Optimize naturalsize algorithm by using math.log

[Zaczero]

Changes proposed in this pull request:

• Optimize naturalsize algorithm by using math.log

[Zaczero]

It's also faster to simply cast to float rather than do isinstance call.

[hugovk]

Yep:

❯ python --version
Python 3.14.0b1

❯ python -m timeit -s "value = 123.456" "bytes_ = float(value) if isinstance(value, str) else value"
10000000 loops, best of 5: 24.9 nsec per loop

❯ python -m timeit -s "value = 123.456" "bytes_ = float(value)"
20000000 loops, best of 5: 13.7 nsec per loop

❯ python -m timeit -s "value = '123.456'" "bytes_ = float(value) if isinstance(value, str) else value"
5000000 loops, best of 5: 47.7 nsec per loop

❯ python -m timeit -s "value = '123.456'" "bytes_ = float(value)"
10000000 loops, best of 5: 36.2 nsec per loop

Although in isolation this change would fail:

AssertionError: assert '1000.0 ZB' == '1.0 YB'

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.49%. Comparing base (b818dbf) to head (a79dafa).
⚠️ Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #253      +/-   ##
==========================================
- Coverage   99.49%   99.49%   -0.01%     
==========================================
  Files          11       11              
  Lines         798      794       -4     
==========================================
- Hits          794      790       -4     
  Misses          4        4              

Flag	Coverage Δ
macos-latest	97.73% <100.00%> (-0.02%)	⬇️
ubuntu-latest	97.73% <100.00%> (-0.02%)	⬇️
windows-latest	95.71% <100.00%> (-0.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[hugovk]

Thank you, this is about twice as fast. 🚀

❯ python -m timeit -s "from humanize import naturalsize" "naturalsize(1000**9 * 39)"
500000 loops, best of 5: 426 nsec per loop

❯ python -m timeit -s "from humanize import naturalsize" "naturalsize(1000**9 * 39)"
500000 loops, best of 5: 956 nsec per loop

[hugovk]

Yep:

❯ python --version
Python 3.14.0b1

❯ python -m timeit -s "value = 123.456" "bytes_ = float(value) if isinstance(value, str) else value"
10000000 loops, best of 5: 24.9 nsec per loop

❯ python -m timeit -s "value = 123.456" "bytes_ = float(value)"
20000000 loops, best of 5: 13.7 nsec per loop

❯ python -m timeit -s "value = '123.456'" "bytes_ = float(value) if isinstance(value, str) else value"
5000000 loops, best of 5: 47.7 nsec per loop

❯ python -m timeit -s "value = '123.456'" "bytes_ = float(value)"
10000000 loops, best of 5: 36.2 nsec per loop

Although in isolation this change would fail:

AssertionError: assert '1000.0 ZB' == '1.0 YB'