# UniAST

**Repository Path**: balababa/uni-ast

## Basic Information

- **Project Name**: UniAST
- **Description**: Define a unified abstract syntax tree （AST）to quickly generate code. 
- **Primary Language**: C++
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-02-18
- **Last Updated**: 2025-02-22

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# What is this?

Using this tool, you can simply write the following grammar to generate a simple compiler.

```
module: statement+;

statement: assign_stmt<assign_stmt>;

assign_stmt: variable_access<variable_access> Assign expression<expression>;

variable_access
    : variable_name<name_access> 
    ;

...
```

The generated compiler will be able to analysis and generate code for input:

```
a = 1 + 2
b = a + 3
```


# How to use?

## 1. install java sdk

```bash
$ sudo apt install openjdk-17-jdk
```

## 2. build llvm-project

```bash
$ git submodule update --init
$ cd thirdparty/LLVM && mkdir build && cd build
$ cmake ../llvm \
    -DLLVM_ENABLE_PROJECTS="mlir" \
    -DLLVM_ENABLE_ASSERTIONS=ON \
    -DLLVM_ENABLE_RTTI=ON \
    -DCMAKE_BUILD_TYPE=RELEASE
$ make -j 32
```

## 3. cmake and make

```bash
$ mkdir build && cd build
$ cmake ..
$ make -j 32
```
## 4. generate grammar, listener, CMakeLists.txt for TestGrammar

```bash
$ ./build/bin/uni-ast ./example/TestGrammar.test -o ./example/test
```

## 5. build TestGrammar and test it~
```bash
$ cd build
$ make -j 32
$ ./bin/TestGrammar_exe ../example/input.txt
```

## 6. Example output

The first part is generated AST:

* `<>` are source code content of AST, `<generated>` means this node is generate by extra decl after parser grammar, such as `{expr_params = expr_param}`
* `root` `assign_stmt`... are predefined AST Node, you can expand this list by add register visitor by using function `REGISTER_FUNCTION` in `BasicVisitor`.


The second part is debug info, you can change log info by function `LOGGER.setLogLevel`, the default level is `INFO`, and it is setted in constructor of `BasicVisitor`.

The third part is mlir code generated by `BasicVisitor`

```
root<generated>(assign_stmt<a=1+2>(variable_access<a>(name_access<a>(TEXT<a> ) ) expression<1+2>(expr_params<generated>(expr_param<1>(literal<1>(NUMBER<1> ) ) expr_param<2>(literal<2>(NUMBER<2> ) expr_params<generated> ) ) ) ) assign_stmt<b=a+3>(variable_access<b>(name_access<b>(TEXT<b> ) ) expression<a+3>(expr_params<generated>(expr_param<a>(variable_access<a>(name_access<a>(TEXT<a> ) ) ) expr_param<3>(literal<3>(NUMBER<3> ) expr_params<generated> ) ) ) ) )
[2025-02-22 10:34:18] [DEBUG] visit Assign_stmt. 
[2025-02-22 10:34:18] [DEBUG] visit Variable_access. 
[2025-02-22 10:34:18] [DEBUG] visit Name_access. 
[2025-02-22 10:34:18] [DEBUG] visit TEXT. 
[2025-02-22 10:34:18] [DEBUG] visit Expression. 
[2025-02-22 10:34:18] [DEBUG] visit Expr_params. 
[2025-02-22 10:34:18] [DEBUG] visit Expr_param. 
[2025-02-22 10:34:18] [DEBUG] visit literal. 
[2025-02-22 10:34:18] [DEBUG] visit NUMBER. 
[2025-02-22 10:34:18] [DEBUG] visit Expr_param. 
[2025-02-22 10:34:18] [DEBUG] visit literal. 
[2025-02-22 10:34:18] [DEBUG] visit NUMBER. 
[2025-02-22 10:34:18] [DEBUG] visit Op. 
[2025-02-22 10:34:18] [DEBUG] visit add. 
[2025-02-22 10:34:18] [DEBUG] visit Assign_stmt. 
[2025-02-22 10:34:18] [DEBUG] visit Variable_access. 
[2025-02-22 10:34:18] [DEBUG] visit Name_access. 
[2025-02-22 10:34:18] [DEBUG] visit TEXT. 
[2025-02-22 10:34:18] [DEBUG] visit Expression. 
[2025-02-22 10:34:18] [DEBUG] visit Expr_params. 
[2025-02-22 10:34:18] [DEBUG] visit Expr_param. 
[2025-02-22 10:34:18] [DEBUG] visit Variable_access. 
[2025-02-22 10:34:18] [DEBUG] visit Name_access. 
[2025-02-22 10:34:18] [DEBUG] visit TEXT. 
[2025-02-22 10:34:18] [DEBUG] visit Expr_param. 
[2025-02-22 10:34:18] [DEBUG] visit literal. 
[2025-02-22 10:34:18] [DEBUG] visit NUMBER. 
[2025-02-22 10:34:18] [DEBUG] visit Op. 
[2025-02-22 10:34:18] [DEBUG] visit add. 
module {
  func.func private @main() -> i32 {
    %c1_i32 = arith.constant 1 : i32
    %c2_i32 = arith.constant 2 : i32
    %0 = arith.addi %c1_i32, %c2_i32 : i32
    %alloca = memref.alloca() : memref<1xi32>
    %c0 = arith.constant 0 : index
    memref.store %0, %alloca[%c0] : memref<1xi32>
    %c1 = arith.constant 1 : index
    %1 = memref.load %alloca[%c1] : memref<1xi32>
    %c3_i32 = arith.constant 3 : i32
    %2 = arith.addi %1, %c3_i32 : i32
    %alloca_0 = memref.alloca() : memref<1xi32>
    %c0_1 = arith.constant 0 : index
    memref.store %2, %alloca_0[%c0_1] : memref<1xi32>
    %c0_i32 = arith.constant 0 : i32
    return %c0_i32 : i32
  }
}
```

# More

The entire job is divided into two parts:
* Generate unified AST (in directory `lib`, `include` and `tools`)
* Analysis unified AST and do code generation (in director `runtime`)