filter_area() selects trips serving a polygon or bounding box. Crops check
endpoint times, shift frequency windows, and prune dependent records.
compare_scenarios() compares scheduled supply by route or network on one
service date, including vehicle-hours, vehicle-kilometres and active fleet.
Trip, route and segment speeds and distances use the same geometry. Detailed distances identify each trip and originating stop sequence. Inferred shapes and their stop measures use metres.
repair_gtfs() supports explicit overnight and stop-time repairs.
repair_report() identifies changed cells and original rows. Use
repair = "none" to disable reference cleanup during import.
tidy_raptor() prepares routing tables internally. Date and time filtering
limits frequency expansion before routing. Its public arguments are unchanged.
Barcelona regression tests exposed and fixed missing parent stations after filtering, which prevented filtered feeds from being reimported or merged.
Frequency-based departures now contribute correctly to calendar counts, service spans, vehicle-hours, planning indicators, and time filtering. System frequency sums routes and directions before averaging over patterns; weekday-hour averages include zero departures on active dates.
The explorer now uses OpenStreetMap tiles and keeps empty-corridor notices in the plot rather than repeating warnings in the console.
Added the embedded bcn_bus_gtfs dataset from the supplied AMB Mobilitat
archive es-barcelona-autos-castellbisbal-gtfs-892.zip: six agencies,
128 routes, 25,122 retained trips, and 3,385 stops. The dataset uses xz
compression; import excludes 48 trips with fewer than two stop-time records.
All supplied calendar years (2017-2026) are retained.
Stop-level headways process one service pattern at a time, and stop frequencies and dwell summaries aggregate services before expanding patterns, reducing memory use on feeds with many patterns. Services without active dates are omitted from service-pattern analyses rather than producing missing-weight warnings.
plot_calendar() adds max_patterns to bound categorical legends without
removing dates. Less frequent patterns share an "Other service patterns" color.
Route-frequency, headway, trip-duration, and service-supply plots also accept
max_patterns, defaulting to the 12 most frequent patterns; Inf shows all.
Selecting the full date range in the explorer preserves every source trip; date comparisons no longer depend on the internal Date storage type. The scheduled-hours slider rounds its upper bound up to include the final hour.
read_gtfs() and write_gtfs() now open native file-selection windows when
their paths are omitted in interactive R sessions. Output filenames receive a
.zip extension when needed.
read_gtfs() converts UTF-16LE and UTF-16BE files to UTF-8 in a temporary
copy, preserving importer options and leaving the source archive unchanged.
A conversion warning is shown when quiet = FALSE.
Added validate_gtfs() using the MobilityData Canonical GTFS Schedule
Validator, with an optional SHA-256-verified download cached outside the
package. It prints a summary, returns findings, and optionally saves HTML and
JSON reports for a ZIP archive or exported feed object.
Time editing functions wrap extreme edited times into the next service-day range while keeping stop times non-decreasing.
split_trip() accepts stop IDs as an alternative to an equal-part split;
boundary stops are included in both adjacent parts.
Refreshed the README figures and package website from the current plotting code, and added a reproducible figure-generation script for maintainers.
Improved the year-faceted service calendar for dense feeds and retained the
supplied object name in the explore_gtfs() dashboard.
Organized the package website by workflow, added task-based function reference sections, and added vignettes for getting started, service analysis, filtering and editing, and the interactive dashboard.
Standardized public method values with underscores: by_trip, by_route,
by_hour, by_stop, and by_shape. Legacy dotted values remain accepted.
Standardized public argument names including file_path, gtfs_x,
gtfs_y, max_dwelltime, and min_length, with compatibility aliases for
previous names.
Improved tidy_raptor() validation, transfer handling, date selection, and
output column names.
Added planning indicators to explore_gtfs(), including scheduled headway,
commercial speed, service span, vehicle-hours, peak fleet, stop spacing, and
route connectivity.
Calendar and service-pattern summaries now represent dates with no active
service explicitly: trip counts are 0, and the service pattern is labeled
"No service".
plot_routefrequency() now uses a route-hour-service pattern tile plot and
limits the default display to the busiest routes through top_n, improving
readability for large feeds.
explore_gtfs() adds a route-count control for route-frequency plots and
keeps filtered service and service-pattern selections aligned.
create_gtfs() introduced.
CRAN resubmission release following archival caused by the archived
hrbrthemes dependency. The dependency has been removed, and the package
dependency set has been substantially reduced.
GTFS editing and selection
filter_stop() and filter_time() now preserve partial trips for
experimentation instead of restoring complete stop sequences.selection() now supports grouping columns and computed groups in a style
similar to dplyr::group_by(), while continuing to support logical and
spatial selections without altering GTFS tables.Plotting and dashboard
-Inf values when
shape_dist_traveled contains only missing values.explore_gtfs() removes duplicated system information, assigns distinct
route colors, and filters non-finite plotting values.Package quality
Now featuring travel time matrices, trip speed and dwell-time editing, corridors, hubs, and more!
New functions
New methods
Bug fix & improvements
trips argument.