Unverified Commit 20990f6e authored by Birte Kristina Friesel's avatar Birte Kristina Friesel
Browse files

xml2json: O(n²) → O(n). hashtables sure are helpful!

parent f15c05d8

share/xml2json

+62 −51
Original line number Diff line number Diff line
@@ -13,6 +13,24 @@ my $json_str = read_file('stations.json'); 
my $stations = JSON->new->utf8->decode($json_str);

@{$stations} = sort { $a->{name} cmp $b->{name} } @{$stations};



my %station_by_ds100;

for my $station ( @{$stations} ) {

	$station_by_ds100{ $station->{ds100} } = $station;

}



my %station_by_eva;

for my $station ( @{$stations} ) {

	$station_by_eva{ $station->{eva} } = $station;

}



my %station_by_name;

for my $station ( @{$stations} ) {

	$station_by_name{ $station->{name} } = $station;

}



my %xml_by_name;

my %xml_by_eva;



my $xml_str = read_file('stations.xml');

my $tree    = XML::LibXML->load_xml( string => $xml_str );
@@ -31,46 +49,51 @@ for my $station ( $tree->findnodes('//station') ) { 
	my $ds100 = $station->getAttribute('ds100');

	my $is_db = $station->getAttribute('db') eq 'true';



	my $found = 0;

	my $xml_station = {

		name  => $name,

		eva   => $eva,

		ds100 => $ds100,

		is_db => $is_db,

	};

	$xml_by_name{$name} = $xml_station;

	$xml_by_eva{$eva}   = $xml_station;



	for my $j_station ( @{$stations} ) {

		my $j_name  = $j_station->{name};

		my $j_ds100 = $j_station->{ds100};

		my $j_eva   = $j_station->{eva};

	my $found = 0;



		if ( $name eq $j_name or $eva == $j_eva ) {

	if ( $station_by_name{$name} or $station_by_eva{$eva} ) {

		$found = 1;

	}



		if ( $j_ds100 eq $ds100 and $j_name ne $name ) {

	if (    $station_by_ds100{$ds100}

		and $station_by_ds100{$ds100}{name} ne $name )

	{

		printf( "%8s has been renamed: %30s -> %30s\n",

				$ds100, $j_name, $name );

			$ds100, $station_by_ds100{$ds100}{name}, $name );



			#$j_station->{name} = $name;

			last;

		#$station_by_ds100{$ds100}{name} = $name;

	}

		elsif ( $j_eva == $eva and $j_name ne $name ) {

			printf( "%d mismatch: (%s -> %s), (%s -> %s)\n",

				$eva, $j_name, $name, $j_ds100, $ds100 );

			last;

	elsif ( $station_by_eva{$eva} and $station_by_eva{$eva}{name} ne $name ) {

		printf(

			"%d mismatch: (%s -> %s), (%s -> %s)\n",

			$eva,  $station_by_eva{$eva}{name},

			$name, $station_by_eva{$eva}{ds100}, $ds100

		);

	}

		elsif ( $j_name eq $name

			and $j_ds100 ne $ds100

	elsif ( $station_by_name{$name}

		and $station_by_name{$name}{ds100} ne $ds100

		and $is_db

		and $ds100 !~ m{ ^ PQ }x )

	{

		printf( "%30s has been recoded: %8s -> %8s\n",

				$name, $j_ds100, $ds100 );

			last;

			$name, $station_by_name{$name}{ds100}, $ds100 );

	}

		elsif ( $j_name eq $name

			and $j_eva != $eva

	elsif ( $station_by_name{$name}

		and $station_by_name{$name}{eva} ne $eva

		and $is_db

		and $ds100 !~ m{ ^ PQ }x )

	{

			printf( "%30s has been recoded: %d -> %d\n", $name, $j_eva, $eva );

			last;

		}

		printf( "%30s has been recoded: %d -> %d\n",

			$name, $station_by_name{$name}{eva}, $eva );

	}



	if ( not $found
@@ -96,19 +119,7 @@ for my $i ( 0 .. $#{$stations} ) { 
	my $j_ds100   = $j_station->{ds100};

	my $j_eva     = $j_station->{eva};



	my $found = 0;



	for my $station ( $tree->findnodes('//station') ) {

		my $name  = $station->getAttribute('name');

		my $eva   = $station->getAttribute('eva');

		my $ds100 = $station->getAttribute('ds100');

		my $is_db = $station->getAttribute('db') eq 'true';

		if ( $name eq $j_name or $eva == $j_eva ) {

			$found = 1;

		}

	}



	if ( not $found ) {

	if ( not( $xml_by_name{$j_name} or $xml_by_eva{$j_eva} ) ) {

		say "station no longer exists: $j_eva  $j_ds100  \"$j_name\"";

		unshift( @to_delete, $i );

	}