cosven/bench_result_analysis.py

## bench_result_analysis.py
import os
import re

from collections import defaultdict


def get_query_per_sec(line):
    """
    >>> s = 'queries:                             17493427 (58309.95 per sec.)'
    >>> get_query_per_sec(s)
    58309.95
    """
    return float(line.split('(')[1].split(' ')[0])


def get_p95_latency(line):
    """
    >>> s = "95th percentile:                        2.35"
    >>> get_p95_latency(s)
    2.35
    """
    return float(line.split(':')[1].strip())


def calc_per_dir(dirname):
    """计算每个目录的结果

    每个目录有三个 sysbench 结果文件
    """
    query_per_sec_sum = 0
    latency_list = []
    for fname in os.listdir(dirname):
        if not fname.endswith('.out'):
            continue
        fpath = os.path.join(dirname, fname)
        with open(fpath) as f:
            for line in f:
                line = line.strip()
                if line.startswith('queries:'):
                    query_per_sec = get_query_per_sec(line)
                    query_per_sec_sum += query_per_sec
                if line.startswith('95th'):
                    latency = get_p95_latency(line)
                    latency_list.append(latency)

    assert len(latency_list) == 3
    latency_mid = sorted(latency_list)[1]
    return query_per_sec_sum, latency_mid


def get_threads_count_from_dirname(dirname):
    """获取 threads 数目"""
    return int(dirname.split('_')[-3])


def get_type_from_dirname(dirname):
    """获取实验类型 -- 比如：oltp-point-select"""
    return '-'.join(dirname.split('_')[1:-3])


def show(result):
    """展示结果"""
    for result_type, results in result.items():
        print(result_type)
        results = sorted(results, key=lambda each: each[0])
        for each in results:
            threads = each[0]
            qps, latency = each[1]
            print('{} \t {:.2f} \t {}'.format(each[0], qps, latency))

def show_as_md_table(result):
    """展示结果 markdown table"""
    for result_type, results in result.items():
        print(result_type)
        print()
        results = sorted(results, key=lambda each: each[0])
        print('| threads | qps | latency |')
        print('| ------- | --- | ------- |')
        for each in results:
            threads = each[0]
            qps, latency = each[1]
            print('| {}   | {:.2f} |  {} |'.format(each[0], qps, latency))
        print()


def calc_all_v21():
    result = defaultdict(list)
    for fname in os.listdir('.'):
        if fname.startswith('2.1_oltp'):
            dirname = fname
            threads_count = get_threads_count_from_dirname(dirname)
            result_type = get_type_from_dirname(dirname)
            _result = (threads_count * 3, calc_per_dir(dirname))
            result[result_type].append(_result)
    show_as_md_table(result)


def calc_all_v3():
    result = defaultdict(list)
    for fname in os.listdir('.'):
        if fname.startswith('oltp'):
            dirname = fname
            threads_count = get_threads_count_from_dirname(dirname)
            result_type = '-'.join(dirname.split('_')[:-3])
            _result = (threads_count * 3, calc_per_dir(dirname))
            result[result_type].append(_result)
    show(result)


if __name__ == '__main__':
    calc_all_v3()
	import os
	import re

	from collections import defaultdict


	def get_query_per_sec(line):
	"""
	>>> s = 'queries: 17493427 (58309.95 per sec.)'
	>>> get_query_per_sec(s)
	58309.95
	"""
	return float(line.split('(')[1].split(' ')[0])


	def get_p95_latency(line):
	"""
	>>> s = "95th percentile: 2.35"
	>>> get_p95_latency(s)
	2.35
	"""
	return float(line.split(':')[1].strip())


	def calc_per_dir(dirname):
	"""计算每个目录的结果

	每个目录有三个 sysbench 结果文件
	"""
	query_per_sec_sum = 0
	latency_list = []
	for fname in os.listdir(dirname):
	if not fname.endswith('.out'):
	continue
	fpath = os.path.join(dirname, fname)
	with open(fpath) as f:
	for line in f:
	line = line.strip()
	if line.startswith('queries:'):
	query_per_sec = get_query_per_sec(line)
	query_per_sec_sum += query_per_sec
	if line.startswith('95th'):
	latency = get_p95_latency(line)
	latency_list.append(latency)

	assert len(latency_list) == 3
	latency_mid = sorted(latency_list)[1]
	return query_per_sec_sum, latency_mid


	def get_threads_count_from_dirname(dirname):
	"""获取 threads 数目"""
	return int(dirname.split('_')[-3])


	def get_type_from_dirname(dirname):
	"""获取实验类型 -- 比如：oltp-point-select"""
	return '-'.join(dirname.split('_')[1:-3])


	def show(result):
	"""展示结果"""
	for result_type, results in result.items():
	print(result_type)
	results = sorted(results, key=lambda each: each[0])
	for each in results:
	threads = each[0]
	qps, latency = each[1]
	print('{} \t {:.2f} \t {}'.format(each[0], qps, latency))

	def show_as_md_table(result):
	"""展示结果 markdown table"""
	for result_type, results in result.items():
	print(result_type)
	print()
	results = sorted(results, key=lambda each: each[0])
	print('\| threads \| qps \| latency \|')
	print('\| ------- \| --- \| ------- \|')
	for each in results:
	threads = each[0]
	qps, latency = each[1]
	print('\| {} \| {:.2f} \| {} \|'.format(each[0], qps, latency))
	print()


	def calc_all_v21():
	result = defaultdict(list)
	for fname in os.listdir('.'):
	if fname.startswith('2.1_oltp'):
	dirname = fname
	threads_count = get_threads_count_from_dirname(dirname)
	result_type = get_type_from_dirname(dirname)
	_result = (threads_count * 3, calc_per_dir(dirname))
	result[result_type].append(_result)
	show_as_md_table(result)


	def calc_all_v3():
	result = defaultdict(list)
	for fname in os.listdir('.'):
	if fname.startswith('oltp'):
	dirname = fname
	threads_count = get_threads_count_from_dirname(dirname)
	result_type = '-'.join(dirname.split('_')[:-3])
	_result = (threads_count * 3, calc_per_dir(dirname))
	result[result_type].append(_result)
	show(result)


	if __name__ == '__main__':
	calc_all_v3()