started work on c compiler

2025-11-14 23:36:51 +00:00
parent 2582ad10fa
commit fd5b305576
10 changed files with 1870 additions and 1 deletions
@@ -0,0 +1,13 @@
+use crate::parser::Program;
+
+pub struct CodeGenerator;
+
+impl CodeGenerator {
+    pub fn new(ast: Program) -> Self {
+        CodeGenerator {}
+    }
+
+    pub fn run(&mut self) -> Result<String, String> {
+        Ok(String::new())
+    }
+}
@@ -0,0 +1,265 @@
+// ============================================================================
+// Token Types
+// ============================================================================
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum TokenType {
+    // Keywords
+    Int,
+    If,
+    Else,
+    While,
+    Return,
+
+    // Identifiers and literals
+    Identifier(String),
+    Number(i32),
+
+    // Operators
+    Plus,
+    Minus,
+    Star,
+    Slash,
+    Assign,
+    Eq,
+    Ne,
+    Lt,
+    Gt,
+    Le,
+    Ge,
+
+    // Delimiters
+    LParen,
+    RParen,
+    LBrace,
+    RBrace,
+    Semicolon,
+    Comma,
+
+    Eof,
+}
+
+#[derive(Debug, Clone)]
+pub struct Token {
+    pub token_type: TokenType,
+    pub line: usize,
+    pub col: usize,
+}
+
+impl Token {
+    pub fn new(token_type: TokenType, line: usize, col: usize) -> Self {
+        Self {
+            token_type,
+            line,
+            col,
+        }
+    }
+}
+
+// ============================================================================
+// Lexer
+// ============================================================================
+
+pub struct Lexer {
+    source: Vec<char>,
+    pos: usize,
+    line: usize,
+    col: usize,
+}
+
+impl Lexer {
+    pub fn new(source: &str) -> Self {
+        Self {
+            source: source.chars().collect(),
+            pos: 0,
+            line: 1,
+            col: 1,
+        }
+    }
+
+    fn error(&self, msg: &str) -> String {
+        format!(
+            "Lexer error at line {}, col {}: {}",
+            self.line, self.col, msg
+        )
+    }
+
+    fn peek(&self, offset: usize) -> Option<char> {
+        self.source.get(self.pos + offset).copied()
+    }
+
+    fn advance(&mut self) -> Option<char> {
+        if self.pos >= self.source.len() {
+            return None;
+        }
+        let ch = self.source[self.pos];
+        self.pos += 1;
+        if ch == '\n' {
+            self.line += 1;
+            self.col = 1;
+        } else {
+            self.col += 1;
+        }
+        Some(ch)
+    }
+
+    fn skip_whitespace(&mut self) {
+        while let Some(ch) = self.peek(0) {
+            if ch.is_whitespace() {
+                self.advance();
+            } else {
+                break;
+            }
+        }
+    }
+
+    fn skip_comment(&mut self) {
+        if self.peek(0) == Some('/') && self.peek(1) == Some('/') {
+            while let Some(ch) = self.peek(0) {
+                if ch == '\n' {
+                    break;
+                }
+                self.advance();
+            }
+        }
+    }
+
+    fn read_number(&mut self) -> i32 {
+        let mut num_str = String::new();
+        while let Some(ch) = self.peek(0) {
+            if ch.is_ascii_digit() {
+                num_str.push(ch);
+                self.advance();
+            } else {
+                break;
+            }
+        }
+        num_str.parse().unwrap_or(0)
+    }
+
+    fn read_identifier(&mut self) -> String {
+        let mut ident = String::new();
+        while let Some(ch) = self.peek(0) {
+            if ch.is_alphanumeric() || ch == '_' {
+                ident.push(ch);
+                self.advance();
+            } else {
+                break;
+            }
+        }
+        ident
+    }
+
+    pub fn tokenize(&mut self) -> Result<Vec<Token>, String> {
+        let mut tokens = Vec::new();
+
+        loop {
+            self.skip_whitespace();
+            self.skip_comment();
+
+            if self.pos >= self.source.len() {
+                break;
+            }
+
+            let line = self.line;
+            let col = self.col;
+            let ch = self.peek(0).unwrap();
+
+            let token_type = if ch.is_ascii_digit() {
+                let num = self.read_number();
+                TokenType::Number(num)
+            } else if ch.is_alphabetic() || ch == '_' {
+                let ident = self.read_identifier();
+                match ident.as_str() {
+                    "int" => TokenType::Int,
+                    "if" => TokenType::If,
+                    "else" => TokenType::Else,
+                    "while" => TokenType::While,
+                    "return" => TokenType::Return,
+                    _ => TokenType::Identifier(ident),
+                }
+            } else {
+                match ch {
+                    '=' if self.peek(1) == Some('=') => {
+                        self.advance();
+                        self.advance();
+                        TokenType::Eq
+                    }
+                    '!' if self.peek(1) == Some('=') => {
+                        self.advance();
+                        self.advance();
+                        TokenType::Ne
+                    }
+                    '<' if self.peek(1) == Some('=') => {
+                        self.advance();
+                        self.advance();
+                        TokenType::Le
+                    }
+                    '>' if self.peek(1) == Some('=') => {
+                        self.advance();
+                        self.advance();
+                        TokenType::Ge
+                    }
+                    '+' => {
+                        self.advance();
+                        TokenType::Plus
+                    }
+                    '-' => {
+                        self.advance();
+                        TokenType::Minus
+                    }
+                    '*' => {
+                        self.advance();
+                        TokenType::Star
+                    }
+                    '/' => {
+                        self.advance();
+                        TokenType::Slash
+                    }
+                    '=' => {
+                        self.advance();
+                        TokenType::Assign
+                    }
+                    '<' => {
+                        self.advance();
+                        TokenType::Lt
+                    }
+                    '>' => {
+                        self.advance();
+                        TokenType::Gt
+                    }
+                    '(' => {
+                        self.advance();
+                        TokenType::LParen
+                    }
+                    ')' => {
+                        self.advance();
+                        TokenType::RParen
+                    }
+                    '{' => {
+                        self.advance();
+                        TokenType::LBrace
+                    }
+                    '}' => {
+                        self.advance();
+                        TokenType::RBrace
+                    }
+                    ';' => {
+                        self.advance();
+                        TokenType::Semicolon
+                    }
+                    ',' => {
+                        self.advance();
+                        TokenType::Comma
+                    }
+                    _ => return Err(self.error(&format!("Unexpected character: {}", ch))),
+                }
+            };
+
+            tokens.push(Token::new(token_type, line, col));
+        }
+
+        tokens.push(Token::new(TokenType::Eof, self.line, self.col));
+        Ok(tokens)
+    }
+}
@@ -0,0 +1,72 @@
+use std::fmt;
+
+use crate::{codegen::CodeGenerator, lexer::Lexer, parser::Parser};
+
+pub mod codegen;
+pub mod lexer;
+pub mod parser;
+
+// ============================================================================
+// Main & Tests
+// ============================================================================
+
+fn main() {
+    // read from input file: syntax "c_compiler <src.c> [output.dsa]"
+    let args: Vec<String> = std::env::args().collect();
+    if args.len() < 2 {
+        eprintln!("Usage: c_compiler <src.c> [output.dsa]");
+        return;
+    }
+
+    let input_file = &args[1];
+    let output_file = if args.len() > 2 {
+        &args[2]
+    } else {
+        "output.dsa"
+    };
+
+    // read input
+    let input = std::fs::read_to_string(input_file).expect("Failed to read input file");
+
+    // Lexing
+    let mut lexer = Lexer::new(&input);
+    let tokens = match lexer.tokenize() {
+        Ok(tokens) => tokens,
+        Err(e) => {
+            eprintln!("Lexing error: {}", e);
+            return;
+        }
+    };
+
+    println!("Tokens:");
+    for token in &tokens {
+        println!("  {:?}", token.token_type);
+    }
+    println!();
+
+    // Parsing
+    let mut parser = Parser::new(tokens);
+    let ast = match parser.parse() {
+        Ok(ast) => ast,
+        Err(e) => {
+            eprintln!("Parsing error: {}", e);
+            return;
+        }
+    };
+
+    println!("AST:");
+    println!("{:#?}", ast);
+
+    // Code Gen
+    let mut generator = CodeGenerator::new(ast);
+    let result = match generator.run() {
+        Ok(code) => code,
+        Err(e) => {
+            eprintln!("Parsing error: {}", e);
+            return;
+        }
+    };
+
+    println!("CODE:");
+    println!("{:#?}", result);
+}
@@ -0,0 +1,558 @@
+// ============================================================================
+// AST Node Types
+// ============================================================================
+
+use std::fmt;
+
+use crate::lexer::{Token, TokenType};
+
+#[derive(Debug, Clone)]
+pub struct Program {
+    pub declarations: Vec<Declaration>,
+}
+
+#[derive(Debug, Clone)]
+pub enum Declaration {
+    Function {
+        name: String,
+        return_type: Type,
+        params: Vec<Parameter>,
+        body: Statement,
+    },
+    Variable {
+        name: String,
+        init: Option<Expression>,
+    },
+}
+
+#[derive(Debug, Clone)]
+pub struct Parameter {
+    pub name: String,
+    pub param_type: Type,
+}
+
+#[derive(Debug, Clone)]
+pub enum Type {
+    Int,
+    Long,
+    Float,
+    Double,
+    Char,
+    Void,
+    Ptr(Box<Type>),
+    Array(Box<Type>, usize),
+    Struct(String),
+}
+
+#[derive(Debug, Clone)]
+pub enum Statement {
+    Compound {
+        statements: Vec<Statement>,
+    },
+    Assign {
+        // left side
+        name: String,
+        declare_type: Option<Type>,
+
+        // right side
+        value: Option<Box<Expression>>,
+    },
+    Expression {
+        expr: Expression,
+    },
+    If {
+        condition: Expression,
+        then_stmt: Box<Statement>,
+        else_stmt: Option<Box<Statement>>,
+    },
+    While {
+        condition: Expression,
+        body: Box<Statement>,
+    },
+    Return {
+        expr: Option<Expression>,
+    },
+}
+
+#[derive(Debug, Clone)]
+pub enum Expression {
+    Empty,
+    Binary {
+        op: BinaryOperator,
+        left: Box<Expression>,
+        right: Box<Expression>,
+    },
+    Unary {
+        op: UnaryOperator,
+        operand: Box<Expression>,
+    },
+    Variable {
+        name: String,
+        expr_type: Option<Type>,
+    },
+    Number {
+        value: i32,
+    },
+    Call {
+        name: String,
+        args: Vec<Expression>,
+    },
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum BinaryOperator {
+    Add,
+    Sub,
+    Mul,
+    Div,
+    Eq,
+    Ne,
+    Lt,
+    Gt,
+    Le,
+    Ge,
+}
+
+impl fmt::Display for BinaryOperator {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        match self {
+            BinaryOperator::Add => write!(f, "+"),
+            BinaryOperator::Sub => write!(f, "-"),
+            BinaryOperator::Mul => write!(f, "*"),
+            BinaryOperator::Div => write!(f, "/"),
+            BinaryOperator::Eq => write!(f, "=="),
+            BinaryOperator::Ne => write!(f, "!="),
+            BinaryOperator::Lt => write!(f, "<"),
+            BinaryOperator::Gt => write!(f, ">"),
+            BinaryOperator::Le => write!(f, "<="),
+            BinaryOperator::Ge => write!(f, ">="),
+        }
+    }
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum UnaryOperator {
+    Plus,
+    Minus,
+}
+
+impl fmt::Display for UnaryOperator {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        match self {
+            UnaryOperator::Plus => write!(f, "+"),
+            UnaryOperator::Minus => write!(f, "-"),
+        }
+    }
+}
+
+// ============================================================================
+// Parser
+// ============================================================================
+
+pub struct Parser {
+    tokens: Vec<Token>,
+    pos: usize,
+}
+
+impl Parser {
+    pub fn new(tokens: Vec<Token>) -> Self {
+        Self { tokens, pos: 0 }
+    }
+
+    fn error(&self, msg: &str) -> String {
+        let token = self.current();
+        format!(
+            "Parser error at line {}, col {}: {}",
+            token.line, token.col, msg
+        )
+    }
+
+    fn current(&self) -> &Token {
+        self.tokens
+            .get(self.pos)
+            .unwrap_or_else(|| self.tokens.last().unwrap())
+    }
+
+    fn peek(&self, offset: usize) -> &Token {
+        self.tokens
+            .get(self.pos + offset)
+            .unwrap_or_else(|| self.tokens.last().unwrap())
+    }
+
+    fn advance(&mut self) -> &Token {
+        if self.pos < self.tokens.len() - 1 {
+            self.pos += 1;
+        }
+        self.current()
+    }
+
+    fn expect(&mut self, expected: TokenType) -> Result<Token, String> {
+        let token = self.current().clone();
+        if std::mem::discriminant(&token.token_type) != std::mem::discriminant(&expected)
+        {
+            return Err(self.error(&format!(
+                "Expected {:?}, got {:?}",
+                expected, token.token_type
+            )));
+        }
+        self.advance();
+        Ok(token)
+    }
+
+    pub fn parse(&mut self) -> Result<Program, String> {
+        let mut declarations = Vec::new();
+
+        while !matches!(self.current().token_type, TokenType::Eof) {
+            declarations.push(self.parse_declaration()?);
+        }
+
+        Ok(Program { declarations })
+    }
+
+    fn parse_declaration(&mut self) -> Result<Declaration, String> {
+        self.expect(TokenType::Int)?;
+
+        let name = match &self.current().token_type {
+            TokenType::Identifier(s) => s.clone(),
+            _ => return Err(self.error("Expected identifier")),
+        };
+        self.advance();
+
+        match &self.current().token_type {
+            TokenType::LParen => {
+                // Function declaration
+                self.advance();
+                let mut params = Vec::<Parameter>::new();
+
+                if !matches!(self.current().token_type, TokenType::RParen) {
+                    self.expect(TokenType::Int)?;
+
+                    match &self.current().token_type {
+                        TokenType::Identifier(s) => {
+                            params.push(Parameter {
+                                name: s.clone(),
+                                param_type: Type::Int,
+                            });
+                            self.advance();
+                        }
+                        _ => return Err(self.error("Expected parameter name")),
+                    }
+
+                    while matches!(self.current().token_type, TokenType::Comma) {
+                        self.advance();
+                        self.expect(TokenType::Int)?;
+
+                        match &self.current().token_type {
+                            TokenType::Identifier(s) => {
+                                params.push(Parameter {
+                                    name: s.clone(),
+                                    param_type: Type::Int,
+                                });
+                                self.advance();
+                            }
+                            _ => return Err(self.error("Expected parameter name")),
+                        }
+                    }
+                }
+
+                self.expect(TokenType::RParen)?;
+                let body = self.parse_compound_stmt()?;
+
+                Ok(Declaration::Function {
+                    name,
+                    params,
+                    body,
+                    return_type: Type::Int,
+                })
+            }
+            _ => {
+                // Variable declaration
+                let init = if matches!(self.current().token_type, TokenType::Assign) {
+                    self.advance();
+                    Some(self.parse_expression()?)
+                } else {
+                    None
+                };
+
+                self.expect(TokenType::Semicolon)?;
+                Ok(Declaration::Variable { name, init })
+            }
+        }
+    }
+
+    fn parse_compound_stmt(&mut self) -> Result<Statement, String> {
+        self.expect(TokenType::LBrace)?;
+        let mut statements = Vec::new();
+
+        while !matches!(self.current().token_type, TokenType::RBrace) {
+            statements.push(self.parse_statement()?);
+        }
+
+        self.expect(TokenType::RBrace)?;
+        Ok(Statement::Compound { statements })
+    }
+
+    fn parse_statement(&mut self) -> Result<Statement, String> {
+        match &self.current().token_type {
+            TokenType::LBrace => Ok(self.parse_compound_stmt()?),
+            TokenType::If => self.parse_if_stmt(),
+            TokenType::While => self.parse_while_stmt(),
+            TokenType::Return => self.parse_return_stmt(),
+            TokenType::Identifier(name) => {
+                let name = name.clone();
+
+                // peek ahead for open paren (func call expr)
+                if matches!(self.peek(1).token_type, TokenType::LParen) {
+                    let expr = self.parse_expression()?; // a function call expr
+                    self.expect(TokenType::Semicolon)?;
+                    return Ok(Statement::Expression { expr });
+                }
+
+                self.advance(); // advance past identifier
+
+                // assignment expression
+                if matches!(self.current().token_type, TokenType::Assign) {
+                    self.advance();
+                    let expr = self.parse_expression()?;
+
+                    self.expect(TokenType::Semicolon)?;
+                    Ok(Statement::Assign {
+                        name,
+                        value: Some(Box::new(expr)),
+                        declare_type: None,
+                    })
+                }
+                // var expression
+                else {
+                    self.expect(TokenType::Semicolon)?;
+                    Ok(Statement::Expression {
+                        expr: Expression::Variable {
+                            name,
+                            expr_type: None,
+                        },
+                    })
+                }
+            }
+            TokenType::Int => {
+                // Local variable declaration
+                self.advance();
+                let name = match &self.current().token_type {
+                    TokenType::Identifier(s) => s.clone(),
+                    _ => return Err(self.error("Expected variable name")),
+                };
+                self.advance();
+
+                let init = if matches!(self.current().token_type, TokenType::Assign) {
+                    self.advance();
+                    Some(self.parse_expression()?)
+                } else {
+                    None
+                };
+
+                self.expect(TokenType::Semicolon)?;
+
+                // Convert to assignment expression statement
+                let expr = if let Some(init_expr) = init {
+                    Statement::Assign {
+                        name,
+                        value: Some(Box::new(init_expr)),
+                        declare_type: Some(Type::Int),
+                    }
+                } else {
+                    Statement::Assign {
+                        name,
+                        value: None,
+                        declare_type: Some(Type::Int),
+                    }
+                };
+
+                Ok(expr)
+            }
+            _ => {
+                let expr = if matches!(self.current().token_type, TokenType::Semicolon) {
+                    Expression::Empty
+                } else {
+                    self.parse_expression()?
+                };
+
+                self.expect(TokenType::Semicolon)?;
+                Ok(Statement::Expression { expr })
+            }
+        }
+    }
+
+    fn parse_if_stmt(&mut self) -> Result<Statement, String> {
+        self.expect(TokenType::If)?;
+        self.expect(TokenType::LParen)?;
+        let condition = self.parse_expression()?;
+        self.expect(TokenType::RParen)?;
+        let then_stmt = Box::new(self.parse_statement()?);
+
+        let else_stmt = if matches!(self.current().token_type, TokenType::Else) {
+            self.advance();
+            Some(Box::new(self.parse_statement()?))
+        } else {
+            None
+        };
+
+        Ok(Statement::If {
+            condition,
+            then_stmt,
+            else_stmt,
+        })
+    }
+
+    fn parse_while_stmt(&mut self) -> Result<Statement, String> {
+        self.expect(TokenType::While)?;
+        self.expect(TokenType::LParen)?;
+        let condition = self.parse_expression()?;
+        self.expect(TokenType::RParen)?;
+        let body = Box::new(self.parse_statement()?);
+
+        Ok(Statement::While { condition, body })
+    }
+
+    fn parse_return_stmt(&mut self) -> Result<Statement, String> {
+        self.expect(TokenType::Return)?;
+
+        let expr = if matches!(self.current().token_type, TokenType::Semicolon) {
+            None
+        } else {
+            Some(self.parse_expression()?)
+        };
+
+        self.expect(TokenType::Semicolon)?;
+        Ok(Statement::Return { expr })
+    }
+
+    fn parse_expression(&mut self) -> Result<Expression, String> {
+        self.parse_comparison()
+    }
+
+    fn parse_comparison(&mut self) -> Result<Expression, String> {
+        let mut expr = self.parse_additive()?;
+
+        while let Some(op) = match &self.current().token_type {
+            TokenType::Eq => Some(BinaryOperator::Eq),
+            TokenType::Ne => Some(BinaryOperator::Ne),
+            TokenType::Lt => Some(BinaryOperator::Lt),
+            TokenType::Gt => Some(BinaryOperator::Gt),
+            TokenType::Le => Some(BinaryOperator::Le),
+            TokenType::Ge => Some(BinaryOperator::Ge),
+            _ => None,
+        } {
+            self.advance();
+            let right = Box::new(self.parse_additive()?);
+            expr = Expression::Binary {
+                op,
+                left: Box::new(expr),
+                right,
+            };
+        }
+
+        Ok(expr)
+    }
+
+    fn parse_additive(&mut self) -> Result<Expression, String> {
+        let mut expr = self.parse_multiplicative()?;
+
+        while let Some(op) = match &self.current().token_type {
+            TokenType::Plus => Some(BinaryOperator::Add),
+            TokenType::Minus => Some(BinaryOperator::Sub),
+            _ => None,
+        } {
+            self.advance();
+            let right = Box::new(self.parse_multiplicative()?);
+            expr = Expression::Binary {
+                op,
+                left: Box::new(expr),
+                right,
+            };
+        }
+
+        Ok(expr)
+    }
+
+    fn parse_multiplicative(&mut self) -> Result<Expression, String> {
+        let mut expr = self.parse_unary()?;
+
+        while let Some(op) = match &self.current().token_type {
+            TokenType::Star => Some(BinaryOperator::Mul),
+            TokenType::Slash => Some(BinaryOperator::Div),
+            _ => None,
+        } {
+            self.advance();
+            let right = Box::new(self.parse_unary()?);
+            expr = Expression::Binary {
+                op,
+                left: Box::new(expr),
+                right,
+            };
+        }
+
+        Ok(expr)
+    }
+
+    fn parse_unary(&mut self) -> Result<Expression, String> {
+        let op = match &self.current().token_type {
+            TokenType::Plus => Some(UnaryOperator::Plus),
+            TokenType::Minus => Some(UnaryOperator::Minus),
+            _ => None,
+        };
+
+        if let Some(op) = op {
+            self.advance();
+            let operand = Box::new(self.parse_unary()?);
+            return Ok(Expression::Unary { op, operand });
+        }
+
+        self.parse_primary()
+    }
+
+    fn parse_primary(&mut self) -> Result<Expression, String> {
+        match &self.current().token_type.clone() {
+            TokenType::Number(n) => {
+                let value = *n;
+                self.advance();
+                Ok(Expression::Number { value })
+            }
+            TokenType::Identifier(name) => {
+                let name = name.clone();
+                self.advance();
+
+                if matches!(self.current().token_type, TokenType::LParen) {
+                    // Function call
+                    self.advance();
+                    let mut args = Vec::new();
+
+                    if !matches!(self.current().token_type, TokenType::RParen) {
+                        args.push(self.parse_expression()?);
+
+                        while matches!(self.current().token_type, TokenType::Comma) {
+                            self.advance();
+                            args.push(self.parse_expression()?);
+                        }
+                    }
+
+                    self.expect(TokenType::RParen)?;
+                    Ok(Expression::Call { name, args })
+                } else {
+                    Ok(Expression::Variable {
+                        name,
+                        expr_type: None,
+                    })
+                }
+            }
+            TokenType::LParen => {
+                self.advance();
+                let expr = self.parse_expression()?;
+                self.expect(TokenType::RParen)?;
+                Ok(expr)
+            }
+            _ => Err(self.error(&format!(
+                "Unexpected token: {:?}",
+                self.current().token_type
+            ))),
+        }
+    }
+}