bug: not skipping about gtfs2gps

[pedro-andrade-inpe]

The problem occurs in filter_day_period, as it works properly for the whole saopaulo data. There are two bugs:

1. In filter_day_period(), line 31, it is

  gtfs$stops <- gtfs$stops[ stop_id %in% stop_id ]

and should be

  gtfs$stops <- gtfs$stops[ stop_id %in% unique_stops ]

isn´t it?

2. Filtering stop_times by time can create trips that have only one stop_time, which is inconsistent as a trip must have at least two stop_times. The implemented algorithm for gtfs2gps supposes each trip has at least two stop_times. Should such trips be removed by filter_day_period? Do you know an easy way to do that using data.table?

[rafapereirabr]

You're right. I've just fixed issue (1).

regarding issue (2), we should be able to discard trips with a single valid departure_time just as we already deal with 'invalid' trips with no valid departure_time.

[rafapereirabr]

note. the problem is in shape_id "52896", which has only one trip_id "8700-21-0"

[rafapereirabr]

I think I've fixed issue (2) with commit 34662d9.

The the number of stops (nstop) is smaller than 2, the gtfs2gps function should ignore that shape_id. There are probably better ways to go around this problem and report the shape_ids that are being ignored, but this should do the work for now.

    # Get the stops sequence with lat long linked to that route
    # each shape_id only has one stop sequence
    nstop <- gtfs_data$stop_times[trip_id %in% all_tripids, .N, by ="trip_id"]$N
    
    # If there is less than two valid stops/departure times, jump this shape_id
    if ( nstop < 2) {   return(NULL) }

[rafapereirabr]

Now fully fixed with

    # If there is less than two valid stops, jump this shape_id
    if( min(nstop) < 2){ return(NULL) } # nocov