msp430-repl/src/parser.rs

// © 2023 John Breaux
//! Parses [Tokens](crate::Token) into an [abstract syntax tree](Root)

use crate::{Error, Hash, TokenStream, Type};
use std::fmt::{Debug, Display, LowerHex};

pub mod preamble {
    //! All the different AST node types
    use super::*;
    // Traits
    pub use parsable::Parsable;

    pub use comment::Comment;
    pub use directive::Directive;
    pub use identifier::Identifier;
    pub use instruction::{
        encoding::{
            encoding_parser::EncodingParser, jump_target::JumpTarget, number::Number, primary_operand::PrimaryOperand,
            register::Register, secondary_operand::SecondaryOperand, width::Width, Encoding,
        },
        opcode::Opcode,
        Instruction,
    };
    pub use label::Label;
    pub use line::Line;
    pub use root::Root;
}
use preamble::*;

pub(crate) mod parsable;

pub(crate) mod comment;
pub(crate) mod directive;
pub(crate) mod identifier;
pub(crate) mod instruction;
pub(crate) mod label;

pub(crate) mod line {
    // © 2023 John Breaux
    use super::*;

    /// A line is one of:
    /// - [`Label`] (definition)
    /// - [`Instruction`]
    /// - [`Directive`]
    /// - [`Comment`]
    #[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
    pub enum Line {
        Empty,
        Insn(Instruction),
        Comment(Comment),
        Directive(Directive),
        Label(Label), // TODO: Label resolution
        EndOfFile,    // Expected end of file
    }

    impl Parsable for Line {
        fn parse<'text, T>(p: &Parser, stream: &mut T) -> Result<Self, Error>
        where T: TokenStream<'text> {
            if let Ok(token) = stream.peek_expect_any_of([Type::Insn, Type::Comment, Type::Directive, Type::Identifier])
            {
                return Ok(match token.variant() {
                    Type::Insn => Self::Insn(Instruction::parse(p, stream)?),
                    Type::Comment => Self::Comment(Comment::parse(p, stream)?),
                    Type::Directive => Self::Directive(Directive::parse(p, stream)?),
                    Type::Identifier => Self::Label(Label::parse(p, stream)?),
                    _ => unreachable!(),
                });
            }
            let token = stream.expect_any_of([Type::EndOfFile])?;
            Ok(match token.variant() {
                Type::EndOfFile => Self::EndOfFile,
                _ => unreachable!(),
            })
        }
    }
    impl Display for Line {
        fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
            match self {
                Self::Empty => writeln!(f, "\n"),
                Self::Label(arg0) => Display::fmt(arg0, f),
                Self::Insn(arg0) => Display::fmt(arg0, f),
                Self::Directive(arg0) => Display::fmt(arg0, f),
                Self::Comment(arg0) => Display::fmt(arg0, f),
                Self::EndOfFile => write!(f, "; End of file."),
            }
        }
    }
    impl LowerHex for Line {
        fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
            match self {
                Line::Insn(arg0) => LowerHex::fmt(arg0, f),
                _ => Ok(()),
            }
        }
    }
}

pub(crate) mod root {
    // © 2023 John Breaux
    use super::*;

    /// Contains the entire AST
    #[derive(Clone, Default, PartialEq, Eq, PartialOrd, Ord, Hash)]
    pub struct Root(pub Vec<Line>);

    // TODO: Get data out of ParseTree
    // TODO: Maybe implement some sort of follower
    impl Parsable for Root {
        fn parse<'text, T>(p: &Parser, stream: &mut T) -> Result<Self, Error>
        where T: TokenStream<'text> {
            let mut lines = vec![];
            loop {
                match Line::parse(p, stream) {
                    Ok(Line::EndOfFile) => break,
                    Ok(line) => lines.push(line),
                    Err(e) => {
                        let ret = Self(lines);
                        eprintln!("{ret}");
                        eprintln!("Error:{e}\n");
                        eprint!("Remaining:");
                        stream.for_each(|t| eprint!("{t}"));
                        eprintln!();
                        return Err(Error::ParseError(ret, Box::new(e)));
                    }
                }
            }
            Ok(Root(lines))
        }
    }

    impl Display for Root {
        fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
            for line in self.0.iter() {
                f.pad(&format!("{line} "))?;
            }
            Ok(())
        }
    }
    impl LowerHex for Root {
        fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
            for line in self.0.iter() {
                LowerHex::fmt(line, f)?;
            }
            Ok(())
        }
    }
    impl Debug for Root {
        fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
            for line in self.0.iter() {
                Display::fmt(line, f)?;
                Debug::fmt(line, f)?;
            }
            Ok(())
        }
    }
}

/// The type for [Parser] callbacks
pub type EmitComment = Box<dyn FnMut(&str)>;
pub type DefineLabel = Box<dyn FnMut(&Identifier) -> Result<(), Error>>;

pub struct Parser {
    radix: u32,
    // TODO: callbacks for emitted token sequences?!
    on_label: Option<DefineLabel>,
    on_comment: Option<EmitComment>,
}

impl Parser {
    pub fn parse_with<'t, T>(self, stream: &'t mut T) -> Result<Root, Error>
    where T: TokenStream<'t> {
        Root::parse(&self, &mut stream.ignore_spaces())
    }
    pub fn parse<T>(self, input: &T) -> Result<Root, Error>
    where T: AsRef<str> + ?Sized {
        Root::parse(&self, &mut super::Tokenizer::new(input).ignore_spaces())
    }
    pub fn parse_one<T>(self, input: &T) -> Result<Line, Error>
    where T: AsRef<str> + ?Sized {
        Line::parse(&self, &mut super::Tokenizer::new(input).ignore_spaces())
    }

    /// Sets the default radix for [Token](crate::tokenizer::token::Token) -> [Number]
    /// conversion
    pub fn radix(mut self, radix: u32) { self.radix = radix; }

    /// Inform the caller of a new identifier definition
    pub fn define_label(&mut self, l: &Identifier) -> Result<(), Error> {
        match self.on_label.as_mut() {
            Some(f) => f(l),
            _ => Ok(()),
        }
    }
    /// Inform the caller of an identifier being used
    pub fn emit_comment(&mut self, d: &str) {
        if let Some(f) = self.on_comment.as_mut() {
            f(d)
        }
    }
}

impl Default for Parser {
    fn default() -> Self { Self { radix: 16, on_label: None, on_comment: None } }
}

impl Debug for Parser {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        f.debug_struct("Parser").field("radix", &self.radix).finish_non_exhaustive()
    }
}