python 元类型编程,实现方法级的时间性能记录(performanceCount)
需求: 系统中的每个操作流程,需要打印所有被执行到的方法的耗时。
实现原理,利用python的元类型编程,动态改变class的创建过程,拦截class的方法,加上自定义的耗时记录。把记录导出到csv文件里面,方便对数据进行删选和排序。
具体代码实现:
#-*- coding: UTF-8 -*- #------------------------------------------------------------------------------- # Name: performanceCountMeta # Purpose: 记录每个方法的时间消耗,帮助程序员方便的查找系统的性能瓶颈 # # Author: ankier # # Created: 26-01-2013 # Copyright: (c) ankier 2013 # Licence: <your licence> #------------------------------------------------------------------------------- from types import FunctionType import time, datetime import csv pfile = file('E:\performanceCount.csv', 'ab') writer = csv.writer(pfile) writer.writerow(['Module', 'method', 'spend time', 'time Start', 'time End']) pfile.close() ## @summary: wrap 方法,AOP实现,记录每个函数的执行时间 def _RecordPerformance(func): def warp(*args): glos = func.func_globals model = glos['__name__'] methodName = func.func_name #Start time starTime = time.time() startStr = datetime.datetime.now().isoformat() result = func(*args) #End time endTime = time.time() endStr = datetime.datetime.now().isoformat() SpendTime = endTime - starTime row = [model, methodName, str(SpendTime), startStr, endStr] pfile = file('E:\performanceCount.csv', 'ab') writer = csv.writer(pfile) writer.writerow(row) pfile.close() return result return warp ## @summary: 性能计数器的元类型,动态修改类的创建过程,wrap 一些特定的方法。实现方法拦截,记录性能 class PerformanceCountMeta(type): def __new__(cls, name, bases, dct): for name, value in dct.iteritems(): if name not in ('__metaclass__', '__init__', '__module__') and \ type(value) == FunctionType: value = _RecordPerformance(value) dct[name] = value return type.__new__(cls, name, bases, dct)
例子如何使用,
## @summary: word host 窗口 class WordHostWindow(wx.Window): __metaclass__ = PerformanceCountMeta def __init__(self, parent): wx.Window.__init__(self, parent)
运行结果: