Simplified-C Compiler for MIPS

This repository contains a C to MIPS compiler developed as a course project for the compiler course CSC4180 at CUHK. The compiler translates a subset of the C language into MIPS assembly code, focusing on basic arithmetic, logical operations, control structures, and array operations.

An NFA regex engine is implemented for lexer, and a LR(1) table driven parser generator (Parser class is an implementation of a bison-like parser generator) is implemented for parsing. Grammar.cpp is the user of Parser class, with grammar rules (productions) along with semantic roitines configured there, which builds the parser ready for parsing.

An one-pass code-gen is adopted and an AST is built on the fly in the middle of parsing. Then, code gen will be conducted on the root of AST, which in effect perfroms a DFS on the tree to complete the code generations recursively.

As for symbol table and stack memory allocations, I adopt an approach similar to my previous project MICRO Language Compiler, using the same SymEntry structure (also for intermediate/temporary value). The compiler is limited and uses no more than 4 registers, leaving a lot of space for register allocation optimization (a possible approach is to leave SymEntry in the generated code, which in effect becomes 3AC, and perform algorithms like graph coloring to schedule the allocation of registers).

Project Structure

The project is organized into the following structure:

.
├── Makefile
├── ast (namespace ASTGen)
│   ├── code.cpp
│   ├── code.hpp
│   ├── symbol_table.cpp
│   ├── symbol_table.hpp
│   ├── syntax_tree.cpp
│   ├── syntax_tree.hpp
│   └── utils.hpp
├── parser
├── parser_src
│   ├── configuration.cpp
│   ├── grammar.cpp
│   ├── include
│   │   ├── configuration.h
│   │   ├── grammar.h
│   │   ├── parser.h
│   │   ├── production.h
│   │   └── symbol.h
│   ├── main.cpp
│   ├── parser.cpp
│   ├── production.cpp
│   └── symbol.cpp
├── readme.md
├── run_compiler.sh
├── scanner_src
│   ├── Makefile
│   ├── README.md
│   ├── dfa.cpp
│   ├── include
│   │   ├── dfa.hpp
│   │   ├── nfa.hpp
│   │   └── scanner.hpp
│   ├── main.cpp
│   ├── nfa.cpp
│   ├── run_scanner.sh
│   └── scanner.cpp

Grammar in CFG

Components

To understand the code, you can start with grammar.cpp/h and syntax_tree.cpp/hpp. You can refer to parser.cpp/h for parser and scanner.cpp/h for scanner.

• ast: Contains the code for abstract syntax tree (AST) generation, symbol table management, and MIPS assembly code generation.
  • code.cpp/hpp: Encapsulates code generation small building blocks.
  • symbol_table.cpp/hpp: Contains the SymbolTable class for managing the symbol table during code generation.
  • syntax_tree.cpp/hpp: Defines the SyntaxTreeNode class and its generateCode() method for generating MIPS assembly code.
• parser: Contains the parser implementation, which builds the parsing table and parses the tokens.
  • configuration.cpp/h: Contains the Configuration class and ConfigurationSet class, representing a configuration (production rule plus dot; a configuration set is a state) in the parsing process.
  • grammar.cpp/h: Contains the Grammar class, which defines the grammar (production) rules for the parser, the semantic routines that trigger the code gen process are also attached to the production rules.
  • parser.cpp/h: Contains the Parser class, which is responsible for building the parsing table and parsing the tokens.
  • production.cpp/h: Contains the Production class, which is an abstraction for a production rule in the grammar.
  • symbol.cpp/h: Contains the Symbol class, which represents a grammar symbol.
• scanner_src: Contains the scanner implementation, which is responsible for scanning the input C code and converting it into tokens.
  • dfa.cpp/hpp: Contains the DFA class, which represents a deterministic finite automaton.
  • nfa.cpp/hpp: Contains the NFA class, which represents a nondeterministic finite automaton.
  • scanner.cpp/hpp: Contains the Scanner class, which implements the main scanning functionality, which utilizes nfa for regular expression matching

Building the Compiler

To build the compiler, simply run make in the root directory of the project:

make

This will generate the compiler executable in the project root.

Please note that the CXX_STANDARD should be at least C++14 as std::hash function for enum type is not provided in C++11 (added since C++14, as was stated here). An alternative is to manually provide specialization in C++11 as:

std::hash<std::underlying_type<Enum>::type>

Usage

To compile a C source file, run the following command:

./compiler  <  input_file.c  >  output_file.asm

This will generate a MIPS assembly file output_file.asm from the C source file input_file.c.

You can then run the MIPS assembly code using a MIPS simulator, such as SPIM or MARS.

Limitations

• The compiler currently only supports a single main function with no parameters and no return value.
• No support for pointers, structures, or other advanced C language features.
• No optimization is performed on the generated assembly code.

Contributing

Contributions are welcome! If you find any bugs, have suggestions for new features, or would like to help improve the project, please feel free to open an issue or create a pull request.