org.opentripplanner.transit

Class TransitLayer

java.lang.Object

org.opentripplanner.transit.TransitLayer

All Implemented Interfaces:

Serializable

public class TransitLayer
extends Object
implements Serializable

A key simplifying factor is that we don't handle overnight trips. This is fine for analysis at usual times of day.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
double	centerLat
double	centerLon
StreetLayer	linkedStreetLayer A transitLayer can only be linked to one StreetLayer, otherwise the street indexes for the transit stops would be ambiguous.
int	nTrips
List<gnu.trove.list.TIntList>	patternsForStop
List<com.conveyal.gtfs.model.Service>	services
List<com.conveyal.gtfs.model.Stop>	stopForIndex
gnu.trove.map.TIntIntMap	stopForStreetVertex
List<String>	stopIdForIndex
gnu.trove.map.TIntIntMap[]	stopTree this is the result of running a search from every stop in the graph, map from stop vertex index to (vertex index, distance)
gnu.trove.list.TIntList	streetVertexForStop
List<gnu.trove.list.TIntList>	transfersForStop
List<TripPattern>	tripPatterns
static int	TYPICAL_NUMBER_OF_STOPS_PER_TRIP

Constructor Summary

Constructors

Constructor and Description
TransitLayer()

Method Summary

Modifier and Type	Method and Description
void	buildStopTree()
static TransitLayer	fromGtfs(String file)
BitSet	getActiveServicesForDate(org.joda.time.LocalDate date)
int	getStopCount()
void	loadFromGtfs(com.conveyal.gtfs.GTFSFeed gtfs) Seems kind of hackish to pass the street layer in.
void	rebuildTransientIndexes() (Re-)build transient indexes of this TripPattern, connecting stops to patterns etc.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

stopIdForIndex

public List<String> stopIdForIndex

TYPICAL_NUMBER_OF_STOPS_PER_TRIP

public static final int TYPICAL_NUMBER_OF_STOPS_PER_TRIP

See Also:

Constant Field Values

tripPatterns

public List<TripPattern> tripPatterns

streetVertexForStop

public gnu.trove.list.TIntList streetVertexForStop

stopForStreetVertex

public transient gnu.trove.map.TIntIntMap stopForStreetVertex

transfersForStop

public List<gnu.trove.list.TIntList> transfersForStop

patternsForStop

public List<gnu.trove.list.TIntList> patternsForStop

services

public List<com.conveyal.gtfs.model.Service> services

stopForIndex

public transient List<com.conveyal.gtfs.model.Stop> stopForIndex

centerLon

public double centerLon

centerLat

public double centerLat

nTrips

public int nTrips

stopTree

public transient gnu.trove.map.TIntIntMap[] stopTree

this is the result of running a search from every stop in the graph, map from stop vertex index to (vertex index, distance)

linkedStreetLayer

public StreetLayer linkedStreetLayer

A transitLayer can only be linked to one StreetLayer, otherwise the street indexes for the transit stops would be ambiguous. It can however be linked to no StreetLayer. So if this field is null there are no known streets, but if this field is set then this TransitLayer has already been linked to a StreetLayer. This field is only public because it has to be set from StreetLayer, which is in another package.

Constructor Detail

TransitLayer

public TransitLayer()

Method Detail

loadFromGtfs

public void loadFromGtfs(com.conveyal.gtfs.GTFSFeed gtfs)

Seems kind of hackish to pass the street layer in. Maybe there should not even be separate street and transit layer classes. Should we have separate stop objects in the transit layer and stop vertices in the street layer? It should be possible to load multiple GTFS feeds into the same transit layer. This might be achieved by loading more than one feed into a MapDB, but it's probably better to keep the one MapDB == one feed equivalence. That way when individual GTFS feeds are updated, we only need to reload one MapDB, there are no identifier collision problems within one MapDB, and we don't need to store which feed each entity belongs to in the MapDB. So loading multiple feeds would be achieved by calling a function on TransitLayer multiple times with multiple GtfsFeed objects. All maps/lists would need to be initialized with empty objects at construction, and all methods would have to be designed to be called multiple times in successtion.

rebuildTransientIndexes

public void rebuildTransientIndexes()

(Re-)build transient indexes of this TripPattern, connecting stops to patterns etc.

buildStopTree

public void buildStopTree()

fromGtfs

public static TransitLayer fromGtfs(String file)

getStopCount

public int getStopCount()

getActiveServicesForDate

public BitSet getActiveServicesForDate(org.joda.time.LocalDate date)