Micro C - um Compilador Acadêmico

Abstract

Compiler construction is a field of great interest in computer science, as it is responsible for translating high-level languages into machine code. However, the complexity of modern compilers, such as GCC and Clang/LLVM, or even the existing literature on the subject, makes learning and understanding them difficult. This work aims to mitigate this difficulty by building a compiler for a minimalist programming language (a subset of the C programming language), incorporating only the theoretical context necessary to understand compiler construction. Therefore, the objective of this work is to build a compiler that implements the compilation phases, translating the defined language, Micro C, into Assembly code and, finally, into an executable binary. As this is an educational compiler, all compilation steps are explicit and can be executed independently, following the compilation pipeline. Thus, it is possible to understand how the language is converted into its intermediate artifacts at each phase of the process. In the compiler developed for the Micro C language, a simple modular design divided into front-end and back-end was used. The front-end is associated with the analysis phases (lexical, syntactic, and semantic) to convert the source code into objects ready for conversion into machine language. On the other hand, the back-end converts these objects into simplified codes and, finally, Assembly code, ready to be converted into binaries. Finally, several test examples were made available to validate the compiler in various scenarios, such as nested loops (Bubble Sort, Sieve of Eratosthenes) and multiple recursion (Fibonacci, Greatest Common Divisor). The compiler for the Micro C language amounts to, in lines of code, less than 0.027% of the GCC compiler, making it a valuable tool for learning and teaching compilers. Furthermore, the source code of the compiler and all its artifacts are publicly available.