Research And Implementation Of Measurement Methods For Automatic Code Generation And Completion

Because the automatic code generation and completion can greatly reduce the programmer's workload,and improve the efficiency of software development,so the automatic code generation and completion has become one of the current research hotspots.Current mainstream software integrated development environment,and part of the code automatic generation tool though to some extent,realized the function of automatically generated code,but they are generally unable to properly according to the requirements of the programmer and efficient automatic generation and code completion,so to measure of code automatically generated and completion,to improve the code automatic generation technology and improve the quality and efficiency of code automatically generated has very important significance.Existing scholars at home and abroad have put forward corresponding measurement indexes and methods to carry out research on the measurement of code automatic generation and completion quality and efficiency.However,according to the current research results,only some measurement indexes of automatic generation and completion of measurement code are proposed,and the corresponding measurement indexs model is not established.However,when measuring the indexes of automatic code generation and completion,only some indexes can be quantitatively analyzed,which does not reflect the overall quantization results of the quality and efficiency of automatic code generation and completion.Based on the above discussion,this paper mainly carried out the following main research work:(1)Analysis code automatically generated and completion techniques or methods to realize the process and result,this article selects and defined the code automatically generated with the completion of the quality and efficiency features,and contain code syntax rules conform to the degree of relevancy,and generate code complexity,generate the code context such as 12 quantitative metrics,and establish the code automatically generated and completion metric model,to measure the code automatically generated with the quality and efficiency of completion.(2)Based on the metric model of automatic code generation and completion,it is necessary to solve such problems as how to obtain quantitative measurement index data and measure the quality and efficiency of automatic code generation and completion.This article used for code automatically generated with the completion of syntax,semantics and similarity analysis method,and establish the code automatically generated and complement a traceability matrix,and then according to the result of analysis defines the rule for calculating the corresponding quantitative metrics,finally will obtain the metrics data for standardization,using combination of two methods of weight,automatic code generation and completion quality and efficiency of metric model of quantitative measure empowerment,calculate the code automatically generated and completion quality and efficiency of measurement.(3)Design and implement a prototype tool for automatic code generation and completion measurement analysis software.The rationality of the metric model of automatic code generation and completion and the effectiveness of the metric method of automatic code generation and completion are verified by an example.