Keywords : Web Services, Test Case Generation, Test Case Mutation, Self-Healing Software, Software Security, Large Language Models

This thesis addresses the challenge of ensuring reliability and security in modern software systems, which increasingly rely on web service compositions. A composition is defined as a set of independently deployed services that communicate with each other to deliver a specific functionality.

While this architectural style offers agility and scalability, it raises significant assurance challenges. Testing such systems is widely recognised as a complex process due to inherent difficulties in controlling or monitoring numerous concurrent components interacting simultaneously. This presents several major challenges: understanding their behaviour without formal models, generating relevant tests from raw execution traces, isolating tightly coupled services, automatically discovering security vulnerabilities in increasingly exposed REST APIs, and enhancing service resilience without disrupting their operation or modifying their source code.

This thesis proposes a comprehensive pipeline of approaches to address these challenges:

• An approach for the automatic generation of test cases for service compositions from execution logs, without requiring prior specifications;
• A method to test each service individually through the automatic generation of partial models, IOTS tests, and mock components;
• An approach to transform existing tests into tests capable of revealing vulnerabilities in REST APIs;
• A dynamic healing approach based on large language models (LLMs) and a dual-mitigation strategy, capable of automatically applying corrective measures through encapsulation without modifying the service code.
  
   
   
  Jury composed of:
   
  M. SEBASTIEN SALVA,  Professeur des universités, Université Clermont Auvergne, Directeur de thèse   
  M. Jean-Paul ARCANGELI, Professeur des universités, Université de Toulouse, IRIT, Rapporteur  
  Mme Pascale LE GALL,  Professeur des universités, Centrale Supélec, Rapporteur
  M. Pascal LAFOURCADE,  Professeur des universités, Université Clermont Auvergne, Examinateur
  Mme Fatiha ZAIDI,  Professeur des universités, Université Paris-Saclay, CNRS, Examinateur.