crsn/crsn/src/asm/parse/arg_parser.rs

use sexp::{Sexp, SourcePosition};

use crate::asm::data::{Rd, RdData, RdObj, Wr, RdWr};
use crate::asm::error::CrsnError;
use crate::asm::parse::parse_data::{parse_rd, parse_wr, parse_value, parse_rdwr_offset, parse_rdwr, parse_rd_offset, parse_wr_offset};
use crate::asm::parse::ParserContext;
use crate::asm::parse::sexp_expect::expect_string_atom;
use crate::asm::data::literal::Value;
use crate::builtin::defs::BitMask;
use crate::asm::patches::ErrWithPos;
use std::fmt::{Debug, Formatter};
use std::fmt;

/// Utility for argument parsing
pub struct TokenParser<'a> {
    orig_len: usize,
    args: Vec<Sexp>,
    start_pos: &'a SourcePosition,
    pub pcx: &'a ParserContext<'a>,
}

impl<'a> Debug for TokenParser<'a> {
    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
        f.debug_struct("TokenParser")
            .field("orig_len", &self.orig_len)
            .field("args", &self.args)
            .field("start_pos", &self.start_pos)
            .finish()
    }
}

impl<'a> IntoIterator for TokenParser<'a> {
    type Item = Sexp;
    type IntoIter = std::vec::IntoIter<Sexp>;

    fn into_iter(mut self) -> Self::IntoIter {
        // The vec is reversed so we can call "pop", but here we will iterate it as a slice, so it
        // must be reversed to its original direction.
        self.args.reverse();
        self.args.into_iter()
    }
}

impl<'a> TokenParser<'a> {
    /// Create a new argument parser
    pub fn new(mut args: Vec<Sexp>, start_pos: &'a SourcePosition, pcx: &'a ParserContext) -> Self {
        args.reverse();
        Self {
            orig_len: args.len(),
            start_pos,
            args,
            pcx,
        }
    }

    /// Prepend a token
    pub fn prepend(&mut self, what: Sexp) {
        // the list is reversed - actually, append it
        self.args.push(what);
        self.orig_len += 1;
    }

    /// Get error if not empty.
    /// The argument is substituted into the phrase "Instruction needs ...!" - i.e. "one Wr argument and a list or string"
    pub fn ensure_empty(&self, what_arguments : &str) -> Result<(), CrsnError> {
        self.ensure_empty_custom(&format!("Instruction needs {}!", what_arguments))
    }

    pub fn ensure_empty_custom(&self, msg : &str) -> Result<(), CrsnError> {
        if self.have_more() {
            Err(CrsnError::Parse(msg.to_string().into(), self.start_pos.clone()))
        } else {
            Ok(())
        }
    }

    pub fn have_more(&self) -> bool {
        self.len() > 0
    }

    /// Get remaining items count
    pub fn len(&self) -> usize {
        self.args.len()
    }

    /// Get len before parsing started
    pub fn orig_len(&self) -> usize {
        self.orig_len
    }

    /// Check if parsing started
    pub fn parsing_started(&self) -> bool {
        self.args.len() != self.orig_len
    }

    /// Get the next entry
    pub fn next(&mut self) -> Option<Sexp> {
        self.args.pop()
    }

    /// Get the next entry, or raise an error
    pub fn next_or_err(&mut self) -> Result<Sexp, CrsnError> {
        match self.next() {
            None => {
                Err(CrsnError::Parse("Unexpected end of token list".into(), self.start_pos.clone()))
            }
            Some(removed) => Ok(removed)
        }
    }

    /// Look at the next entry
    pub fn peek(&mut self) -> Option<&Sexp> {
        self.args.last()
    }

    /// Get the next string entry
    pub fn next_string(&mut self) -> Result<(String, SourcePosition), CrsnError> {
        let next = self.next_or_err()?;
        let esa = expect_string_atom(next)?;
        Ok(esa)
    }

    /// Get the next value entry
    pub fn next_value(&mut self) -> Result<(Value, SourcePosition), CrsnError> {
        let next = self.next_or_err()?;
        parse_value(next, self.pcx)
    }

    /// Get the next entry as read location
    pub fn next_rd(&mut self) -> Result<Rd, CrsnError> {
        let next = self.next_or_err()?;
        parse_rd(next, self.pcx)
    }

    /// Get the next entry as an object location
    pub fn next_rdobj(&mut self) -> Result<RdObj, CrsnError> {
        match parse_rd(self.next_or_err()?, self.pcx)? {
            Rd(RdData::RegObject(reg)) => {
                Ok(RdObj::from_reg(reg))
            }
            other => {
                Err(CrsnError::Parse(
                    format!("Not a valid object handle syntax: {:?}", other).into(),
                    self.start_pos.clone(),
                ))
            }
        }
    }

    /// Get the next entry as write location
    pub fn next_wr(&mut self) -> Result<Wr, CrsnError> {
        parse_wr(self.next_or_err()?, self.pcx)
    }

    /// Get the next entry as write location
    pub fn next_rdwr(&mut self) -> Result<RdWr, CrsnError> {
        let next = self.next_or_err()?;
        parse_rdwr(next, self.pcx)
    }

    /// Get the next entry as write location with optional :offset
    pub fn next_rdwr_offset(&mut self) -> Result<(RdWr, u32), CrsnError> {
        let next = self.next_or_err()?;
        parse_rdwr_offset(next, self.pcx)
    }

    pub fn next_rd_offset(&mut self) -> Result<(Rd, u32), CrsnError> {
        let next = self.next_or_err()?;
        parse_rd_offset(next, self.pcx)
    }

    pub fn next_wr_offset(&mut self) -> Result<(Wr, u32), CrsnError> {
        let next = self.next_or_err()?;
        parse_wr_offset(next, self.pcx)
    }

    /// Parse combining binary instruction operands (i.e. add) with bit masks
    /// Accepts (Wr, Rd, Rd) and (RdWr, Rd)
    pub fn parse_masked_wr_rd_rd(&mut self, keyword: &str, prefix: &str) -> Result<Option<(Wr, Rd, Rd, BitMask)>, CrsnError> {
        if let Some(s) = keyword.strip_prefix(prefix) {
            let width = if s.is_empty() {
                (std::mem::size_of::<Value>() as u32) * 8
            } else {
                s.parse().err_pos(self.start_pos)?
            };
            if self.len() == 2 {
                let (rw, dst_pos) = self.next_rdwr_offset()?;
                let (rd, src_pos) = self.next_rd_offset()?;

                let mask = BitMask {
                    width,
                    dst_pos,
                    src_pos: dst_pos,
                    src2_pos: src_pos,
                };
                mask.validate(self.start_pos)?;

                Ok(Some((rw.wr(), rw.rd(), rd, mask)))
            } else if self.len() == 3 {
                let (wr, dst_pos) = self.next_wr_offset()?;
                let (rd1, src_pos) = self.next_rd_offset()?;
                let (rd2, src2_pos) = self.next_rd_offset()?;

                let mask = BitMask {
                    width,
                    dst_pos,
                    src_pos,
                    src2_pos,
                };
                mask.validate(self.start_pos)?;

                Ok(Some((wr, rd1, rd2, mask)))
            } else {
                Err(CrsnError::Parse("Instruction needs 2 (RW,Rd) or 3 (Wr,Rd,Rd) arguments!".into(), self.start_pos.clone()))
            }
        } else {
            Ok(None)
        }
    }


    /// Parse unary instruction operands (i.e. complement) with bit masks.
    /// Accepts (Wr, Rd) and (RdWr)
    pub fn parse_masked_wr_rd(&mut self, keyword: &str, prefix: &'static str) -> Result<Option<(Wr, Rd, BitMask)>, CrsnError> {
        if let Some(s) = keyword.strip_prefix(prefix) {
            let width = if s.is_empty() {
                (std::mem::size_of::<Value>() as u32) * 8
            } else {
                s.parse().err_pos(self.start_pos)?
            };
            if self.len() == 1 {
                let (rw, dst_pos) = self.next_rdwr_offset()?;

                let mask = BitMask {
                    width,
                    dst_pos,
                    src_pos: dst_pos,
                    src2_pos: 0,
                };
                mask.validate(self.start_pos)?;

                Ok(Some((rw.wr(), rw.rd(), mask)))
            } else if self.len() == 2 {
                let (wr, dst_pos) = self.next_wr_offset()?;
                let (rd, src_pos) = self.next_rd_offset()?;

                let mask = BitMask {
                    width,
                    dst_pos,
                    src_pos,
                    src2_pos: 0,
                };
                mask.validate(self.start_pos)?;

                Ok(Some((wr, rd, mask)))
            } else {
                Err(CrsnError::Parse("Instruction needs 1 (RW) or 2 (Wr,Rd) arguments!".into(), self.start_pos.clone()))
            }
        } else {
            Ok(None)
        }
    }

    /// Parse a swap-type binary instruction operands (i.e. exchange) with bit masks
    /// Accepts (RdWr, RdWr)
    pub fn parse_masked_rdwr_rdwr(&mut self, keyword: &str, prefix: &str) -> Result<Option<(RdWr, RdWr, BitMask)>, CrsnError> {
        if let Some(s) = keyword.strip_prefix(prefix) {
            let width = if s.is_empty() {
                (std::mem::size_of::<Value>() as u32) * 8
            } else {
                s.parse().err_pos(self.start_pos)?
            };
            if self.len() == 2 {
                let (wr1, dst_pos) = self.next_rdwr_offset()?;
                let (wr2, src_pos) = self.next_rdwr_offset()?;

                let mask = BitMask {
                    width,
                    dst_pos,
                    src_pos,
                    src2_pos: 0,
                };
                mask.validate(self.start_pos)?;

                Ok(Some((wr1, wr2, mask)))
            } else {
                Err(CrsnError::Parse("Instruction needs 2 (RW,RW) arguments!".into(), self.start_pos.clone()))
            }
        } else {
            Ok(None)
        }
    }

    /// Parse combining binary instruction operands (i.e. add) without masks
    /// Accepts (Wr, Rd, Rd) and (RdWr, Rd)
    pub fn parse_wr_rd_rd(&mut self) -> Result<(Wr, Rd, Rd), CrsnError> {
        if self.len() == 2 {
            let rw = self.next_rdwr()?;
            let rd = self.next_rd()?;
            Ok((rw.wr(), rw.rd(), rd))
        } else if self.len() == 3 {
            let wr = self.next_wr()?;
            let rd1 = self.next_rd()?;
            let rd2 = self.next_rd()?;
            Ok((wr, rd1, rd2))
        } else {
            Err(CrsnError::Parse("Instruction needs 2 (RW,Rd) or 3 (Wr,Rd,Rd) arguments!".into(), self.start_pos.clone()))
        }
    }

    /// Parse combining binary instruction operands (i.e. add) without masks
    /// Accepts (Wr, Rd, Rd) and (RdWr, Rd)
    pub fn parse_wr_rd_rd_or_n(&mut self, n : Value) -> Result<(Wr, Rd, Rd), CrsnError> {
        if self.len() == 1 {
            let rw = self.next_rdwr()?;
            Ok((rw.wr(), rw.rd(), Rd::new_imm(n)))
        } else if self.len() == 2 {
            let rw = self.next_rdwr()?;
            let rd = self.next_rd()?;
            Ok((rw.wr(), rw.rd(), rd))
        } else if self.len() == 3 {
            let wr = self.next_wr()?;
            let rd1 = self.next_rd()?;
            let rd2 = self.next_rd()?;
            Ok((wr, rd1, rd2))
        } else {
            Err(CrsnError::Parse("Instruction needs 1 (RW,1), 2 (RW,Rd) or 3 (Wr,Rd,Rd) arguments!".into(), self.start_pos.clone()))
        }
    }

    /// Parse unary instruction operands (i.e. complement) without masks
    /// Accepts (Wr, Rd) and (RdWr)
    pub fn parse_wr_rd(&mut self) -> Result<(Wr, Rd), CrsnError> {
        if self.len() == 1 {
            let rw = self.next_rdwr()?;
            Ok((rw.wr(), rw.rd()))
        } else if self.len() == 2 {
            let wr = self.next_wr()?;
            let rd = self.next_rd()?;
            Ok((wr, rd))
        } else {
            Err(CrsnError::Parse("Instruction needs 1 (RW) or 2 (Wr,Rd) arguments!".into(), self.start_pos.clone()))
        }
    }

    /// Parse a swap-type binary instruction operands (i.e. exchange) without masks
    /// Accepts (RdWr, RdWr)
    pub fn parse_rdwr_rdwr(&mut self, _keyword: &str, _prefix: &str, _pos: &SourcePosition) -> Result<(RdWr, RdWr), CrsnError> {
        if self.len() == 2 {
            let rw1 = self.next_rdwr()?;
            let rw2 = self.next_rdwr()?;
            Ok((rw1, rw2))
        } else {
            Err(CrsnError::Parse("Instruction needs 2 (RW,RW) arguments!".into(), self.start_pos.clone()))
        }
    }
}