X-Git-Url: http://dxcluster.org/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FPrefix.pm;h=a07ebdbdde8ff5dc8fd727bd5784829b1729001c;hb=f87323c2926605792ee02b84783d8f3d4dbd605f;hp=4e0b454698408a0e4764c0194f978a0f1d61d9d0;hpb=65744dd1e30165cd280502ee1a05c0c640023303;p=spider.git diff --git a/perl/Prefix.pm b/perl/Prefix.pm index 4e0b4546..a07ebdbd 100644 --- a/perl/Prefix.pm +++ b/perl/Prefix.pm @@ -8,85 +8,401 @@ package Prefix; -use Carp; +use IO::File; use DXVars; use DB_File; +use Data::Dumper; +use DXDebug; +use DXUtil; +use LRU; use strict; -my $db; # the DB_File handle -my %prefix_loc; # the meat of the info -my %prefix; # the prefix list +use vars qw($VERSION $BRANCH); +$VERSION = sprintf( "%d.%03d", q$Revision$ =~ /(\d+)\.(\d+)/ ); +$BRANCH = sprintf( "%d.%03d", q$Revision$ =~ /\d+\.\d+\.(\d+)\.(\d+)/ || (0,0)); +$main::build += $VERSION; +$main::branch += $BRANCH; + +use vars qw($db %prefix_loc %pre $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', $lrusize); sub load { - if ($db) { - untie %prefix; - %prefix = (); - %prefix_loc = (); - } - $db = tie(%prefix, "DB_File", undef, O_RDWR|O_CREAT, 0, $DB_BTREE) or confess "can't tie %prefix ($!)"; - do "$main::data/prefix_data.pl"; - confess $@ if $@; + # untie every thing + if ($db) { + undef $db; + untie %pre; + %pre = (); + %prefix_loc = (); + } + + # 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 $@; + + return $out; } sub store { - my ($k, $l); - my $fh = new FileHandle; - my $fn = "$main::data/prefix_data.pl"; + my ($k, $l); + my $fh = new IO::File; + my $fn = "$main::data/prefix_data.pl"; - confess "Prefix system not started" if !$db; + confess "Prefix system not started" if !$db; - # save versions! - rename "$fn.oooo", "$fn.ooooo" if -e "$fn.oooo"; - rename "$fn.ooo", "$fn.oooo" if -e "$fn.ooo"; - rename "$fn.oo", "$fn.ooo" if -e "$fn.oo"; - rename "$fn.o", "$fn.oo" if -e "$fn.o"; - rename "$fn", "$fn.o" if -e "$fn"; + # save versions! + rename "$fn.oooo", "$fn.ooooo" if -e "$fn.oooo"; + rename "$fn.ooo", "$fn.oooo" if -e "$fn.ooo"; + rename "$fn.oo", "$fn.ooo" if -e "$fn.oo"; + rename "$fn.o", "$fn.oo" if -e "$fn.o"; + rename "$fn", "$fn.o" if -e "$fn"; - $fh->open(">$fn") or die "Can't open $fn ($!)"; - - # prefix location data - $fh->print("%prefix_loc = (\n"); - foreach $l (sort {$a <=> $b} keys %prefix_loc) { - my $r = $prefix_loc{$l}; - $fh->printf(" $l => { name => '%s', dxcc => %d, itu => %d, utcoff => %d, lat => %f, long => %f },\n", - $r->{name}, $r->{dxcc}, $r->{itu}, $r->{cq}, $r->{utcoff}, $r->{lat}, $r->{long}); - } - $fh->print(");\n\n"); - - # prefix data - $fh->print("%prefix = (\n"); - foreach $k (sort keys %prefix) { - $fh->print(" '$k' => ["); - my @list = @{$prefix{$k}}; - my $l; - my $str; - foreach $l (@list) { - $str .= " $l,"; - } - chop $str; - $fh->print("$str ],\n"); - } - $fh->print(");\n"); - $fh->close; + $fh->open(">$fn") or die "Can't open $fn ($!)"; + + # prefix location data + $fh->print("%prefix_loc = (\n"); + foreach $l (sort {$a <=> $b} keys %prefix_loc) { + my $r = $prefix_loc{$l}; + $fh->printf(" $l => bless( { name => '%s', dxcc => %d, itu => %d, utcoff => %d, lat => %f, long => %f }, 'Prefix'),\n", + $r->{name}, $r->{dxcc}, $r->{itu}, $r->{cq}, $r->{utcoff}, $r->{lat}, $r->{long}); + } + $fh->print(");\n\n"); + + # prefix data + $fh->print("%pre = (\n"); + foreach $k (sort keys %pre) { + $fh->print(" '$k' => ["); + my @list = @{$pre{$k}}; + my $l; + my $str; + foreach $l (@list) { + $str .= " $l,"; + } + chop $str; + $fh->print("$str ],\n"); + } + $fh->print(");\n"); + undef $fh; + untie %pre; } -# this may return several entries, be warned! -# -# what you get is a list of pairs of:- +# what you get is a list that looks like:- # -# prefix => \[ @list of references to prefix_locs ] +# prefix => @list of blessed references to prefix_locs # # 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 $ref; + my $gotkey = $key; + return () if $db->seq($gotkey, $ref, R_CURSOR); + return () if $key ne substr $gotkey, 0, length $key; + + return ($gotkey, map { $prefix_loc{$_} } split ',', $ref); +} + +# +# get the next key that matches, this assumes that you have done a 'get' first +# +# +sub next +{ + my $key = shift; + my $ref; + my $gotkey; + return () if $db->seq($gotkey, $ref, R_NEXT); + return () if $key ne substr $gotkey, 0, length $key; + + return ($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 (); } +# +# extract a 'prefix' from a callsign, in other words the largest entity that will +# obtain a result from the prefix table. +# +# This is done by repeated probing, callsigns of the type VO1/G1TLH or +# G1TLH/VO1 (should) return VO1 +# + +sub extract +{ + my $calls = uc shift; + my @out; + my $p; + my @parts; + my ($call, $sp, $i); + +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: $call = " . join(' ', @parts)) if isdbg('prefix'); + + # remove any /0-9 /P /A /M /MM /AM suffixes etc + if (@parts > 1) { + @parts = grep { !/^\d+$/ && !/^[PABM]$/ && !/^(?:|AM|MM|BCN|JOTA|SIX|WEB|NET|Q\w+)$/; } @parts; + + # can we resolve them by direct lookup + 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'); + + # 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; + } + } + + # 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; +} + +my %valid = ( + lat => '0,Latitude,slat', + long => '0,Longitude,slong', + dxcc => '0,DXCC', + name => '0,Name', + itu => '0,ITU', + cq => '0,CQ', + utcoff => '0,UTC offset', + cont => '0,Continent', + ); + +no strict; +sub AUTOLOAD +{ + my $self = shift; + my $name = $AUTOLOAD; + + return if $name =~ /::DESTROY$/; + $name =~ s/.*:://o; + + confess "Non-existant field '$AUTOLOAD'" if !$valid{$name}; + # this clever line of code creates a subroutine which takes over from autoload + # from OO Perl - Conway + *{$AUTOLOAD} = sub {@_ > 1 ? $_[0]->{$name} = $_[1] : $_[0]->{$name}} ; + if (@_) { + $self->{$name} = shift; + } + return $self->{$name}; +} +use strict; + +# +# return a prompt for a field +# + +sub field_prompt +{ + my ($self, $ele) = @_; + return $valid{$ele}; +} 1; __END__