pybenchx 0.1.0

A tiny, precise microbenchmarking framework for Python

PyBench — a microbenchmarking framework for Python

Write small, focused benchmarks with automatic file discovery, simple parameterization, and a clean, minimal CLI.

Run benchmarks with a single command:

pybench path/to/file_or_dir [-k keyword] [-P key=value ...]

✨ Features

• One-line registration: register cases with bench.bench(...); optionally use BenchContext.start()/end() to measure only the critical section.
• Auto-discovery: pybench <dir> expands **/*bench.py.
• Parameterization: generate multiple cases via params={...} (cartesian product) or per-case args/kwargs.
• CLI overrides: -P key=value to tweak parameters at runtime (no code edits).
• Sound timing: high-resolution, monotonic clock; built-in warmup and repeat.
• Nice output: medians, means, and stdev per call (ns) with iterations/repeats.

📦 Installation

• With pip:

  pip install pybench
  

• With uv:

  uv pip install pybench
  

🚀 Usage

• Run benchmarks in a file or directory:

  pybench path/to/file_or_dir [-k keyword] [-P key=value ...]
  

• Examples:

  # Filter by keyword in case/file name
  pybench benches/ -k json
  
  # Pass runtime parameters
  pybench benches/ -P size=1000 -P repeats=10
  

Tip: use BenchContext.start()/end() when you only want to measure the critical section of your benchmark.