Design And Implementation Of The Automated Testing Tool For Web System Based On Flow-Playback

This paper designed and implemented an automated testing tool for Web system based on flow-playback.The aim is to intercept the real online traffic and playback it on test environment,and then check results of the playback according to test requirements.This tool is implemented by shell script and Python,its function modules were written in Python,the invoked commands were packaged by shell.Finally,the command line is offered to users,combining Jenkins can easily achieve project's continuous integration construction.The main part of automated testing tools are traffic playback and subsequent inspects.Flow acquisition and playback are two basic steps,followed by the analysis of playback result.The result generated during playback can be departed into several fields according to analysis requirements.Analysis items include log checking,card recall,data evaluation.The author divided the system to five modules and implemented each of them.(1)Traffic Acquisition:Include interception of online traffic and mock of offline traffic,respectively,to obtain the test traffic of existing and new functions in order to achieve a comprehensive coverage of the card.Get test traffic of existing functions from production environment,and test traffic of new function is implemented by mocking a list of request url.(2)Traffic Playback:Replay the traffic which is got by different ways in offline environment,resends request data to the test environment,waiting for the reaction of system,to gain access records.Online traffic uses TCPCopy to achieve playback,and offline traffic directly uses 'Curl' to deal with the request link.(3)Check Logs:Include log specification checking and exception analysis.The log specification check is with respect to business logs,user behavior analysis logs and their joint logs,according to their rules to see if there are missing records or error values of the fields and so on.Exception parsing tool monitors business exception,operation error logs,programming errors,records errors generated during operation and then reports to the responsible person.(4)Card Recall:The page consists of cards.Check card recall situation of first page and middle page in each subproject fold.The user can enter the trade,card type,card number and other parameters to determine whether the search results have shown completely;implement judgment of the length of time consuming for each page,including the overall time-consuming analysis of each page and specific time-consuming distribution of each card cutting the degree of 10ms.It can help the user locate the page over time and obtain the timeout period distribution of the page.(5)Data Assessment:By examining the situation of replaying real online traffic,check the difference of showing the card or not.Compare the income between control group and test group to see if you can get more money,which can help make the formal decision whether deliver a card or not.Automated testing tools has already been applied in production environments,basically meet the test requirements of existing projects.In addition,through the trial in the new types of projects,found that some places can be optimized,and the program has been extended accordingly.In effect,this automated testing tools can cover the testing needs of a project in early development period,bring great convenience to testers and developers.