Snap points to a line string keeping a given order or spacing intact.
The PySnapping python library helps to solve the problem of snapping an ordered sequence of points to a line string, respecting the given input order.
The motivation for this library is to solve the problem of snapping stops onto vehicle trajectories in the GTFS format with inaccurate or missing kilometrage.
In order to be able to work with metric parameters and to treat data with large extent located anywhere on the world, we use the EPSG:4978 Cartesian 3D geocentric coordinate system for the internal representation.
The library aims to automatically classify kilometrage information as trusted or untrusted. Trusted points are always snapped as given by the kilometrage. In between trusted points, untrusted points are snapped by minimizing the sum of square snapping distances among all admissible solutions within certain radii. Admissible solutions are those that respect the order and minimum spacing between points.
Install the latest stable release from PyPI with pip:
pip install pysnapping
The pysnapping.linear_referencing
module contains low-level classes and functions for referencing points along
linear features in N-dimensions built on top of numpy
. This contains interpolation/extrapolation along a line string
and data structures optimized for projecting points to substrings of a line string.
The pysnapping.util
module contains common helper functions used in other parts of the library.
The pysanpping.snap
module is the main entry point for users of the pysnapping
library and
provides the classes needed to use the library.
The typical usage pattern is to create a pysnapping.snap.DubiousTrajectory
instance which represents
a vehicle trajectory with dubious kilometrage and a pysnapping.snap.DubiousTrajectoryTrip
dtrip
which represents
a trip along such a trajectory with dubious kilometrage for the stops.
Then dtrip.to_trajectory_trip
can be used to get a pysnapping.snap.TrajectorTrip
trip
with well defined
metric kilometrage. trip.trajectory
then refers to a pysnapping.snap.Trajectory
instance with well defined metric
kilometrage.
The trip.snap_trip_points
method can be used to snap the trip points onto the
trajectory, resulting in a pysnapping.snap.SnappedTripPoints
instance snapped
. Then
you can e.g. split the trajectory at the stops using the
snapped.get_inter_point_ls_coords_in_travel_direction
method. Snapping can be
controlled with parameters given by a pysnapping.snap.SnappingParams
instance.
A more detailed usage example (e.g. how to process GTFS input) is planned but not available yet. Until then, please also check the docstrings and source code for additional usage hints/possibilities.
Please use the GitHub issue tracker to report bugs/issues.
If you want to contribute to the pysnapping library, you can make a pull request at GitHub. Before working on major features/changes, please consider contacting us about your plans. See our GitHub page for contact details.
Clone this repo and enter the corresponding directory. Create a virtual environment, then install frozen requirements, dev-requirements and this library in editable mode:
python3.9 -m venv env
. env/bin/activate
pip install -U pip
pip install -r requirements.txt -r dev-requirements.txt -e .
Keep env activated for all following instructions.
Enable pre-commit hooks:
pre-commit install
From time to time (not automated yet) run
pre-commit autoupdate
to update frozen revs.
Run tests and analyze code coverage:
pytest --cov=pysnapping --cov-report term --cov-fail-under=85 pysnapping
- Snapping for untrusted points is now done by an exact algorithm instead of the
iterative approximate solution. This implies breaking changes to the parameters in
snap.SnappingParams
as well as to parts of the API of the classes in thesnap
module. For example: Timing information to guide the initial conditions of the iterative solution is not necessary and thus not supported any more. - The
ordering
module has been removed since it is not needed any more.
- The development toolchain now uses ruff where possible.