path: root/bin/db-iris

#!perl
use strict;
use warnings;
use 5.014;
use utf8;

no if $] >= 5.018, warnings => 'experimental::smartmatch';

our $VERSION = '1.59';

use DateTime;
use DateTime::Format::Strptime;
use Encode qw(decode);
use Getopt::Long qw(:config no_ignore_case bundling);
use JSON;
use List::Util qw(first max);
use List::MoreUtils qw(none);
use Travel::Status::DE::IRIS;
use Travel::Status::DE::IRIS::Stations;

my ( $date, $time );
my $datetime       = DateTime->now( time_zone => 'Europe/Berlin' );
my $developer_mode = 0;
my $lookahead      = 2 * 60;
my $realtime       = 0;
my $with_related   = 1;
my $json_output    = 0;
my $use_cache      = 1;
my ( $schedule_cache, $realtime_cache );
my ( $filter_via,     $track_via, $status_via );
my ( @grep_class,     @grep_type, @grep_platform );
my ( %edata,          @edata_pre );

my @output;

binmode( STDOUT, ':encoding(utf-8)' );

@ARGV = map { decode( 'UTF-8', $_ ) } @ARGV;

GetOptions(
	'c|class=s@'         => \@grep_class,
	'd|date=s'           => \$date,
	'h|help'             => sub { show_help(0) },
	'l|lookahead=i'      => \$lookahead,
	'o|output=s@'        => \@edata_pre,
	'p|platform=s@'      => \@grep_platform,
	'r|realtime'         => \$realtime,
	't|time=s'           => \$time,
	'T|type=s'           => \@grep_type,
	'v|via=s'            => \$filter_via,
	'V|track-via=s'      => \$track_via,
	'x|exact|no-related' => sub { $with_related = 0 },
	'cache!'             => \$use_cache,
	'devmode'            => \$developer_mode,
	'json'               => \$json_output,
	'version'            => \&show_version,

) or show_help(1);

if ( @ARGV != 1 ) {
	show_help(1);
}

# opt=foo,bar support
@edata_pre     = split( qr{,}, join( q{,}, @edata_pre ) );
@grep_class    = split( qr{,}, join( q{,}, @grep_class ) );
@grep_platform = split( qr{,}, join( q{,}, @grep_platform ) );
@grep_type     = split( qr{,}, join( q{,}, @grep_type ) );

my ($station) = @ARGV;
$station = get_station($station);

if ($track_via) {
	$track_via = get_station($track_via);
}

if ($date) {
	my ( $day, $month, $year ) = split( qr{ [.] }x, $date );

	if ( $date eq 'tomorrow' ) {
		$datetime->add( days => 1 );
	}
	elsif (not( defined $day and defined $month )
		or ( $day < 1 )
		or ( $day > 31 )
		or ( $month < 1 )
		or ( $month > 12 ) )
	{
		say STDERR
"-d/--date: Please specify a valid date (dd.mm. / dd.mm.YYYY / tomorrow)";
		exit(3);
	}
	else {
		$datetime->set(
			day   => $day,
			month => $month,
			year  => $year || $datetime->year,
		);
	}
}
if ($time) {
	my ( $hour, $minute, $second ) = split( qr{ : }x, $time );

	if (   not defined $hour
		or not defined $minute
		or ( $hour < 0 )
		or ( $hour > 23 )
		or ( $minute < 0 )
		or ( $minute > 59 )
		or ( defined $second and ( ( $second < 0 ) or ( $second > 59 ) ) ) )
	{
		say STDERR "-t/--time: Please specify a valid time";
		exit(3);
	}

	$datetime->set(
		hour   => $hour,
		minute => $minute,
		second => $second || $datetime->second,
	);
}

for my $efield (@edata_pre) {
	given ($efield) {
		when ('a') { $edata{additional}   = 1 }
		when ('c') { $edata{canceled}     = 1 }
		when ('d') { $edata{delay}        = 1 }
		when ('D') { $edata{delays}       = 1 }
		when ('f') { $edata{fullroute}    = 1 }
		when ('m') { $edata{messages}     = 1 }
		when ('q') { $edata{qos}          = 1 }
		when ('r') { $edata{route}        = 1 }
		when ('R') { $edata{replacements} = 1 }
		when ('t') { $edata{times}        = 1 }
		when ('!') { $edata{debug}        = 1 }
		default    { $edata{$efield}      = 1 }
	}
}

if ($use_cache) {
	my $cache_path          = $ENV{XDG_CACHE_HOME} // "$ENV{HOME}/.cache";
	my $schedule_cache_path = "${cache_path}/db-iris-schedule";
	my $realtime_cache_path = "${cache_path}/db-iris-realtime";
	eval {
		use Cache::File;
		$schedule_cache = Cache::File->new(
			cache_root      => $schedule_cache_path,
			default_expires => '6 hours',
			lock_level      => Cache::File::LOCK_LOCAL(),
		);
		$realtime_cache = Cache::File->new(
			cache_root      => $realtime_cache_path,
			default_expires => '180 seconds',
			lock_level      => Cache::File::LOCK_LOCAL(),
		);
	};
	if ($@) {
		$schedule_cache = undef;
		$realtime_cache = undef;
	}
}

my $status = Travel::Status::DE::IRIS->new(
	datetime       => $datetime,
	developer_mode => $developer_mode,
	lookahead      => $lookahead,
	main_cache     => $schedule_cache,
	realtime_cache => $realtime_cache,
	station        => $station,
	with_related   => $with_related,
);
if ($track_via) {
	$status_via = Travel::Status::DE::IRIS->new(
		datetime       => $datetime,
		lookahead      => $lookahead + 3 * 60,
		main_cache     => $schedule_cache,
		realtime_cache => $realtime_cache,
		station        => $track_via,
	);
}

sub get_arrival {
	my ( $result, $fmt ) = @_;

	my $dt_arrival = $realtime ? $result->arrival : $result->sched_arrival;

	if ($fmt) {
		return $dt_arrival ? $dt_arrival->strftime($fmt) : q{};
	}
	return $dt_arrival;
}

sub get_departure {
	my ( $result, $fmt ) = @_;

	my $dt_dep = $realtime ? $result->departure : $result->sched_departure;

	if ($fmt) {
		return $dt_dep ? $dt_dep->strftime($fmt) : q{};
	}
	return $dt_dep;
}

sub get_station {
	my ($input_name) = @_;

	if ( $input_name =~ m{ ^ [[:digit:]]+ $ }x ) {
		return $input_name;
	}

	if ( $input_name
		=~ m{ ^ (?<lon> [[:digit:].]+ ) , (?<lat> [[:digit:].]+ ) }x )
	{
		my @candidates
		  = Travel::Status::DE::IRIS::Stations::get_station_by_location(
			$+{lon}, $+{lat} );

		if ( not @candidates ) {
			say STDERR
			  "Found no stations inside a 70km radius around $+{lon},$+{lat}";
			exit(1);
		}

		say STDERR "Geolocation candidates for $+{lon},$+{lat} are:";
		say STDERR join(
			"\n",
			map {
				sprintf( "%-30s %-5s  %4.1fkm",
					$_->[0][1], $_->[0][0], $_->[1] )
			} @candidates
		);
		exit(1);
	}

	my @stations = Travel::Status::DE::IRIS::Stations::get_station($input_name);

	if ( @stations == 0 ) {
		say STDERR "No station matches '$input_name'";
		exit(1);
	}
	elsif ( @stations == 1 ) {
		return $stations[0][0];
	}
	else {
		say STDERR "The input '$input_name' is ambiguous. Please choose one "
		  . 'of the following:';
		say STDERR
		  join( "\n", map { $_->[1] . ' (' . $_->[0] . ')' } @stations );
		exit(1);
	}
}

sub show_help {
	my ($code) = @_;

	print 'Usage: db-iris [-rx] [-d <date>] [-o <output-flags>]'
	  . '[-t <time>] [-v|-V <via>] [other options ...] <station>' . "\n"
	  . "See also: man db-iris\n";

	exit $code;
}

sub show_version {
	say "db-iris version ${VERSION}";

	exit 0;
}

sub sanitize_options {
	if ( $track_via and $edata{times} ) {
		say STDERR 'Note: --track-via cannot be combined with --output=times';
		say STDERR 'Disabling option --output=times';
		delete $edata{times};
	}
	if ( $realtime and $edata{times} ) {
		say STDERR 'Note: --realtime cannot be combined with --output=times';
		say STDERR 'Disabling option --realtime';
		$realtime = 0;
	}
	return;
}

sub format_delay {
	my ($d) = @_;
	my $delay = q{};

	if ( $d->delay ) {
		$delay = ( $d->delay > 0 ? ' +' : q{ } ) . $d->delay;
	}
	if ( $d->is_cancelled ) {
		$delay = ' CANCELED';
	}
	elsif ( $d->departure_is_cancelled ) {
		$delay .= ' ⊖';
	}
	elsif ( $d->start < $datetime and not $d->has_realtime ) {
		$delay = '  ?';
	}

	return $delay;
}

sub display_result {
	my (@lines) = @_;

	my @line_length;

	if ( not @lines ) {
		die("Nothing to show\n");
	}

	# the " !" suffixes might change a column's maximum line length, so we
	# need to add them before calculating it
	for my $line (@lines) {
		my $d = $line->[5];

		if ( not $edata{canceled} and $d->canceled_stops ) {
			$line->[3] .= q{ !};
		}

		if ( ( $d->platform // q{} ) ne ( $d->sched_platform // q{} ) ) {
			$line->[4] .= q{ !};
		}
	}

	for my $i ( 0 .. 4 ) {
		$line_length[$i] = max map { length( $_->[$i] ) } @lines;
	}

	for my $line (@lines) {

		my $d = $line->[5];

		printf(
			join( q{  }, ( map { "%-${_}s" } @line_length ) ),
			@{$line}[ 0 .. 4 ]
		);

		if (    $edata{delays}
			and $d->delay_messages )
		{
			printf( '  %s', join( q{  }, map { $_->[1] } $d->delay_messages ) );
		}
		if (    $edata{delay}
			and ( $d->delay or $d->is_cancelled )
			and $d->delay_messages )
		{
			printf( '  %s', ( $d->delay_messages )[-1]->[1] );
		}
		if ( $edata{qos} and $d->qos_messages ) {
			printf( '  %s', join( q{  }, map { $_->[1] } $d->qos_messages ) );
		}
		print "\n";

		if ( $edata{times} ) {
			if ( not defined $d->delay ) {
				print "\n";
			}
			elsif ( $d->delay == 0 and ( $d->arrival_delay // 0 ) == 0 ) {
				printf( "%s+0\n", q{ } x 15 );
			}
			else {
				printf(
					"%5s → %5s  %+d\n",
					$d->arrival   ? $d->arrival->strftime('%H:%M')   : q{},
					$d->departure ? $d->departure->strftime('%H:%M') : q{},
					$d->delay,
				);
			}

		}

		if ( $edata{debug} ) {
			if ( $d->{unk_ar_hi} ) {
				say "[DEBUG] arr:hi = $d->{unk_ar_hi}";
			}
			if ( $d->{unk_dp_hi} ) {
				say "[DEBUG] dep:hi = $d->{unk_dp_hi}";
			}
		}

		if ( $edata{messages} ) {
			for my $message ( $d->messages ) {

				# leading spaces to align with regular output
				printf( " %s  %s\n",
					$message->[0]->strftime('%d.%m. %H:%M'),
					$message->[1] );
			}
			print "\n";
		}

		if ( $edata{replacements} ) {
			for my $e ( $d->replaced_by ) {
				printf(
					"Ersatzzug: %s%s %s\n",
					$e->type, $e->line_no // q{},
					$e->train_no
				);
			}
			for my $e ( $d->replacement_for ) {
				printf(
					"Ersatzzug für: %s%s %s\n",
					$e->type, $e->line_no // q{},
					$e->train_no
				);
			}
		}

		if ( $edata{additional} and $d->additional_stops ) {
			printf( "Zusätzlicher Halt in: %s\n",
				join( q{, }, $d->additional_stops ) );
		}

		if ( $edata{canceled} and $d->canceled_stops ) {
			printf( "Ohne Halt in: %s\n", join( q{, }, $d->canceled_stops ) );
		}

		if ( $edata{fullroute} ) {
			print "\n"
			  . join( "\n", $d->route_pre )
			  . "\n - - - -\n"
			  . join( "\n", $d->route_post ) . "\n\n";
		}
	}

	return;
}

if ( my $err = $status->errstr ) {
	say STDERR "Request error: ${err}";
	exit 2;
}
if ( my $warning = $status->warnstr ) {
	say STDERR "Request warning: ${warning}";
	say STDERR '                 Information may be incomplete';
}

if ( $status_via and $status_via->errstr ) {
	my $err = $status_via->errstr;
	say STDERR "Request error in --via : ${err}";
	exit 2;
}
if ( $status_via and $status_via->warnstr ) {
	my $warning = $status_via->warnstr;
	say STDERR "Request warning in --via : ${warning}";
	say STDERR '                           Information may be incomplete';
}

sanitize_options();

if ($json_output) {
	say JSON->new->convert_blessed->encode( [ $status->results ] );
	exit 0;
}

for my $d ( $status->results() ) {

	my @via;

	# route may be incomplete, so check route_end as well
	@via = ( $d->route_post, $d->route_end );

	if (   ( $filter_via and not( first { $_ =~ m{$filter_via}io } @via ) )
		or ( @grep_class and none { $_ ~~ \@grep_class } $d->classes )
		or ( @grep_platform and not( $d->platform ~~ \@grep_platform ) )
		or ( @grep_type     and not( $d->type     ~~ \@grep_type ) )
		or $d->is_wing )
	{
		next;
	}

	my $delay = format_delay($d);

	my $platformstr = $d->platform // q{};
	my $timestr;
	if ($track_via) {
		$timestr = get_departure( $d, '%H:%M' ) || get_arrival( $d, '%H:%M' );
		if ( not $d->departure ) {
			next;
		}
		my $d_via = first {
			$_->train_id eq $d->train_id
			  or ( $_->old_train_id and $_->old_train_id eq $d->train_id );
		}
		$status_via->results;
		if (   not $d_via
			or not $d_via->sched_arrival
			or $d_via->sched_arrival < $d->departure )
		{
			next;
		}
		my $timestr_via = get_arrival( $d_via, '%H:%M' ) . $delay;
		$timestr .= ' → ' . $timestr_via;
		$platformstr = sprintf( '%2s → %2s', $d->platform // q{},
			$d_via->platform // q{} );
	}
	elsif ( $edata{times} ) {
		$timestr = sprintf( '%5s → %5s',
			$d->sched_arrival   ? $d->sched_arrival->strftime('%H:%M')   : q{},
			$d->sched_departure ? $d->sched_departure->strftime('%H:%M') : q{},
		);
	}
	else {
		$timestr
		  = ( get_departure( $d, '%H:%M' ) || get_arrival( $d, '%H:%M' ) )
		  . $delay;
	}

	push(
		@output,
		[
			$timestr, $d->train,
			$edata{route} ? join( q{  }, $d->route_interesting ) : q{},
			$d->route_end, $platformstr // q{}, $d
		]
	);

	my @processed_wings;
	for my $wing ( $d->departure_wings ) {
		my $wingdelay = format_delay($wing);
		push(
			@output,
			[
				'├'
				  . '─' x ( length($timestr) - 1 - length($delay) )
				  . $wingdelay,
				$wing->train,
				$edata{route} ? join( q{  }, $wing->route_interesting ) : q{},
				$wing->route_end,
				$platformstr // q{},
				$wing
			]
		);
		push( @processed_wings, $wing->wing_id );
	}
	for my $wing ( $d->arrival_wings ) {
		if ( not $wing->wing_id ~~ \@processed_wings ) {
			my $wingdelay = format_delay($wing);
			push(
				@output,
				[
					'├'
					  . '─' x ( length($timestr) - 1 - length($delay) )
					  . $wingdelay,
					$wing->train,
					$edata{route}
					? join( q{  }, $wing->route_interesting )
					: q{},
					$wing->route_end,
					$platformstr // q{},
					$wing
				]
			);
		}
	}
	if ( $d->departure_wings or $d->arrival_wings ) {
		substr( $output[-1][0], 0, 1 ) = '└';
	}
}

display_result(@output);

__END__

=head1 NAME

db-iris - Interface to the DeutscheBahn online departure monitor

=head1 SYNOPSIS

B<db-iris> [B<-rx>] [B<-d> I<date>] [B<-o> I<output-flags>]
[B<-t> I<time>] [B<-v>|B<-V> I<via>] [I<other options ...>] I<station>

=head1 VERSION

version 1.59

=head1 DESCRIPTION

db-iris is an interface to the DeutscheBahn departure monitor
available at L<https://iris.noncd.db.de/wbt/js/index.html>.

It requests all trains departing from (or arriving at) I<station> in the next
two hours and lists them on stdout.  I<station> can be a DS100 station code
(such as "EE"), a normal station name (such as "Essen Hbf" or "Dortmund
UniversitE<auml>t"), or an IBNR / european station number (such as 8000098). If
no exact match is found, B<db-iris> will try to find station names similar to
I<station>.

By default, db-iris shows the following data for each train:

=over

=item * scheduled departure time (see also B<-ot>, B<-r>).

=item * delay in minutes, cancellation, or a question mark (C<< ? >>)
indicating that no real-time data is available.

=item * train line or number.

=item * destination (see also B<-or>). An exclamation mark (C<< ! >>)
indicates that at least one stop has been cancelled (see B<-oc>).

=item * platform. An exclamation mark (C<< ! >>) indicates that it is not
the scheduled one.

=back

=head1 OPTIONS

=over

=item B<-c>, B<--class> I<classlist>

Comma-separated list of train classes to filter by. Using this option
causes all trains whose class is not in I<classlist> to be discarded.

Valid classes are:

    D    Non-DB train. Usually local transport
    F    "Fernverkehr", long-distance transport
    N    "Nahverkehr", local and regional transport
    S    S-Bahn, rather slow local/regional transport

=item B<-d>, B<--date> I<date>

Request results for I<date>, which is either a date string in in I<dd>.I<mm>.
or I<dd>.I<mm>.I<YYYY> format, or C<< tomorrow >>.  Note that typically only
slight (a few hours max) deviations from the current time are supported by the
IRIS backend, larger ones will not return data.

=item B<--json>

List results as JSON, see Travel::Status::DE::IRIS::Result(3pm) for a partial
documentation of arrival/departure keys. The B<--output> option has no effect
when using B<--json>.

Note that JSON entries not mentioned in Travel::Status::DE::IRIS::Result(3pm)
are NOT guaranteed to be compatible between releases. Their structure is not
part of the db-iris / Travel::Status::DE::IRIS versioning scheme; it may change
in backwards-incompatible ways anytime.

=item B<-l>, B<--lookahead> I<int>

Do not return results which are more than I<int> minutes in the future.
Defaults to 120 (2 hours).

Note that this is only an upper limit, not a guarantee to get every train
with a departure in less than I<int> minutes. This guarantee holds only for
I<int> below 120. However, any non-negative number is accepted for this
option.

=item B<--no-cache>

If the Cache::File module is available, server replies are cached in
F<~/.cache/db-iris-schedule> and F<~/.cache/db-iris-realtime> (or paths
relative to C<$XDG_CACHE_HOME>, if set). Use this option to disable caching
altogether. Note that this will significantly decrease db-iris responsiveness,
especially on mobile networks such as WifiOnICE.

=item B<-o>, B<--output> I<outputtypes>

For each result, output I<outputtypes> in addition to the normal time, delay,
line and destination information. I<outputtypes> is a comma-separated list,
this option may be repeated. Each output type has both a short and long form,
so both C<< -ot,d >> and C<< --output=times,delay >> are valid.

Valid output types are:

=over

=item a / additional

If a train's route deviates from its schedule: Print a list of additional
(unscheduled) stops it will serve.

=item c / canceled

If a train's route deviates from its schedule: Print a list of canceled
stops (scheduled stops which will not be served).

=item d / delay

If a train is delayed, show the most recent reason for this delay.

=item D / delays

List all delay reasons entered into the IRIS for each train, even if the
particular train is on time by now.

=item f / fullroute

Show the entire route (both before and after I<station>).

=item m / messages

List all messages (delay and qos) entered into the IRIS with timestamps.

=item q / qos

List all quality of service messages entered into the IRIS. These contain
information like "Missing carriage" or "Broken air conditioning".

Note that some qos messages may supersede older ones. superseded messages are
omitted, use the m / messages type to see those as well.

=item r / route

Show up to three stops between I<station> and the train's destination.

=item R / replacements

For cancelled trains: Print their replacement train(s), if present.
For unplanned trains: Print the train(s) they replace, if present.

=item t / times

Show both scheduled and expected arrival and departure times.

=back

=item B<-p>, B<--platforms> I<platforms>

Only show arrivals/departures at I<platforms> (comma-separated list, option may
be repeated).  This applies to actual departures, not schedules.

=item B<-r>, B<--realtime>

Show estimated instead of scheduled time where available. Cannot be combined
with C<< --output=times >>.

=item B<-t>, B<--time> I<time>

Request results for I<time> in HH:MM or HH:MM:SS format. Note that only
slight deviations (a few hours max) from the current time are supported by the
IRIS backend, larger ones will not return data.

=item B<-T>, B<--type> I<typelist>

Comma-separated list of train types to filter by. Using this option
causes all arrivals/departures whose type is not in I<typelist> to be discarded.

The following valid values are known:

    local transport:
    IRE  Inter-Regio Express (rare)
    RB   Regionalbahn (slower than RE)
    RE   Regional-Express
    S    S-Bahn

    regional/interregional transport:
    D    "Schnellzug" (generic fast train, rare)
    EC   Eurocity
    IC   Intercity
    IR   Inter-Regio (rare in Germany, mostly used in Switzerland)
    ICE  Intercity-Express
    THA  Thalys

Depending on the city and country, other types may be used as well. Examples
include "ABR" / "NWB" (private trains included in the local transport tariff
system), "HKX" (private train not included in any DB tariffs) and
"SBB" (unknown swiss train class)

=item B<-v>, B<--via> I<viastation>

Only show trains serving I<viastation> after I<station>. In this case,
I<viastation> must match the station as contained in the train's route
(see B<-of>), DS100 codes are not supported.

=item B<-V>, B<--track-via> I<viastation>

Only show trains serving I<viastation> after I<station>. Show result
timestamps as "HH:MM -> HH:MM +x", where the first time is the scheduled
departure (without delay) at I<station> and the second the scheduled arrival
(also without delay) at I<viastation>. If a delay is known, it will be indicated
by +x.

Note that here, I<viastation> must be a regular station name or DS100 code.

Caveat: Some trains may change their identity along the route. B<track-via> is
not able to handle those and will miss trains changing their identifier between
I<station> and I<viastation>

=item B<-x>, B<--exact>, B<--no-related>

Sometimes, Deutsche Bahn splits up major stations in their IRIS interface.
For instance, "KE<ouml>ln Messe/Deutz" actually consists of
"KE<ouml>ln Messe/Deutz" (KKDZ), "KE<ouml>ln Messe/Deutz Gl. 9-10"
(KKDZB) and "KE<ouml>ln Messe/Deutz (tief)" (KKDT).

By default, B<db-iris> will show departures for all of these stations when
queried for any of them. When this option is set, only the departures for the
station part specified on the commandline are shown.

=item B<--version>

Show version information.

=back

=head1 EXIT STATUS

=over

=item 0: Normal operation

=item 1: Invalid arguments or unknown station

=item 2: Cannot get departures from backend (network issues?)

=item 3: Invalid date/time specified

=back

=head1 CONFIGURATION

None.

=head1 DEPENDENCIES

=over

=item * Class::Accessor(3pm)

=item * DateTime(3pm)

=item * LWP::UserAgent(3pm)

=item * XML::LibXML(3pm)

=back

=head1 RECOMMENDS

=over

=item * Cache::File(3pm)

=back

=head1 BUGS AND LIMITATIONS

There are no known bugs at the moment.

=head1 AUTHOR

Copyright (C) 2013-2020 by Daniel Friesel E<lt>derf@finalrewind.orgE<gt>

The station data used by this script is provided by DB
Station&Service AG, Europaplatz 1, 10557 Berlin, Germany and available
under a CC-BY 4.0 license on
L<https://data.deutschebahn.com/dataset/data-haltestellen>.

=head1 LICENSE

This program is licensed under the same terms as Perl itself.