your
 246 youre
 247 yours
 248 youve