glsl-lexer/src/lib.rs

//! A simple lexer for GLSL.
//!
//! Adheres to the GLSL 440. Read the spec
//! [here](https://registry.khronos.org/OpenGL/specs/gl/GLSLangSpec.4.40.pdf).
//! ## Example
//! ```
//! use glsl_lexer::*;
//!
//! fn main() {
//!     let source = r#"
//!         #version 440
//!         uniform float time;
//!         void main() {
//!             gl_FragColor = vec4(1.0, 0.5, 0.2, 1.0);
//!         }
//!     "#;
//!     let mut lexer = glsl_lexer::Lexer::new(&source);
//!     let tokens = lexer.get_tokens();
//!     dbg!("{}", tokens);
//! }
//! ```

//! # WIP THAT SHIT STILL WONKY AF

use std::sync::Arc;

mod handlers;

#[derive(Debug, PartialEq, Clone)]
pub enum Token {
    /// Something like `float`
    Identifier(String),
    /// Something like `uniform`
    Keyword(String),
    /// Something like `13`
    IntegerLiteral(i64),
    /// Something like `3.5` or `.5`
    FloatLiteral(f64),
    /// Something like `+`
    Operator(String),
    /// Something like `{`
    Symbol(char),
    /// Should be self-explanatory
    Whitespace,
    /// Something like `// uwu`
    Comment(String),
    /// Shrouded in mystery
    Unknown(char),
    /// End Of File
    EOF,
}

pub struct Lexer {
    /// GLSL source
    pub input: Vec<char>,
    /// Position in source
    pub position: usize,
    /// [`char`] under position
    pub current_char: Option<char>,
}

/// Instantiates Lexer and retrieves Tokens of given source.
/// Returns Arc<[Token]>
/// # Example:
/// ```
/// let lexed: Arc<[Token]> = lex!(r#"
///         #version 440
///         uniform float time;
///         void main() {
///             gl_FragColor = vec4(1.0, 0.5, 0.2, 1.0);
///         }
/// "#);
/// ```
/// Equivalent to:
/// ```
/// let source  = "some source";
/// let lexed = Lexer::new(source).get_tokens();
/// ```
#[macro_export]
macro_rules! lex {
    ($source:expr) => {{
        $crate::Lexer::get_tokens(&mut $crate::Lexer::new($source))
    }};
}

impl Lexer {
    /// Instantiates the [`Lexer`]
    pub fn new(input: &str) -> Self {
        let mut lexer = Lexer {
            input: input.chars().collect(),
            position: 0,
            current_char: None,
        };
        lexer.current_char = if lexer.position < lexer.input.len() {
            Some(lexer.input[lexer.position])
        } else {
            None
        };
        lexer
    }

    fn advance(&mut self) {
        self.position += 1;
        self.current_char = if self.position < self.input.len() {
            Some(self.input[self.position])
        } else {
            None
        };
    }

    /// Parses the source given the [`Lexer`] upon initialization and returns a vector of [`Token`].
    /// # Example:
    /// ```
    /// use glsl_lexer::*;
    /// let source = r#"
    ///     #version 440
    ///     uniform float time;
    ///     void main() {
    ///         gl_FragColor = vec4(1.0, 0.5, 0.2, 1.0);
    ///     }
    /// "#;
    /// let mut lexer = glsl_lexer::Lexer::new(&source);
    /// let tokens = lexer.get_tokens();
    /// dbg!("{}", tokens);
    ///```
    // We are using Arc<[Token]> as return type for cheaper cloning of the returned value
    pub fn get_tokens(&mut self) -> Arc<[Token]> {
        let mut tokens = Vec::new();
        while let Some(c) = self.current_char {
            if c.is_whitespace() {
                self.consume_whitespace();
                tokens.push(Token::Whitespace);
            } else if c.is_alphabetic() || c == '_' {
                tokens.push(self.consume_identifier_or_keyword());
            } else if c.is_ascii_digit() {
                tokens.push(self.consume_number());
            } else if c == '/' && self.peek() == Some('/') {
                tokens.push(self.consume_comment());
            } else {
                match c {
                    // TODO Implement operands like +=
                    '+' | '-' | '*' | '/' | '%' | '&' | '|' | '^' | '!' | '=' | '<' | '>' | '?' => {
                        tokens.push(self.consume_operator());
                    }

                    '{' | '}' | '(' | ')' | '#' | ',' | ';' => {
                        tokens.push(self.consume_symbol());
                    }

                    '.' => {
                        tokens.push(self.consume_number());
                    }

                    _ => {
                        tokens.push(self.consume_unknown());
                    }
                }
            }
        }
        tokens.push(Token::EOF);
        let ret: Arc<[Token]> = tokens.into();
        ret
    }

    fn peek(&self) -> Option<char> {
        if self.position + 1 < self.input.len() {
            Some(self.input[self.position + 1])
        } else {
            None
        }
    }

    // fn error(&self, message: &str) -> ! {
    //     panic!("Lexer error at position {}: {}", self.position, message);
    // }
}

fn is_keyword(word: &str) -> bool {
    matches!(
        word,
        "void"
            | "int"
            | "float"
            | "bool"
            | "if"
            | "else"
            | "for"
            | "while"
            | "return"
            | "struct"
            | "uniform"
            | "varying"
    )
}

#[cfg(test)]
mod tests {
    use super::*;
    use log::info;

    fn init() {
        std::env::set_var("RUST_LOG", "INFO");
        let _ = env_logger::builder().is_test(true).try_init();
    }

    #[test]
    fn whitespace() {
        init();
        let source_code = " \t\n";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[Whitespace] Tokens: {:#?}", tokens);
        assert_eq!(tokens, vec![Token::Whitespace, Token::EOF].into());
    }

    #[test]
    fn identifier() {
        init();
        let source_code = "variableName";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[Identifier] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![Token::Identifier("variableName".to_string()), Token::EOF].into()
        );
    }

    #[test]
    fn keyword() {
        init();
        let source_code = "uniform";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[Keyword] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![Token::Keyword("uniform".to_string()), Token::EOF].into()
        );
    }

    #[test]
    fn integer_literal() {
        init();
        let source_code = "12345";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[IntegerLiteral] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![Token::IntegerLiteral(12345), Token::EOF].into()
        );
    }

    #[test]
    fn float_literal() {
        init();
        let source_code = "123.4504";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[FloatLiteral] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![Token::FloatLiteral(123.4504), Token::EOF].into()
        );
    }

    #[test]
    fn float_shorthand() {
        init();
        let source_code = ".4504";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[FloatLiteral Shorthand] Tokens: {:#?}", tokens);
        assert_eq!(tokens, vec![Token::FloatLiteral(0.4504), Token::EOF].into());
    }

    #[test]
    fn swizzling() {
        init();
        let source_code = "abcd.xyz";
        let mut lexer = Lexer::new(source_code);
        let tokens = lexer.get_tokens();
        info!("[Swizzling] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![
                Token::Identifier("abcd".to_string()),
                Token::Symbol('.'),
                Token::Identifier("xyz".to_string()),
                Token::EOF
            ]
            .into()
        );
    }

    #[test]
    fn test_operator() {
        init();
        let source = "+-*/%&|^!=<>?";
        let mut lexer = Lexer::new(source);
        let tokens = lexer.get_tokens();
        info!("[Operator] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![
                Token::Operator("+".to_string()),
                Token::Operator("-".to_string()),
                Token::Operator("*".to_string()),
                Token::Operator("/".to_string()),
                Token::Operator("%".to_string()),
                Token::Operator("&".to_string()),
                Token::Operator("|".to_string()),
                Token::Operator("^".to_string()),
                Token::Operator("!".to_string()),
                Token::Operator("=".to_string()),
                Token::Operator("<".to_string()),
                Token::Operator(">".to_string()),
                Token::Operator("?".to_string()),
                Token::EOF,
            ]
            .into()
        );
    }

    #[test]
    fn test_single_line_comment() {
        init();
        let source = "// This is a comment\n";
        let mut lexer = Lexer::new(source);
        let tokens = lexer.get_tokens();
        info!("[Comment] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![
                Token::Comment("// This is a comment".to_string()),
                Token::Whitespace,
                Token::EOF,
            ]
            .into()
        );
    }

    // I hope that does it. Writing this test was pain.
    #[test]
    fn complex_source() {
        init();
        let source = r#"
            uniform float time;
            void main() {
                gl_FragColor = vec4(1.0, 0.5, 0.2, 1.0);
            }
        "#;
        let mut lexer = Lexer::new(source);
        let tokens = lexer.get_tokens();
        info!("[Complex Source] Tokens: {:#?}", tokens);
        assert_eq!(
            tokens,
            vec![
                Token::Whitespace,
                Token::Keyword("uniform".to_string()),
                Token::Whitespace,
                Token::Keyword("float".to_string()),
                Token::Whitespace,
                Token::Identifier("time".to_string()),
                Token::Symbol(';'),
                Token::Whitespace,
                Token::Keyword("void".to_string()),
                Token::Whitespace,
                Token::Identifier("main".to_string()),
                Token::Symbol('('),
                Token::Symbol(')'),
                Token::Whitespace,
                Token::Symbol('{'),
                Token::Whitespace,
                Token::Identifier("gl_FragColor".to_string()),
                Token::Whitespace,
                Token::Operator('='.to_string()),
                Token::Whitespace,
                Token::Identifier("vec4".to_string()),
                Token::Symbol('('),
                Token::FloatLiteral(1.0),
                Token::Symbol(','),
                Token::Whitespace,
                Token::FloatLiteral(0.5),
                Token::Symbol(','),
                Token::Whitespace,
                Token::FloatLiteral(0.2),
                Token::Symbol(','),
                Token::Whitespace,
                Token::FloatLiteral(1.0),
                Token::Symbol(')'),
                Token::Symbol(';'),
                Token::Whitespace,
                Token::Symbol('}'),
                Token::Whitespace,
                Token::EOF,
            ]
            .into()
        );
    }
}