HCC C0-like Compiler

本项目是一个类C0语言的编译器。特性如下：

• 扩充C0文法，包含了常量、变量定义，支持条件、循环、开关等多种语句，支持函数递归调用
• 目标代码格式是x86汇编代码，可经MASM汇编器生成Win32可执行文件
• 实现了一些代码优化，如公共子表达式删除（DAG图）、数据流分析、全局寄存器分配等
• 编译器代码本身平台无关

扩充C0语言文法定义见：docs/designed_BNF.txt，详细设计文档：docs/designing_report.doc。

编译

可以直接打开在Code::Blocks中打开hcc.cbp项目文件，在编译器设置里勾选C++11支持，然后单击build按钮。

或者可以直接在命令行中输入：

g++ -std=c++11  -c actvar.cpp     -o obj\Debug\actvar.o
g++ -std=c++11  -c genasm.cpp     -o obj\Debug\genasm.o
g++ -std=c++11  -c lex.cpp        -o obj\Debug\lex.o
g++ -std=c++11  -c main.cpp       -o obj\Debug\main.o
g++ -std=c++11  -c opt.cpp        -o obj\Debug\opt.o
g++ -std=c++11  -c opt_dag.cpp    -o obj\Debug\opt_dag.o
g++ -std=c++11  -c semantics.cpp  -o obj\Debug\semantics.o
g++ -std=c++11  -c syntax.cpp     -o obj\Debug\syntax.o
g++ -std=c++11  -c util.cpp       -o obj\Debug\util.o
g++ -o bin\Debug\hcc.exe obj\Debug\*.o

运行测试样例

tests目录中包含了一些测试样例。例如运行bin\Debug\hcc.exe tests\3.fibo.c即可输出对应x86汇编代码。

hcc.exe的用法如下

Usage: hcc [-c FILENAME|-l FILENAME|-s FILENAME|
            -t FILENAME] [-o] [-D] [-R] [-r] [-d] SOURCEFILE

       Read extended C0 source code from SOURCEFILE file and compile
       to MASM source code. The generated code will be output into
       std-out by default.

  -c FILENAME
       Output the generated code to FILENAME. Assign FILENAME as - to
       indicate std-out.

  -l FILENAME
       Only do lexical analysis and output the result to FILENAME.
       Assign FILENAME as - to indicate std-out.

  -s FILENAME
       Only do syntax analysis and output the result to FILENAME.
       Assign FILENAME as - to indicate std-out.

  -t FILENAME
       Do semantic analysis and dump the relevant tables to FILENAME.
       Assign FILENAME as - to indicate std-out.

  -o   Perform optimization.
  -D   Do NOT use DAG to simplify repeated expression.
  -R   Do NOT use reduce code using data flow, redundant deletion, copy-.
       spread, or constant combination.
  -r   Use global register.
  -d   Output verbose debug information.

HCC C0-like Compiler

This is a compiler for a C0-like language. Features include:

1. Extended C0 syntax, including constant and variable definitions, supporting multiple statements such as conditions, loops, switches, and supporting recursive function calls
2. The target code is x86 assembly, and Win32 executable file can be generated by MASM assembler
3. Some code optimizations have been implemented, such as common sub-expression deletion (DAG graph), data flow analysis, global register allocation, etc.
4. The compiler code itself is platform independent

The language definition can be found in docs/designed_BNF.txt. A full documentation (in Chinese) can be found in docs/designing_report.doc.

Compilation

You can directly open hcc.cbp in Code::Blocks, check C++11 standard support in "Compiler settings", and click build button.

Or, you can run following commands in your shell.

g++ -std=c++11  -c actvar.cpp     -o obj\Debug\actvar.o
g++ -std=c++11  -c genasm.cpp     -o obj\Debug\genasm.o
g++ -std=c++11  -c lex.cpp        -o obj\Debug\lex.o
g++ -std=c++11  -c main.cpp       -o obj\Debug\main.o
g++ -std=c++11  -c opt.cpp        -o obj\Debug\opt.o
g++ -std=c++11  -c opt_dag.cpp    -o obj\Debug\opt_dag.o
g++ -std=c++11  -c semantics.cpp  -o obj\Debug\semantics.o
g++ -std=c++11  -c syntax.cpp     -o obj\Debug\syntax.o
g++ -std=c++11  -c util.cpp       -o obj\Debug\util.o
g++ -o bin\Debug\hcc.exe obj\Debug\*.o

Test samples

There is some test samples intests directory. For example, run bin\Debug\hcc.exe tests\3.fibo.c to get the x86 assembly code.

The usage of hcc.exe:

Usage: hcc [-c FILENAME|-l FILENAME|-s FILENAME|
            -t FILENAME] [-o] [-D] [-R] [-r] [-d] SOURCEFILE

       Read extended C0 source code from SOURCEFILE file and compile
       to MASM source code. The generated code will be output into
       std-out by default.

  -c FILENAME
       Output the generated code to FILENAME. Assign FILENAME as - to
       indicate std-out.

  -l FILENAME
       Only do lexical analysis and output the result to FILENAME.
       Assign FILENAME as - to indicate std-out.

  -s FILENAME
       Only do syntax analysis and output the result to FILENAME.
       Assign FILENAME as - to indicate std-out.

  -t FILENAME
       Do semantic analysis and dump the relevant tables to FILENAME.
       Assign FILENAME as - to indicate std-out.

  -o   Perform optimization.
  -D   Do NOT use DAG to simplify repeated expression.
  -R   Do NOT use reduce code using data flow, redundant deletion, copy-.
       spread, or constant combination.
  -r   Use global register.
  -d   Output verbose debug information.