Minimal python benchmark class

benchmark.py

import time
import logging
from logging import StreamHandler, Formatter

minimal_logger = logging.getLogger(__name__ + "minimal")
minimal_logger.setLevel(logging.INFO)

_handler = StreamHandler()
_handler.setFormatter(Formatter('%(message)s'))
minimal_logger.addHandler(_handler)

class Benchmark:
    def __init__(self):
        self.benchmarks = {}

    def run(self, name):
        self.running_benchmark = { "name": name }
        return self

    def __enter__(self):
        self.running_benchmark["start"] = time.time() * 1e3
        return self

    def __exit__(self, *args):
        self.running_benchmark["end"] = time.time() * 1e3
        self.running_benchmark["interval"] = self.running_benchmark["end"] - self.running_benchmark["start"]

        name = self.running_benchmark["name"]
        if name in self.benchmarks:
            self.benchmarks[name].append(self.running_benchmark)
        else:
            self.benchmarks[name] = [self.running_benchmark]

    def summarize(self):
        for name, bmarks in self.benchmarks.items():
            intervals = sorted([bmark['interval'] for bmark in bmarks])
            total = reduce(lambda m, i: m + i, intervals)
            avg = total / len(intervals)
            median = intervals[len(intervals) / 2]
            min_i = intervals[0]
            max_i = intervals[-1]

            minimal_logger.info(
                "%d iterations of %s took %f milliseconds\n\tmin: %f\n\tmax: %f\n\tmedian: %f\n\tavg: %f" %
                (len(intervals), name, total, min_i, max_i, median, avg))


iterations = 10
sub_iterations = 100

def benchmark_expo(benchmark):
    import random
    with benchmark.run("expovariate"):
        for i in range(sub_iterations):
            random.expovariate(0.5)

def benchmark_gamma(benchmark):
    import random
    with benchmark.run("gammavariate"):
        for i in range(sub_iterations):
            random.gammavariate(0.5, 1.5)

if __name__=='__main__':
    benchmark = Benchmark()

    for i in range(iterations):
        benchmark_expo(benchmark)
        benchmark_gamma(benchmark)

    benchmark.summarize()