adjust startup descriptions a bit

[spider.git] / perl / AnnTalk.pm

diff --git a/perl/AnnTalk.pm b/perl/AnnTalk.pm
index a38f9cb3b7459a98317da876b61cabd07263f0df..defc0110aa250848e7a30e7925ee89ff8200a489 100644 (file)
--- a/perl/AnnTalk.pm
+++ b/perl/AnnTalk.pm
@@ -10,15 +10,18 @@ package AnnTalk;
 
 use strict;
 
+use DXVars;
 use DXUtil;
 use DXDebug;
 use DXDupe;
-use DXVars;
+use DXLog;
+use DXLogPrint;
+use Time::HiRes qw(gettimeofday tv_interval);
 
 use vars qw(%dup $duplth $dupage $filterdef);
 
-$duplth = 60;                                  # the length of text to use in the deduping
-$dupage = 5*24*3600;                   # the length of time to hold spot dups
+$duplth = 30;                                  # the length of text to use in the deduping
+$dupage = 18*3600;                             # the length of time to hold ann dups
 $filterdef = bless ([
                          # tag, sort, field, priv, special parser 
                          ['by', 'c', 0],
@@ -36,7 +39,25 @@ $filterdef = bless ([
                          ['origin_zone', 'nz', 12],
                          ['by_state', 'nz', 13],
                          ['origin_state', 'nz', 14],
-                        ], 'Filter::Cmd');
+                  ], 'Filter::Cmd');
+
+our $maxcache = 130;
+our @anncache;
+
+sub init
+{
+       my $t0 = [gettimeofday];
+       @anncache = DXLog::search(0, $maxcache, $main::systime, 'ann');
+       shift @anncache while @anncache > $maxcache;
+       my $l = @anncache;
+       dbg("AnnTalk: loaded last $l announcements into cache in " . _diffms($t0) . "mS");
+}
+
+sub add_anncache
+{
+       push @anncache, [ $main::systime, @_ ];
+       shift @anncache while @anncache > $maxcache;
+}
 
 # enter the spot for dup checking and return true if it is already a dup
 sub dup
@@ -47,11 +68,10 @@ sub dup
        chomp $text;
        unpad($text);
        $text =~ s/\%([0-9A-F][0-9A-F])/chr(hex($1))/eg;
-       $text = Encode::encode("iso-8859-1", $text) if $main::can_encode && Encode::is_utf8($text, 1);
-       $text = pack("C*", map {$_ & 127} unpack("C*", $text));
+#      $text = Encode::encode("iso-8859-1", $text) if $main::can_encode && Encode::is_utf8($text, 1);
        $text =~ s/[^\#a-zA-Z0-9]//g;
        $text = substr($text, 0, $duplth) if length $text > $duplth; 
-       my $dupkey = "A$to|\L$text";
+       my $dupkey = "A$call|$to|\L$text";
        return DXDupe::check($dupkey, $t);
 }
 
@@ -67,6 +87,7 @@ sub is_talk_candidate
 {
        my ($from, $text) = @_;
        my $call;
+
        ($call) = $text =~ /^\s*(?:[Xx]|[Tt][Oo]?:?)\s+([\w-]+)/;
        ($call) = $text =~ /^\s*>\s*([\w-]+)\b/ unless $call;
        ($call) = $text =~ /^\s*([\w-]+):?\b/ unless $call;