timezonefinderL

Warning

This package is deprecated. Use the TimezoneFinderL class of the timezonefinder package instead.

timezonefinderL is the faster and lightweight, but inaccurate version of the original timezonefinder. Use this package in favour of timezonefinder when memory usage and speed matter more to you than accuracy.

Only the function timezone_at() is being supported and numba cannot be used for precompilation. The commands need to modified:

pip install timezonefinderL

from timezonefinderL import TimezoneFinder

tf = TimezoneFinder()

longitude, latitude = 13.358, 52.5061
tf.timezone_at(lng=longitude, lat=latitude)  # returns 'Europe/Berlin'

For everything else please refer to the original Documentation.

Operating Principle

Instead of storing timezone polygons and checking which polygon a query point is included in, like with the vanilla timezonefinder, this package uses only the precomputed "shortcuts" to instantly lookup a timezone. Shortcuts are defined by splitting the coordinate system into rectangles. The zone which has the highest amount of timezone polygons (not covered surface!) within such a shortcut rectangle is instantly being returned.

This requires far less memory and computing time, but of course is not accurate close to the borders of two neighbouring timezones.

The size of the shortcut rectangles (<-> accuracy) is equal to the one used in the vanilla timezonefinder (1 shortcut per degree longitude, 2 per degree latitude, 260KB binary file size). In order to increase the accuracy (more and smaller shortcut rectangles), increment the parameters NR_SHORTCUTS_PER_LNG and NR_SHORTCUTS_PER_LAT in global_settings.py and compile a new binary shortcut file by running file_converter.py.

Speed Test Results:

obtained on MacBook Pro (15-inch, 2017), 2,8 GHz Intel Core i7 It can be seen that timezonefinderL is roughly one order of magnitude faster than timezonefinder:

Speed Tests:
-------------
"realistic points": points included in a timezone

in memory mode: False

testing 100000 realistic points
total time: 0.5513s
avg. points per second: 1.8 * 10^5

testing 100000 random points
total time: 0.5682s
avg. points per second: 1.8 * 10^5


in memory mode: True

testing 100000 realistic points
total time: 0.1688s
avg. points per second: 5.9 * 10^5


testing 100000 random points
total time: 0.1837s
avg. points per second: 5.4 * 10^5

Contact

If you notice that the tz data is outdated, encounter any bugs, have suggestions, criticism, etc. feel free to open an Issue, add a Pull Requests on Git or ...

contact me: [python] {-at-} [michelfe] {-*dot-} [it]*

License

timezonefinderL is distributed under the terms of the MIT license (see LICENSE.txt).

Also see: GitHub, PyPI, GUI and API of the outdated timezonefinderL timezonefinder,