tzfpy 1.3.0

Probably the fastest Python package to convert longitude/latitude to timezone name

tzfpy

• 
• 
• 
• 
• 
• 

[!NOTE]

0. It's probably the fastest Python package to convert longitude/latitude to timezone name.
1. This package use a simplified polygon data and not so accurate around borders.
2. Rust use lazy init, so first calling will be a little slow.
3. Use about 120MB memory.
4. It's tested under Python 3.10+.
5. Try it online:
   • https://ringsaturn.github.io/tzf-web/, powered by tzf-rs and WebAssembly

Usage

Please note that new timezone names may be added to tzfpy, which could be incompatible with old version package like pytz or tzdata. As an option, tzfpy supports install compatible version of those packages with extra params.

# Install just tzfpy
pip install tzfpy

# Install with pytz
pip install "tzfpy[pytz]"

# Install with tzdata. https://github.com/python/tzdata
pip install "tzfpy[tzdata]"

# Install via conda, see more in https://github.com/conda-forge/tzfpy-feedstock
conda install -c conda-forge tzfpy

>>> from tzfpy import get_tz, get_tzs
>>> get_tz(116.3883, 39.9289)  # in (longitude, latitude) order.
'Asia/Shanghai'
>>> get_tzs(87.4160, 44.0400)  # in (longitude, latitude) order.
['Asia/Shanghai', 'Asia/Urumqi']

Index mode env vars

tzfpy follows current tzf-rs behavior: DefaultFinder enables y_stripes by default. If you need to disable y_stripes, use this environment variable:

export _TZFPY_DISABLE_Y_STRIPES=1

The index requires about 120MB memory, but can speed up query missing from pre-index, especially around borders.

Export to GeoJSON

For data visualization, you can get timezone polygon GeoJSON data from tzfpy:

from tzfpy import get_tz, get_tz_index_geojson, get_tz_polygon_geojson

lng = -74.0060
lat = 40.7128
tz = get_tz(lng, lat)
print(f"Timezone for ({lng}, {lat}): {tz}")

with open("tz_nyc_polygon.geojson", "w") as f:
    geojson_data = get_tz_polygon_geojson(tz)
    f.write(geojson_data)

with open("tz_nyc_index.geojson", "w") as f:
    geojson_data = get_tz_index_geojson(tz)
    f.write(geojson_data)

Best practices

1. Always install tzfpy with tzdata extra: pip install tzfpy[tzdata]

2. Use Python's zoneinfo package(import zoneinfo, aka tzdata in PyPI) to handle timezone names, even if you are using arrow:

   examples/tzfpy_with_datetime.py:

   from datetime import datetime, timezone
   from zoneinfo import ZoneInfo
   
   from tzfpy import get_tz
   
   tz = get_tz(139.7744, 35.6812)  # Tokyo
   
   now = datetime.now(timezone.utc)
   now = now.replace(tzinfo=ZoneInfo(tz))
   print(now)
   # 2025-04-29 01:33:56.325194+09:00
   

   examples/tzfpy_with_arrow.py:

   from zoneinfo import ZoneInfo
   
   import arrow
   from tzfpy import get_tz
   
   tz = get_tz(139.7744, 35.6812)  # Tokyo
   
   arrow_now = arrow.now(ZoneInfo(tz))
   print(arrow_now.format("YYYY-MM-DD HH:mm:ss ZZZ"))
   # 2025-04-29 01:33:56.325194+09:00
   

   If you are using whenever, since whenever use tzdata internally, so it's compatible with tzfpy:

   examples/tzfpy_with_whenever.py:

   from whenever import Instant
   from tzfpy import get_tz
   
   now = Instant.now()
   
   tz = get_tz(139.7744, 35.6812)  # Tokyo
   
   now = now.to_tz(tz)
   
   print(now)
   # 2025-04-29T10:33:28.427784+09:00[Asia/Tokyo]
   

Performance

Benchmark runs under v1.2.0 on my MacBook Pro with Apple M3 Max.

Benchmark with _TZFPY_DISABLE_Y_STRIPES=1
.

---------------------------------------------- benchmark: 1 tests ----------------------------------------------
Name (time in us)        Min     Max    Mean  StdDev  Median     IQR  Outliers  OPS (Kops/s)  Rounds  Iterations
----------------------------------------------------------------------------------------------------------------
test_tzfpy            2.3709  2.9560  2.5338  0.0794  2.5213  0.0728     86;24      394.6709     500       10000
----------------------------------------------------------------------------------------------------------------

Legend:
  Outliers: 1 Standard Deviation from Mean; 1.5 IQR (InterQuartile Range) from 1st Quartile and 3rd Quartile.
  OPS: Operations Per Second, computed as 1 / Mean
Results (14.42s):
         1 passed
Benchmark with default index mode
.

---------------------------------------------- benchmark: 1 tests ----------------------------------------------
Name (time in us)        Min     Max    Mean  StdDev  Median     IQR  Outliers  OPS (Kops/s)  Rounds  Iterations
----------------------------------------------------------------------------------------------------------------
test_tzfpy            1.7190  2.4869  1.8321  0.1139  1.7934  0.0627     53;56      545.8076     500       10000
----------------------------------------------------------------------------------------------------------------

Legend:
  Outliers: 1 Standard Deviation from Mean; 1.5 IQR (InterQuartile Range) from 1st Quartile and 3rd Quartile.
  OPS: Operations Per Second, computed as 1 / Mean
Results (10.35s):
         1 passed

Or you can view more benchmark results on GitHub Action summary page.

More benchmarks compared with other packages can be found in ringsaturn/tz-benchmark.

Background

tzfpy was originally written in Go named tzf and use CGO compiled to .so to be used by Python. Since v0.11.0 it's rewritten in Rust built on PyO3 and tzf-rs, a tzf's Rust port.

I have written an article about the history of tzf, its Rust port, and its Rust port's Python binding; you can view it here.

Compare with other packages

Please note that directly compare with other packages is not fair, because they have different use cases and design goals, for example, the precise.

TimezoneFinder

I got lots of inspiration from it. Timezonefinder is a very good package and it's mostly written in Python, so it's easy to use. And it's much more widely used compared with tzfpy if you care about that.

However, it's slower than tzfpy, especially around the borders, and I have lots of API requests from there. That's the reason I created tzf originally. And then tzf-rs and tzfpy.

pytzwhere

I recommend to read timezonefinder's Comparison to pytzwhere since it's very detailed.

Contributing

Install:

• Rust
• Python
• uv

Available commands:
  build            - Build the project using uv
  build-ext        - Rebuild and install local Rust extension into venv
  fmt              - Format the code using ruff
  lint             - Lint the code using ruff
  sync             - Sync and compile the project using uv
  lock             - Lock dependencies using uv
  upgrade          - Upgrade dependencies using uv
  all              - Run lock, sync, fmt, lint, and test
  test             - Run non-benchmark tests
  test-all         - Run all tests including benchmark
  test-bench       - Run benchmark test with current env
  test-bench-index - Run benchmark in default/disable-y-stripes modes

make all

LICENSE

This project is licensed under the MIT license and Anti CSDN License[^anti_csdn]. The data is licensed under the ODbL license, same as evansiroky/timezone-boundary-builder

[^anti_csdn]: This license is to prevent the use of this project by CSDN, has no effect on other use cases.