PROGRAM QUERY LANGUAGE PQL SYSTEM

Abstract

The continuous increasing maintenance cost draws wide attention to methods and tools for program understanding. With the evolution of software system, program source code gradually becomes the only reliable definition of the system. Analyzing source code consequently becomes one of the key activities to understand an aged software system. Static program analysis (SPA for short), in this case, becomes one of the key techniques which can help software engineers in understanding existing software systems. An SPA tool extracts relevant information from program source code and stores it into a Program Knowledge Base (PKB for short). This dissertation documents the design and implementation of a SPA tool, named PQL system. PQL stands for Program Query Language which is a general-purpose notation to specify program views in different abstraction levels. To support PQL queries, we first built a hybrid PKB system which integrates a relational database with abstract syntax trees. Global program information is pre-extracted and stored in the relational database, while detailed program information is represented in the abstract syntax tree form. The hybrid PKB approach provides an efficient and flexible way of answering PQL queries, Interfaces at different abstraction levels provided by our hybrid PKB system makes it easier for other tools to be built upon it. We, then, present a design of a simplified PQL interpreter. It parses PQL queries with its user-end, and retrieves the PKB system with its query-end. As no experiences with implementing this kind of systems have been reported, our PQL system implementation is innovative and experimental. Finally, we proposed a distributed PKB system, which works basically in a client/server computing model. This model, in one way, coped with today's enterprise computing environment. In another way, it provides a greater flexibility and scalability of the system.