Examples with StopMergeStrategy org.onebusaway.gtfs_merge.strategies.StopMergeStrategy used on opensource projects

Example 1 with StopMergeStrategy

use of org.onebusaway.gtfs_merge.strategies.StopMergeStrategy in project onebusaway-gtfs-modules by OneBusAway.

the class GtfsMergerTest method testAgencyPreference.

@Test
public void testAgencyPreference() throws IOException {
    // lowest priority feed (first) to highest priority feed (last)
    _oldGtfs.putLines("agency.txt", "agency_id,agency_name,agency_url,agency_timezone", "3,Pierce,http://p.us/,America/Los_Angeles");
    _oldGtfs.putLines("routes.txt", "route_id,route_short_name,route_long_name,route_type", "R11,11,The Eleven,3");
    _oldGtfs.putLines("stops.txt", "stop_id,stop_name,stop_lat,stop_lon", "100,The Stop,47.654403,-122.305211", "200,Pierce Other Stop,47.668594,-122.298859", "400,Pierce Only Stop,47.669563,-122.305420");
    _oldGtfs.putLines("calendars.txt", "service_id,monday,tuesday,wednesday,thursday,friday,saturday,sunday,start_date,end_date", "sid1,1,1,1,1,1,0,0,20110101,20111231");
    _oldGtfs.putLines("trips.txt", "route_id,service_id,trip_id", "R11,sid1,T11-0", "R11,sid1,T11-1");
    // stop conflict only
    _oldGtfs.putStopTimes("T11-0", "100,200");
    _oldGtfs.putStopTimes("T11-1", "100,400");
    _oldGtfs.putLines("stop_times.txt", "trip_id,stop_id,stop_sequence,arrival_time,departure_time", "T11-0,100,0,08:00:00,08:00:00", "T11-0,200,1,09:00:00,09:00:00", "T11-1,100,1,08:00:00,08:00:00", "T11-1,400,1,09:00:00,09:00:00");
    _newGtfs.putLines("agency.txt", "agency_id,agency_name,agency_url,agency_timezone", "1,Metro,http://metro.gov/,America/Los_Angeles", "3,Pierce,http://p.us/,America/Los_Angeles");
    _newGtfs.putLines("routes.txt", "agency_id,route_id,route_short_name,route_long_name,route_type", "1,R10,10,The Ten,3");
    _newGtfs.putLines("stops.txt", "stop_id,stop_name,stop_lat,stop_lon", "100,The Stop,47.654403,-122.305211", "200,The Other Stop,47.656303,-122.315436", "300,The Third Stop,47.668575,-122.283653");
    _newGtfs.putCalendars(1, "mask=1111100", "start_date=20120504", "end_date=20120608");
    _newGtfs.putLines("trips.txt", "route_id,service_id,trip_id", "R10,sid0,T10-0");
    _newGtfs.putLines("stop_times.txt", "trip_id,stop_id,stop_sequence,arrival_time,departure_time", "T10-0,100,0,08:00:00,08:00:00", "T10-0,200,1,09:00:00,09:00:00", "T10-0,300,1,10:00:00,10:00:00");
    _pugetGtfs = MockGtfs.create();
    _pugetGtfs.putLines("agency.txt", "agency_id,agency_name,agency_url,agency_timezone", "0,Puget Sound Region,http://puget-sound.gov/,America/Los_Angeles");
    _pugetGtfs.putLines("routes.txt", "route_id,route_short_name,route_long_name,route_type", "");
    _pugetGtfs.putLines("stops.txt", "stop_id,stop_name,stop_lat,stop_lon", "");
    _pugetGtfs.putLines("calendars.txt", "service_id,monday,tuesday,wednesday,thursday,friday,saturday,sunday,start_date,end_date", "");
    _pugetGtfs.putCalendars(1, "mask=1111100", "start_date=20120504", "end_date=20120608");
    _pugetGtfs.putLines("trips.txt", "route_id,service_id,trip_id", "");
    _pugetGtfs.putLines("stop_times.txt", "trip_id,stop_id,stop_sequence,arrival_time,departure_time", "");
    AgencyMergeStrategy agencyStrategy = new AgencyMergeStrategy();
    agencyStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    _merger.setAgencyStrategy(agencyStrategy);
    TripMergeStrategy tripStrategy = new TripMergeStrategy();
    tripStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    _merger.setTripStrategy(tripStrategy);
    StopMergeStrategy stopStrategy = new StopMergeStrategy();
    stopStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    stopStrategy.setDuplicateRenamingStrategy(EDuplicateRenamingStrategy.AGENCY);
    stopStrategy.setLogDuplicatesStrategy(ELogDuplicatesStrategy.WARNING);
    _merger.setStopStrategy(stopStrategy);
    GtfsRelationalDao dao = merge();
    // pierce is included twice, it should not show up as a duplicate
    assertTrue(dao.getAllAgencies().size() == 3);
    for (Trip trip : dao.getAllTrips()) {
        assertTrue(dao.getStopTimesForTrip(trip).size() > 0);
    }
    boolean pugetStopFound = false;
    for (Stop stop : dao.getAllStops()) {
        if ("0".equals(stop.getId().getAgencyId())) {
            pugetStopFound = true;
        }
    }
    assertTrue("expect a puget stop", pugetStopFound);
}

Example 2 with StopMergeStrategy

use of org.onebusaway.gtfs_merge.strategies.StopMergeStrategy in project onebusaway-gtfs-modules by OneBusAway.

the class GtfsMergerTest method testRenameStrategy.

@Test
public void testRenameStrategy() throws IOException {
    // lowest priority feed (first) to highest priority feed (last)
    _oldGtfs.putLines("agency.txt", "agency_id,agency_name,agency_url,agency_timezone", "3,Pierce,http://p.us/,America/Los_Angeles");
    _oldGtfs.putLines("routes.txt", "route_id,route_short_name,route_long_name,route_type", "R10,10,The Pierce Ten,3");
    _oldGtfs.putLines("stops.txt", "stop_id,stop_name,stop_lat,stop_lon", "100,The Stop,47.654403,-122.305211", "200,Pierce Other Stop,47.668594,-122.298859", "400,Pierce Only Stop,47.669563,-122.305420");
    _oldGtfs.putLines("calendars.txt", "service_id,monday,tuesday,wednesday,thursday,friday,saturday,sunday,start_date,end_date", "sid0,1,1,1,1,1,0,0,20110101,20111231");
    _oldGtfs.putLines("trips.txt", "route_id,service_id,trip_id", "R10,sid0,T10-0", "R10,sid0,T10-1");
    // stop conflict only
    _oldGtfs.putStopTimes("T10-0", "100,200");
    _oldGtfs.putStopTimes("T10-1", "100,400");
    _oldGtfs.putLines("stop_times.txt", "trip_id,stop_id,stop_sequence,arrival_time,departure_time", "T10-0,100,0,08:00:00,08:00:00", "T10-0,200,1,09:00:00,09:00:00", "T10-1,100,1,08:00:00,08:00:00", "T10-1,400,1,09:00:00,09:00:00");
    _newGtfs.putLines("agency.txt", "agency_id,agency_name,agency_url,agency_timezone", "1,Metro,http://metro.gov/,America/Los_Angeles", "3,Pierce,http://p.us/,America/Los_Angeles");
    _newGtfs.putLines("routes.txt", "agency_id,route_id,route_short_name,route_long_name,route_type", "1,R10,10,The KCM Ten,3");
    _newGtfs.putLines("stops.txt", "stop_id,stop_name,stop_lat,stop_lon", "100,The Stop,47.654403,-122.305211", "200,The Other Stop,47.656303,-122.315436", "300,The Third Stop,47.668575,-122.283653");
    _newGtfs.putCalendars(1, "mask=1111100", "start_date=20120504", "end_date=20120608");
    _newGtfs.putLines("trips.txt", "route_id,service_id,trip_id", "R10,sid0,T10-0");
    _newGtfs.putLines("stop_times.txt", "trip_id,stop_id,stop_sequence,arrival_time,departure_time", "T10-0,100,0,08:00:00,08:00:00", "T10-0,200,1,09:00:00,09:00:00", "T10-0,300,1,10:00:00,10:00:00");
    _pugetGtfs = MockGtfs.create();
    _pugetGtfs.putLines("agency.txt", "agency_id,agency_name,agency_url,agency_timezone", "0,Puget Sound Region,http://puget-sound.gov/,America/Los_Angeles");
    _pugetGtfs.putLines("routes.txt", "route_id,route_short_name,route_long_name,route_type", "");
    _pugetGtfs.putLines("stops.txt", "stop_id,stop_name,stop_lat,stop_lon", "");
    _pugetGtfs.putLines("calendars.txt", "service_id,monday,tuesday,wednesday,thursday,friday,saturday,sunday,start_date,end_date", "");
    _pugetGtfs.putCalendars(1, "mask=1111100", "start_date=20120504", "end_date=20120608");
    _pugetGtfs.putLines("trips.txt", "route_id,service_id,trip_id", "");
    _pugetGtfs.putLines("stop_times.txt", "trip_id,stop_id,stop_sequence,arrival_time,departure_time", "");
    AgencyMergeStrategy agencyStrategy = new AgencyMergeStrategy();
    agencyStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    _merger.setAgencyStrategy(agencyStrategy);
    TripMergeStrategy tripStrategy = new TripMergeStrategy();
    tripStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    tripStrategy.setDuplicateRenamingStrategy(EDuplicateRenamingStrategy.AGENCY);
    _merger.setTripStrategy(tripStrategy);
    StopMergeStrategy stopStrategy = new StopMergeStrategy();
    stopStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    stopStrategy.setDuplicateRenamingStrategy(EDuplicateRenamingStrategy.AGENCY);
    stopStrategy.setLogDuplicatesStrategy(ELogDuplicatesStrategy.WARNING);
    _merger.setStopStrategy(stopStrategy);
    RouteMergeStrategy routeStrategy = new RouteMergeStrategy();
    routeStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    routeStrategy.setDuplicateRenamingStrategy(EDuplicateRenamingStrategy.AGENCY);
    _merger.setRouteStrategy(routeStrategy);
    ServiceCalendarMergeStrategy serviceStrategy = new ServiceCalendarMergeStrategy();
    serviceStrategy.setDuplicateDetectionStrategy(EDuplicateDetectionStrategy.FUZZY);
    serviceStrategy.setDuplicateRenamingStrategy(EDuplicateRenamingStrategy.AGENCY);
    _merger.setServiceCalendarStrategy(serviceStrategy);
    GtfsRelationalDao dao = merge();
    // pierce is included twice, it should not show up as a duplicate
    assertTrue(dao.getAllAgencies().size() == 3);
    for (Trip trip : dao.getAllTrips()) {
        String tripId = trip.getId().getId();
        // AGENCY renaming strategy
        assertTrue(!tripId.matches("^[a-j]-.*"));
        assertTrue(dao.getStopTimesForTrip(trip).size() > 0);
    }
    boolean pugetStopFound = false;
    for (Stop stop : dao.getAllStops()) {
        if ("0".equals(stop.getId().getAgencyId())) {
            pugetStopFound = true;
        }
        String stopId = stop.getId().getId();
        // AGENCY renaming strategy
        assertTrue(!stopId.matches("^[a-j]-.*"));
    }
    for (Route route : dao.getAllRoutes()) {
        String routeId = route.getId().getId();
        // AGENCY renaming strategy
        assertTrue(!routeId.matches("^[a-j]-.*"));
    }
    for (ServiceCalendar service : dao.getAllCalendars()) {
        String serviceId = service.getServiceId().getId();
        assertTrue(!serviceId.matches("^[a-j]-.*"));
    }
    assertTrue("b-sid0".matches("[a-j]-.*"));
    assertTrue("expect a puget stop", pugetStopFound);
}