remove any leading ::ffff: on ipv4 addresses

[spider.git] / perl / AnnTalk.pm

diff --git a/perl/AnnTalk.pm b/perl/AnnTalk.pm
index 02bb0a9bedca8393479b4ef40668ff6eb7b6b97b..ee7ea51570164916e925f08f34eb24c79dae46ee 100644 (file)
--- a/perl/AnnTalk.pm
+++ b/perl/AnnTalk.pm
@@ -17,8 +17,8 @@ use DXVars;
 
 use vars qw(%dup $duplth $dupage $filterdef);
 
-$duplth = 60;                                  # the length of text to use in the deduping
-$dupage = 5*3600;                      # the length of time to hold ann 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],
@@ -47,8 +47,7 @@ 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 $dupkey = "A$call|$to|\L$text";
@@ -67,6 +66,7 @@ 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;