SableCC, an Object-Oriented Compiler Framework

In this thesis, we introduce SableCC, an object-oriented framework that generates compilers (and interpreters) in the Java programming language. This framework is based on two fundamental design decisions. Firstly, the framework uses objectoriented techniques to automatically build a strictly-typed abstract syntax tree that matches the grammar of the compiled language and simpli es debugging. Secondly, the framework generates tree-walker classes using an extended version of the visitor design pattern which enables the implementation of actions on the nodes of the abstract syntax tree using inheritance. These two design decisions lead to a tool that supports a shorter development cycle for constructing compilers. To demonstrate the simplicity of the framework, we discuss the implementation of a state-of-the-art almost linear time points-to analysis. We also provide a brief description of other systems that have been implemented using the SableCC tool. We conclude that the use of object-oriented techniques signi cantly reduces the length of the programmer written code, can shorten the development time and nally, makes the code easier to read and maintain. ii R esum e Dans cette th ese, nous pr esentons SableCC, un environnement orient e-objet qui sert a construire des compilateurs et des interpr eteurs dans le langage de programmation Java. Deux d ecisions fondamentales forment les assises de cet environnement. En premier lieux, l'utilisation de techniques orient ees-objet sert a construire un arbre syntaxique strictement typ e qui est conforme a la syntaxe du langage compil e et qui simpli e le d everminage des programmes. En second lieux, l'environnement g en ere des classes qui traversent l'arbre syntaxique. Ces classes utilisent une version amend ee du mod ele de conception orient e-objet \le visiteur". Ceci permet d'ajouter des actions a ex ecuter sur les noeuds de l'arbre syntaxique en utilisant les techniques d'h eritage objet. Ces deux d ecisions font de SableCC un outil qui permet d'abr eger le cycle de programmation du d eveloppement d'un compilateur. Pour d emontrer la simplicit e de SableCC, nous expliquons les etapes de programmation d'une analyse a la ne pointe des techniques de compilation appel ee \analyse de pointeurs en temps presque lin eaire". De plus, nous d ecrivons bri evement d'autres syst emes qui ont et e bâtis avec SableCC. Nous concluons que l'utilisation de techniques orient ees-objet permet de r eduire substantiellement la quantit e du code ecrit par le programmeur tout en ecourtant possiblement le temps de d eveloppement. Le code s'en trouve plus lisible et facile a maintenir. iii Acknowledgments I cannot thank enough Professor Laurie Hendren, for her technical and nancial support throughout the course of my study at McGill University. She was always there to listen and help. THANKS Laurie. I would like to thanks the students of the fall 1997 edition of the CS-308-621A course that used SableCC and gave me the essential feedback to improve SableCC. Specially, I would like to thank Amar Goudjil for implementing his course project using SableCC, and for all the long discussions that led to many changes in the design of SableCC. I would like to also thank the students of the winter 1998 edition of the CS-308-521B course. More specially, I am thankful to Dmitri Pasyutin, Monica Potlog, Navindra Umanee, Vijay Sundaresan, Youhua Chen and Xiaobo Fan for implementing their object-oriented compiler for the WIG (Web Interface Generator) language using SableCC. I would like to thank them for their help nding and xing a few discrepancies in SalbeCC and testing the framework. I would like to say a special thank to the Sable Research Group members who made a friendly environment to work together, in particular, Raja Vallee-Rai, Fady Habra, Laleh Tajrobehkar, Vijay Sundaresan and Chrislain Raza mahefa. I would like to thank also Zhu Yingchun, Rakesh Ghyia and the unforgettable Christopher Lapkowski of the ACAPS Research Group. Further, I would like to thank my wife Gladys Iodio for her incredible patience and her support in the long hours spent writing this thesis. Without her constant presence, this thesis would not be. iv Dedicated to my wife, Gladys V. Iodio v