upoints is a collection of GPL v3 licensed modules for working with points on Earth, or other near spherical objects. It allows you to calculate the distance and bearings between points, mangle xearth/xplanet data files, work with online UK trigpoint databases, NOAA's weather station database and other such location databases.
Previous versions of upoints were called earth_distance, but the name was changed as it no longer reflected the majority of uses the packages was targeted at.
upoints does not strictly depend on any modules that aren't included in Python's standard library, and as such should run with Python 2.5 or newer [1]. If upoints doesn't work with the version of Python you have installed, drop me a mail and I'll endeavour to fix it.
However, upoints can make use of lxml if it is installed on systems that don't have the cElementTree module. If neither cElementTree nor lxml are installed it falls back to ElementTree, but will operate much slower on XML data.
The modules have been tested on many UNIX-like systems, including Linux, Solaris and OS X, but it should work fine on other systems too. The modules and scripts contain a large collection of doctest tests that can be checked with ./setup test_code, and the code examples in the documentation can be tested with ./setup test_doc.
[1] | If you still use Python v2.4 only small changes are required, for example to the base class definitions and the unrolling of conditional expressions. |
[2] | Some tests may fail due to rounding errors depending on the system the tests are being run on, but such instances should be obvious even to the casual user and some effort has been put in to reduce the likelihood of such failures. |
The simplest way to show how upoints works is by example, and here goes:
>>> from upoints import point
>>> Home = point.Point(52.015, -0.221)
>>> Telford = point.Point(52.6333, -2.5000)
>>> print("%d kM, with an initial bearing of %d°"
... % (Home.distance(Telford), Home.bearing(Telford)))
169 kM, with an initial bearing of 294°
All the class definitions, methods and independent functions contain hopefully useful usage examples in the docstrings. The html/ directory contains the preprocessed epydoc output for reference.
There is some accompanying text and examples for point.py, formerly edist.py, available in geolocation and path cross. More examples are available for xearth.py in xearth and path cross. Some background and more examples for trigpoints.py is online in Trigpointing and point.py. Usage examples for cities.py is available in Cities and cities.py. And finally, Pythons on a plane contains information on weather_stations.py. All the examples in the doc/ directory can be executed using ./setup.py test_doc.
The following people have submitted patches, testing and feedback:
- Cédric Dufour - edist.py's CSV import, and flight plan output
- Kelly Turner - Xearth import idea, and copious testing
- Simon Woods - Testing
API stability isn't guaranteed across versions, although frivolous changes won't be made.
When upoints 1.0 is released the API will be frozen, and any changes which aren't backwards compatible will force a major version bump.
The modules assume the caller will take care of significant digits, and output formatting [2]. All results are returned with whatever precision your Python install or system generates; unintuitive float representation, rounding errors, warts and all.
The reasoning is simple, the caller should always know what is required and any heuristics added to the code would be just that -- guesses, which can and will be wrong.
The upoints modules do not take flattening in to account, as in calculations based in most populated areas of the earth the errors introduced by ignoring the earth's flattening are quite small. Future versions may change if the limitation becomes an issue in real use.
Although not really a limitation one should also be careful to use data sources that are based around the same datum, and even within two data sources that use the same datum you should make sure they use the same representations. It isn't unusual to find data sources from the USA that specify longitudes west of Greenwich as positive for example.
[3] | A future release may include more standard output definitions, but there is no intention to add "magic" data mangling. |
Patches are most welcome, but I'd appreciate it if you could follow the guidelines below to make it easier to integrate your changes. These are guidelines however, and as such can be broken if the need arises or you just want to convince me that your style is better.
- PEP 8, the style guide, should be followed where possible.
- While support for Python versions prior to v2.5 may be added in the future if such a need were to arise, you are encouraged to use v2.5 features now.
- All new classes and methods should be accompanied by new doctest examples, and epydoc's epytext formatted descriptions if at all possible.
- Tests must not span network boundaries, see test.mock for workarounds.
- doctest tests in modules are only for unit testing in general, and should not rely on any modules that aren't in Python's standard library.
- Functional tests should be in the doc directory in reStructuredText formatted files, with actual tests in doctest blocks. Functional tests can depend on external modules, but they must be Open Source.
New examples for the doc directory are as appreciated as code changes.