Evoluindo o método ChAPTER em direção à geração de casos de teste

Abstract

A Software Product Line (SPL) facilitates the development of applications within the same domain as it enables the reuse of artifacts. A Context-Aware Software Product Line (CASPL), in turn, aims to develop applications that change their behavior dynamically based on context information. One of the problems associated with CASPL is the quality assurance, because its complexity is greater than a traditional application since a CASPL adds the challenges inherent to both the software product line and the context-aware applications. Software testing is one of the ways to ensure quality and to reduce the costs involved in the testing phase, the automatic generation of test cases is a solution found in the literature. This generation can be done using a Textual Use Case (TUC). However, there are challenges involved, because a CUT is described in natural language (NL), which can lead to ambiguity and imprecision in its description, making it difficult to generate test cases. To solve the problem of using an NL to describe a TUC, the vocabulary and the grammar of a natural language can be restricted, leading to the use of a Controlled Natural Language (CNL). In the literature, only one proposal was found to automate the tests for LPSSC from use cases, the ChAPTER method, which uses natural language to describe a TUC. Given the problems mentioned before when using a CUT described in NL, this work first proposes a CNL, called CARNAUbA, to help in the description and identification of the TUC information to be used in the tests. This work then extends the ChAPTER method and the tool that implements it to allow the use of CARNAUbA to generate the necessary structure of the tests for LPSSC to execute them. To verify the feasibility of the proposed extension in the method and implemented in the tool, two software products lines are used as case studies, and it is checked if the tests generated are in accordance with the use cases.