How to monitor code cyclomatic complexity by refactoring indicators

1 引言

Software applications are being developed at the right time,”重构”,It is a work that inevitably needs to be carried out in the development process.重构代码,In order to adapt to the diverse scenarios that were not considered at the beginning of the current module design,And increase the maintainability of the module、健壮性、可测试性.那么,How to clarify the direction of refactoring,And what about the results of quantitative reconstruction？Code cyclomatic complexity can be an alternative metric.The following describes how to obtain the code circle complexity of the application for online monitoring,Provide data support for the complexity of the review program.

2 背景知识

2.1 圈复杂度

圈复杂度（Cyclomatic complexity,简写CC）也称为条件复杂度,是一种代码复杂度的衡量标准.由托马斯·J·麦凯布（Thomas J. McCabe, Sr.）于1976年提出,用来表示程序的复杂度,其符号为VG或是M.它可以用来衡量一个模块判定结构的复杂程度,数量上表现为独立现行路径条数,也可理解为覆盖所有的可能情况最少使用的测试用例数.圈复杂度大说明程序代码的判断逻辑复杂,可能质量低且难于测试和维护.程序的可能错误和高的圈复杂度有着很大关系.

2.2 Cyclomatic complexity calculation method

Commonly used structural cyclomatic complexity calculations

• 顺序结构：The order structure complexity is 1.

• if-else-else、switch-case：每增加一个分支,复杂度增加1,&& 、|| The operation is also a branch.

• 循环结构：Add a loop structure,复杂度增加1.

• return：增加一条return语句,Complexity will increase1.

2.3 Cyclomatic complexity metrics

Relatively recognized metrics in the industry are listed above,In fact, this completely depends on the size of your business and the situation of the project.Suppose your business is simple,And it's a monolithic application,The functions are very simpleCRUD,Then your cyclomatic complexity is not so easy even if you think about it.At this point you can choose to set the reconstruction threshold for cyclomatic complexity to 10.

Suppose your business is very complex,And it involves multiple other microservice system calls,Plus a variety of businessescorner case的判断,cyclomatic complexity100Might not even matter.

2.4 Cyclomatic complexity reduction method

1）Function extraction and splitting,单一职责

• Split into subfunctions

• Each function must have a clear function implementation,Don't merge functional implementations in pursuit of fewer lines

• Logic modules and data modules should be written separately

2）优化算法

• Reduce unnecessary conditions、循环分支,尽量少用 if …else … ,Use a ternary expression to replaceif else

3）表达式逻辑优化

• 合并条件表达式,比如使用a || b || c

4）Reduce in advancereturn

3 方案概述

3.1 脚本设计

1）开发语言

• python

2）依赖环境

• lizard

• APScheduler

• smtplib

• pymysql

3）脚本架构

3.2 功能介绍

1）Supports retrieval of language ranges：

支持15种开发语言,Contains commonly used languages ​​as follows

• C/C++ (works with C++14)

• Java

• C# (C Sharp)

• JavaScript (With ES6 and JSX)

• Python

• Golang

2）Scan parameter configuration description：利用lizard执行扫描,常用命令如下：

Configure the inspection scope：

• List the programming languages ​​to analyze.如果留空,All supported languages ​​will be searched.

-l LANGUAGES, --languages LANGUAGES

• 排除与模式匹配的文件.
  match everything？匹配任何单个字符,“/folder/
  ”Recursively exclude everything in a folder.可以指定多个模式.Don't forget to add around the pattern“”号.

-x EXCLUDE, --exclude EXCLUDE

• 设置白名单, 默认’./whitelizard.txt’

 -W WHITELIST, --whitelist WHITELIST

Configure threshold warnings：

• Threshold for cyclomatic complexity warnings,默认值为15,>15会产生警告.

-C CCN, --CCN CCN

• Sets the limit on the number of fields.number of lines of code,圈复杂度,令牌数,Number of parameters or custom fields.If the function setting exceeds the limit, an alarm will be issued.

-T THRESHOLDS, --Threshold THRESHOLDS

Configure report output：

• Output to file according to format

-o OUTPUT_FILE, --output_file OUTPUT_FILE

官网地址：

http://www.lizard.ws

源码地址：

https://github.com/terryyin/lizard

3）定时执行扫描任务：

• 通过BackgroundScheduler创建调度任务,Automatic trigger scan method,result write library

def dojob(): scheduler = BackgroundScheduler() scheduler.add_job(func, "cron", hour=21, minute=30) scheduler.start()

3.3 结果展示

3.3.1 Explanation of report nomenclature

• Cyclomatic complexity,Cyclomatic complexity is also known as branch complexity,最好保持在15 以下,The current script sets the threshold10.

• LOC,The number of lines of code that contain comments,目前设置200阀值.

• Token count ,token的个数,A program can have at most 8192 个令牌, Each token is a word,例如关键字,标识符,常量,标点符号,操作符. Counts parentheses and strings as 1 个令牌. 逗号、句点、LOCAL、分号、END and comments do not count.

• Parameter count,Parameter statistics is the number of parameters of a function,The current script sets the threshold10.

3.3.2 执行结果展示

• Windows环境运行脚本,输入file_root（文件地址）执行扫描,Supports the automatic pop-up of the browser to display the current runningHtml报告

• Execute on a weekly basis,Scan according to the system dimension,Support trigger email notification corresponding to system development to view the method name that exceeds the threshold

3.3.3 Application data monitoring

• Regularly pull the latest code of the specified branch every week,执行文件分析,存储扫描结果,Displayed through data charts

4 总结

A measure of how good or bad software code is,Cyclomatic complexity can be used as a reference indicator,R&D can split functions by refining、优化算法、Optimizing logical expressions and other methods to reduce modules（函数）圈复杂度.The above describes an online monitoring method of cyclomatic complexity,Take advantage of online data,Combined with existing team projects,In order to form a better practice mechanism.