Python API

• Decorators
• State
• Runner
• Reporters
• Compare
• Profiling
• Context
• Registry
• Configuration
• Regressions

Top-level re-exports

The public surface lives at the package root; all names below are importable from mew directly:

benchmark([fn, name, tags])	Register a function as a single benchmark.
parametrize(parameters, *[, name, ids, tags])	Register a parametrized benchmark family.
product(*[, name, ids, tags, min_time, ...])	Register a benchmark family from the cartesian product of iterables.
State(*args, **kwargs)	Structural state passed into benchmark targets.
BenchmarkFn(*args, **kwargs)	A callable benchmark target.
run([entries, reporter, min_time, ...])	Run benchmarks via the C++ Google Benchmark backend.
Reporter(*args, **kwargs)	Duck-typed reporter interface consumed by the C++ runner.
RichReporter(*[, terminal, show_memory, ...])	Stream one row per benchmark to a terminal as runs complete.
JSONReporter(*[, output])	Emit a single {"context": ..., "benchmarks": [...]} document, GB-style.
JSONLReporter(*[, output, append, header])	Stream one self-contained JSON object per Run, one per line, flushed as runs land.
Fanout(reporters)	Broadcast reporter callbacks to a list of underlying reporters.
RunRow	One benchmark run, serialized: the contract every reporter consumes.
TimeUnit(*values)	Time unit used for reported per-iteration durations.
Registry()
Entry(name, fn[, file, options, tags, ...])
REGISTRY
BENCHMARK_VERSION	str(object='') -> str str(bytes_or_buffer[, encoding[, errors]]) -> str
BENCHMARK_COMMIT	str(object='') -> str str(bytes_or_buffer[, encoding[, errors]]) -> str
set_context(key, value)	Set a single value in the global benchmark context.
update_context(*mapping, **kwargs)	Set many context values at once.
get_context()	Return a deep copy of the current global benchmark context.
clear_context()	Drop every entry from the global benchmark context.