How to write a Compiler in C for C

Question

Possible Duplicate:
Learning to write a compiler

I want to write a Compiler for C. This is a Project for my College i am doing as per my University. I am an intermediate programmer in C, with understanding of Data Structures.

Now i know a Compiler has the following parts: 1. Lexer 2. Parser 3. Intermediate Code Generator 4. Optimizer 5. Code Generator

I want to begin with the Lexer part and move on to Parser. I am consulting the following book: Compilers: Principles, Techniques, and Tools by Alfred V. Aho, Ravi Sethi, Jeffrey D. Ullman. The thing is that this book is highly theoretical and perplexing to me. I really appreciate the authors. But the point is i am not able to begin my project, as if i am blinded where to go. Need guidance please help.

Answer 1

The first stages to get working are your lexer and parser. Unless you are required to build a recursive descent parser, the most straight-forward way to get your teeth into this practically is to use a lexer-generator and parser-generator (aka compiler-compiler).

Both parser-generators and lexer-generators take a DSL in, and produce compilable C code. Lexer-generators such as flex take can tokenise regular languages, taking as input specially formatted regular expressions, and converting them into a finite-state automota in C that is capable of accepting the specified language. Parser-generators such as Bison take a LALR(1) grammar and associated actions perform on matching them as input, and produce a C program as output.

I am assuming that if you are taking a heavily theoretical course you will be able to understand the technical terminology. A detailed tutorial can help through the practicalities of using flex and bison.

Answer 2

If you find the dragon book hard going I suggest you try a different book. Andew Appel's "Modern Compiler Implementation in C" is a very good alternative. I've found it very accessible and hands on. It has a homepage: http://www.cs.princeton.edu/~appel/modern/c/

What will be the target of your compiler? x86 assembler? Something else? Learning about the target architecture can be a significant undertaking when writing a compiler. Often it's not enough to just have a general compiler book, they may not describe the target you want to generate code for. Make sure you dig up some relevant material for the code generation.

Good luck!