use DB_File;
use Data::Dumper;
use DXDebug;
+use DXUtil;
+use LRU;
use strict;
use vars qw($VERSION $BRANCH);
$VERSION = sprintf( "%d.%03d", q$Revision$ =~ /(\d+)\.(\d+)/ );
-$BRANCH = sprintf( "%d.%03d", q$Revision$ =~ /\d+\.\d+\.(\d+)\.(\d+)/ ) || 0;
+$BRANCH = sprintf( "%d.%03d", q$Revision$ =~ /\d+\.\d+\.(\d+)\.(\d+)/ || (0,0));
$main::build += $VERSION;
$main::branch += $BRANCH;
-use vars qw($db %prefix_loc %pre);
+use vars qw($db %prefix_loc %pre $lru $lrusize $misses $hits $matchtotal);
$db = undef; # the DB_File handle
%prefix_loc = (); # the meat of the info
%pre = (); # the prefix list
+$hits = $misses = $matchtotal = 1; # cache stats
+$lrusize = 1000; # size of prefix LRU cache
+
+$lru = LRU->newbase('Prefix', $lrusize);
sub load
{
+ # untie every thing
if ($db) {
undef $db;
untie %pre;
%pre = ();
%prefix_loc = ();
}
- $db = tie(%pre, "DB_File", undef, O_RDWR|O_CREAT, 0666, $DB_BTREE) or confess "can't tie \%pre ($!)";
+
+ # tie the main prefix database
+ $db = tie(%pre, "DB_File", undef, O_RDWR|O_CREAT, 0664, $DB_BTREE) or confess "can't tie \%pre ($!)";
my $out = $@ if $@;
do "$main::data/prefix_data.pl" if !$out;
$out = $@ if $@;
- # print Data::Dumper->Dump([\%pre, \%prefix_loc], [qw(pre prefix_loc)]);
- return $out;
+
+ return $out;
}
sub store
# This routine will only do what you ask for, if you wish to be intelligent
# then that is YOUR problem!
#
+
sub get
{
my $key = shift;
- my @out;
- my @outref;
my $ref;
- my $gotkey;
-
- $gotkey = $key;
+ my $gotkey = $key;
return () if $db->seq($gotkey, $ref, R_CURSOR);
return () if $key ne substr $gotkey, 0, length $key;
- @outref = map { $prefix_loc{$_} } split ',', $ref;
- return ($gotkey, @outref);
+ return ($gotkey, map { $prefix_loc{$_} } split ',', $ref);
}
#
sub next
{
my $key = shift;
- my @out;
- my @outref;
my $ref;
my $gotkey;
return () if $db->seq($gotkey, $ref, R_NEXT);
return () if $key ne substr $gotkey, 0, length $key;
- @outref = map { $prefix_loc{$_} } split ',', $ref;
- return ($gotkey, @outref);
+ return ($gotkey, map { $prefix_loc{$_} } split ',', $ref);
+}
+
+#
+# search for the nearest match of a prefix string (starting
+# from the RH end of the string passed)
+#
+
+sub matchprefix
+{
+ my $pref = shift;
+ my @partials;
+
+ for (my $i = length $pref; $i; $i--) {
+ $matchtotal++;
+ my $s = substr($pref, 0, $i);
+ push @partials, $s;
+ my $p = $lru->get($s);
+ if ($p) {
+ $hits++;
+ if (isdbg('prefix')) {
+ my $percent = sprintf "%.1f", $hits * 100 / $misses;
+ dbg("Partial Prefix Cache Hit: $s Hits: $hits/$misses of $matchtotal = $percent\%");
+ }
+ $lru->put($_, $p) for @partials;
+ return @$p;
+ } else {
+ $misses++;
+ my @out = get($s);
+ if (isdbg('prefix')) {
+ my $part = $out[0] || "*";
+ $part .= '*' unless $part eq '*' || $part eq $s;
+ dbg("Partial prefix: $pref $s $part" );
+ }
+ if (@out && $out[0] eq $s) {
+ $lru->put($_, \@out) for @partials;
+ return @out;
+ }
+ }
+ }
+ return ();
}
#
my $p;
my @parts;
my ($call, $sp, $i);
-
- foreach $call (split /,/, $calls) {
- # first check if the whole thing succeeds
- my @nout = get($call);
- push @out, @nout if @nout;
- next if @nout > 0 && $nout[0] eq $call;
-
+
+LM: foreach $call (split /,/, $calls) {
+
+ # first check if the whole thing succeeds either because it is cached
+ # or because it simply is a stored prefix as callsign (or even a prefix)
+ $matchtotal++;
+ $call =~ s/-\d+$//; # ignore SSIDs
+ my $p = $lru->get($call);
+ my @nout;
+ if ($p) {
+ $hits++;
+ if (isdbg('prefix')) {
+ my $percent = sprintf "%.1f", $hits * 100 / $misses;
+ dbg("Prefix Cache Hit: $call Hits: $hits/$misses of $matchtotal = $percent\%");
+ }
+ push @out, @$p;
+ next;
+ } else {
+ @nout = get($call);
+ if (@nout && $nout[0] eq $call) {
+ $misses++;
+ $lru->put($call, \@nout);
+ dbg("got exact prefix: $nout[0]") if isdbg('prefix');
+ push @out, @nout;
+ next;
+ }
+ }
+
# now split the call into parts if required
@parts = ($call =~ '/') ? split('/', $call) : ($call);
+ dbg("Parts: $call = " . join(' ', @parts)) if isdbg('prefix');
# remove any /0-9 /P /A /M /MM /AM suffixes etc
if (@parts > 1) {
- $p = $parts[0];
- shift @parts if $p =~ /^(WEB|NET)$/o;
- $p = $parts[$#parts];
- pop @parts if $p =~ /^(\d+|[JPABM]|AM|MM|BCN|JOTA|SIX|WEB|NET|Q\w+)$/o;
- $p = $parts[$#parts];
- pop @parts if $p =~ /^(\d+|[JPABM]|AM|MM|BCN|JOTA|SIX|WEB|NET|Q\w+)$/o;
-
+ @parts = grep { !/^\d+$/ && !/^[PABM]$/ && !/^(?:|AM|MM|BCN|JOTA|SIX|WEB|NET|Q\w+)$/; } @parts;
+
# can we resolve them by direct lookup
- foreach $p (@parts) {
- @nout = get($p);
- push @out, @nout if @nout;
- next if @nout > 0 && $nout[0] eq $call;
+ my $s = join('/', @parts);
+ @nout = get($s);
+ if (@nout && $nout[0] eq $s) {
+ dbg("got exact multipart prefix: $call $s") if isdbg('prefix');
+ $misses++;
+ $lru->put($call, \@nout);
+ push @out, @nout;
+ next;
}
}
+ dbg("Parts now: $call = " . join(' ', @parts)) if isdbg('prefix');
- # which is the shortest part (first if equal)?
- $sp = $parts[0];
- foreach $p (@parts) {
- $sp = $p if length $sp > length $p;
+ # at this point we should have two or three parts
+ # if it is three parts then join the first and last parts together
+ # to get an answer
+
+ # first deal with prefix/x00xx/single letter things
+ if (@parts == 3 && length $parts[0] <= length $parts[1]) {
+ @nout = matchprefix($parts[0]);
+ if (@nout) {
+ my $s = join('/', $nout[0], $parts[2]);
+ my @try = get($s);
+ if (@try && $try[0] eq $s) {
+ dbg("got 3 part prefix: $call $s") if isdbg('prefix');
+ $misses++;
+ $lru->put($call, \@try);
+ push @out, @try;
+ next;
+ }
+
+ # if the second part is a callsign and the last part is one letter
+ if (is_callsign($parts[1]) && length $parts[2] == 1) {
+ pop @parts;
+ }
+ }
+ }
+
+ # if it is a two parter
+ if (@parts == 2) {
+
+ # try it as it is as compound, taking the first part as the prefix
+ @nout = matchprefix($parts[0]);
+ if (@nout) {
+ my $s = join('/', $nout[0], $parts[1]);
+ my @try = get($s);
+ if (@try && $try[0] eq $s) {
+ dbg("got 2 part prefix: $call $s") if isdbg('prefix');
+ $misses++;
+ $lru->put($call, \@try);
+ push @out, @try;
+ next;
+ }
+ }
+ }
+
+ # remove the problematic /J suffix
+ pop @parts if @parts > 1 && $parts[$#parts] eq 'J';
+
+ # single parter
+ if (@parts == 1) {
+ @nout = matchprefix($parts[0]);
+ if (@nout) {
+ dbg("got prefix: $call = $nout[0]") if isdbg('prefix');
+ $misses++;
+ $lru->put($call, \@nout);
+ push @out, @nout;
+ next;
+ }
}
-# # now start to resolve it from the left hand end
-# for ($i = 1; $i <= length $sp; ++$i) {
- # now start to resolve it from the right hand end
- for ($i = length $sp; $i >= 1; --$i) {
- my @wout = get(substr($sp, 0, $i));
- last if @wout > 0 && $wout[0] gt $sp;
-# last if @wout == 0;
- push @out, @wout;
- last if @wout;
+
+ # try ALL the parts
+ my @checked;
+ my $n;
+L1: for ($n = 0; $n < @parts; $n++) {
+ my $sp = '';
+ my ($k, $i);
+ for ($i = $k = 0; $i < @parts; $i++) {
+ next if $checked[$i];
+ my $p = $parts[$i];
+ if (!$sp || length $p < length $sp) {
+ dbg("try part: $p") if isdbg('prefix');
+ $k = $i;
+ $sp = $p;
+ }
+ }
+ $checked[$k] = 1;
+ $sp =~ s/-\d+$//; # remove any SSID
+
+ # now start to resolve it from the right hand end
+ @nout = matchprefix($sp);
+
+ # try and search for it in the descriptions as
+ # a whole callsign if it has multiple parts and the output
+ # is more two long, this should catch things like
+ # FR5DX/T without having to explicitly stick it into
+ # the prefix table.
+
+ if (@nout) {
+ if (@parts > 1) {
+ $parts[$k] = $nout[0];
+ my $try = join('/', @parts);
+ my @try = get($try);
+ if (isdbg('prefix')) {
+ my $part = $try[0] || "*";
+ $part .= '*' unless $part eq '*' || $part eq $try;
+ dbg("Compound prefix: $try $part" );
+ }
+ if (@try && $try eq $try[0]) {
+ $misses++;
+ $lru->put($call, \@try);
+ push @out, @try;
+ } else {
+ $misses++;
+ $lru->put($call, \@nout);
+ push @out, @nout;
+ }
+ } else {
+ $misses++;
+ $lru->put($call, \@nout);
+ push @out, @nout;
+ }
+ next LM;
+ }
}
+
+ # we are a pirate!
+ @nout = matchprefix('Q');
+ $misses++;
+ $lru->put($call, \@nout);
+ push @out, @nout;
+ }
+
+ if (isdbg('prefixdata')) {
+ my $dd = new Data::Dumper([ \@out ], [qw(@out)]);
+ dbg($dd->Dumpxs);
}
return @out;
}
projects / spider.git / blobdiff