crsn/crsn/src/asm/parse/parse_data.rs

use std::borrow::Cow;
use std::convert::TryFrom;

use sexp::{Atom, Sexp, SourcePosition};

use crate::asm::data::{DataDisp, Rd, RdData, reg, Wr, WrData, RdWr};
use crate::asm::data::literal::{ConstantName, Label, RegisterAlias, Value};
use crate::asm::error::CrsnError;
use crate::asm::parse::{ParserContext, parse_instructions};
use crate::asm::parse::sexp_expect::expect_string_atom;
use crate::asm::patches::ErrWithPos;
use std::sync::atomic::AtomicU32;
use crate::module::OpTrait;
use crate::asm::instr::Cond;

fn is_valid_identifier(name: &str) -> bool {
    // ascii symbols "!\"#$_&'()*+,-./:;<=>?@[\\]^_`{|}~"
    const BLACKLIST : &str = "!\"#$&'()*+,./;<=>?@[\\]^`{|}~";
    name != "_"
        && !name.starts_with(|c: char| c.is_ascii_digit() || BLACKLIST.contains(c))
        && !name.contains(|c : char| c.is_whitespace() || BLACKLIST.contains(c))
}

/// Parse register alias
pub fn parse_reg_alias(name: Sexp) -> Result<(RegisterAlias, SourcePosition), CrsnError> {
    // trace!("parse reg alias: {:?}", name);

    let (name, namepos) = expect_string_atom(name)?;

    if !is_valid_identifier(&name) {
        return Err(CrsnError::Parse(format!("\"{}\" is not an allowed register alias.", name).into(), namepos));
    }

    Ok((name, namepos))
}

/// Parse constant name
pub fn parse_constant_name(name: Sexp) -> Result<(ConstantName, SourcePosition), CrsnError> {
    // trace!("parse const name: {:?}", name);

    let (name, namepos) = expect_string_atom(name)?;

    if !is_valid_identifier(&name) {
        return Err(CrsnError::Parse(format!("\"{}\" is not an allowed constant name.", name).into(), namepos));
    }

    Ok((name, namepos))
}

/// Parse a label
pub fn parse_label(name: Sexp) -> Result<Label, CrsnError> {
    let (name, namepos) = expect_string_atom(name)?;
    Ok(parse_label_str(&name, &namepos)?)
}

pub fn parse_label_str(name: &str, pos: &SourcePosition) -> Result<Label, CrsnError> {
    if !name.starts_with(':') {
        return Err(CrsnError::Parse("Label must start with a colon".into(), pos.clone()));
    }
    let label = name.trim_start_matches(':');

    Ok(if label.starts_with('#') {
        let val = parse_u64(&label[1..], pos)?;
        Label::Numbered(u32::try_from(val).err_pos(pos)?)
    } else {
        if !is_valid_identifier(&label) {
            return Err(CrsnError::Parse(format!("\"{}\" is not an allowed label name.", name).into(), pos.clone()));
        }

        Label::Named(label.to_string())
    })
}

/// Parse data disposition (address/value, without the read/write restriction)
pub fn parse_data_disp(tok: Sexp, pcx: &ParserContext) -> Result<DataDisp, CrsnError> {
    // trace!("parse data: {:?}", tok);
    match tok {
        Sexp::Atom(Atom::I(val), _pos) => {
            Ok(DataDisp::Immediate(unsafe { std::mem::transmute(val) }))
        }
        Sexp::Atom(Atom::F(val), _pos) => {
            Ok(DataDisp::Immediate(unsafe { std::mem::transmute(val) }))
        }
        Sexp::Atom(Atom::U(val), _pos) => {
            Ok(DataDisp::Immediate(val))
        }
        Sexp::Atom(Atom::C(val), _pos) => {
            Ok(DataDisp::Immediate(val as u64))
        }
        Sexp::Atom(Atom::QS(_s), pos) => {
            Err(CrsnError::Parse("Quoted string not expected here".into(), pos))
        }
        Sexp::List(list, pos) => {
            if let Some(Sexp::Atom(Atom::S(s), s_pos)) = list.first() {
                let parsing_expr = pcx.state.borrow().parsing_expr;

                if !s.starts_with('=') && !parsing_expr {
                    return Err(CrsnError::Parse(format!("Invalid syntax, did you mean \"={}\"?", s).into(), s_pos.clone()));
                }

                // start expr parsing mode
                let orig_parsing_expr = pcx.state.borrow().parsing_expr;
                pcx.state.borrow_mut().parsing_expr = true;

                let lmut = AtomicU32::new(0);

                let mut expr_ops = parse_instructions(
                    // TODO avoid this madness
                    vec![Sexp::List(list, pos.clone())].into_iter(),
                    &pos, pcx)?
                    .flatten(&lmut)?;

                if expr_ops.len() != 1 {
                    return Err(CrsnError::Parse(format!("Expected one top level operation in an expression, got: {:?}", expr_ops).into(), pos.clone()));
                }

                let op = expr_ops.remove(0);

                let mut state_mut = pcx.state.borrow_mut();
                state_mut.const_eval.clear_all();

                let ticl = state_mut.const_eval.thread_info.clone();
                op.execute(&ticl, &mut state_mut.const_eval)
                    .map_err(|f| CrsnError::ParseOther(Box::new(f), pos.clone()))?;

                if state_mut.const_eval.test_cond(Cond::Invalid) {
                    return Err(CrsnError::Parse("Expression evaluation failed - invalid flag set!".into(), pos.clone()));
                }

                if !orig_parsing_expr {
                    // exit expr parsing mode
                    state_mut.parsing_expr = false;
                }

                return Ok(DataDisp::Immediate(state_mut.const_eval.frame.gen[0]));
            }

            Err(CrsnError::Parse("List not expected here".into(), pos))
        }
        Sexp::Atom(Atom::S(s), pos) => {
            parse_data_disp_from_str(&s, &pos, pcx)
        }
    }
}

/// Parse data disp from a string token (a &str is used so the string can be preprocessed)
pub fn parse_data_disp_from_str(s: &str, pos: &SourcePosition, pcx: &ParserContext) -> Result<DataDisp, CrsnError> {
    if s == "_" {
        return Ok(DataDisp::Discard);
    }

    if let Some(reference) = s.strip_prefix('@') {
        /* extension constants (pre-defined handles) */
        for p in pcx.parsers {
            if let Some(val) = p.get_constant_value(reference) {
                return Ok(DataDisp::ImmObject(val));
            }
        }

        /* register aliases */
        let pstate = pcx.state.borrow();
        if let Some(reg) = pstate.reg_aliases.get(reference) {
            return Ok(DataDisp::RegObject(*reg))
        } else if let Some(reg) = pstate.global_reg_aliases.get(reference) {
            return Ok(DataDisp::RegObject(*reg))
        }

        let reg = reg::parse_reg(reference, pos)?;

        if pstate.reg_aliases.values().find(|v| **v == reg).is_some() {
            Err(CrsnError::Parse(format!("Sym exists for register {}, direct access denied. Unsym it if needed.", reg).into(), pos.clone()))
        } else if pstate.global_reg_aliases.values().find(|v| **v == reg).is_some() {
            Err(CrsnError::Parse(format!("Global sym exists for register {}, direct access denied. Unsym it if needed.", reg).into(), pos.clone()))
        } else {
            Ok(DataDisp::RegObject(reg))
        }
    } else {
        /* extension constants */
        for p in pcx.parsers {
            if let Some(val) = p.get_constant_value(&s) {
                return Ok(DataDisp::Immediate(val));
            }
        }

        /* program constants */
        let pstate = pcx.state.borrow();
        if let Some(val) = pstate.constants.get(s) {
            return Ok(DataDisp::Immediate(*val));
        }

        /* register aliases */
        if let Some(val) = pstate.reg_aliases.get(s) {
            return Ok(DataDisp::Register(*val));
        } else if let Some(val) = pstate.global_reg_aliases.get(s) {
            return Ok(DataDisp::Register(*val));
        }

        /* It can also be a plain old number */

        if s.starts_with('#')
            || s.starts_with('-')
            || s.starts_with('.')
            || s.starts_with("0x")
            || s.starts_with("0b")
            || s.starts_with(|c : char| c.is_ascii_digit())
        {
            match sexp::atom_of_string(s.to_string()) {
                Atom::C(v) => return Ok(DataDisp::Immediate(v as Value)),
                Atom::I(v) => return Ok(DataDisp::Immediate(unsafe { std::mem::transmute::<_, u64>(v) } as Value)),
                Atom::U(v) => return Ok(DataDisp::Immediate(v as Value)),
                Atom::F(v) => return Ok(DataDisp::Immediate(unsafe { std::mem::transmute::<_, u64>(v) } as Value)),
                Atom::S(_) | Atom::QS(_) => {
                    // this will probably fail validation down the line, but it is definitely not a number
                }
            }
        }

        /* register */
        let reg = reg::parse_reg(&s, &pos)?;

        let pstate = pcx.state.borrow();
        if pstate.reg_aliases.values().find(|v| **v == reg).is_some() {
            Err(CrsnError::Parse(format!("Sym exists for register {}, direct access denied. Unsym it if needed.", reg).into(), pos.clone()))
        } else if pstate.global_reg_aliases.values().find(|v| **v == reg).is_some() {
            Err(CrsnError::Parse(format!("Global sym exists for register {}, direct access denied. Unsym it if needed.", reg).into(), pos.clone()))
        } else {
            Ok(DataDisp::Register(reg))
        }
    }
}

/// Parse immediate value
pub fn parse_value(tok: Sexp, pcx: &ParserContext) -> Result<(Value, SourcePosition), CrsnError> {
    match tok {
        Sexp::Atom(Atom::I(val), pos) => {
            Ok((unsafe { std::mem::transmute(val) }, pos))
        }
        Sexp::Atom(Atom::U(val), pos) => {
            Ok((val, pos))
        }
        Sexp::Atom(Atom::C(val), pos) => {
            Ok((val as u64, pos))
        }
        Sexp::Atom(Atom::QS(_), pos) => {
            Err(CrsnError::Parse("quoted string not expected here".into(), pos))
        }
        Sexp::Atom(Atom::F(val), pos) => {
            Ok((unsafe { std::mem::transmute(val) }, pos))
        }
        Sexp::Atom(Atom::S(s), pos) => {
            let pstate = pcx.state.borrow();
            if let Some(val) = pstate.constants.get(&s) {
                return Ok((*val, pos));
            }

            /* extension constants */
            for p in pcx.parsers {
                if let Some(val) = p.get_constant_value(&s) {
                    return Ok((val, pos));
                }
            }

            Err(CrsnError::Parse(format!("unknown constant: {}", s).into(), pos))
        }
        Sexp::List(_, pos) => {
            Err(CrsnError::Parse("expected a value".into(), pos))
        }
    }
}


pub fn parse_u64(literal: &str, pos: &SourcePosition) -> Result<u64, CrsnError> {
    // trace!("parse u64 from {}", literal);
    let mut without_underscores = Cow::Borrowed(literal);
    if without_underscores.contains('_') {
        without_underscores = without_underscores.replace('_', "").into();
    }

    if let Some(hex) = without_underscores.strip_prefix("0x") {
        Ok(u64::from_str_radix(hex, 16).err_pos(pos)?)
    } else if let Some(hex) = without_underscores.strip_prefix("0b") {
        Ok(u64::from_str_radix(hex, 2).err_pos(pos)?)
    } else {
        Ok(u64::from_str_radix(&without_underscores, 10).err_pos(pos)?)
    }
}

pub fn parse_rd(tok: Sexp, pcx: &ParserContext) -> Result<Rd, CrsnError> {
    let pos = tok.pos().clone();
    Ok(Rd::new(RdData::try_from(parse_data_disp(tok, pcx)?).err_pos(&pos)?))
}

pub fn parse_wr(tok: Sexp, pcx: &ParserContext) -> Result<Wr, CrsnError> {
    let pos = tok.pos().clone();
    Ok(Wr::new(WrData::try_from(parse_data_disp(tok, pcx)?).err_pos(&pos)?))
}

pub fn parse_rdwr(tok: Sexp, pcx: &ParserContext) -> Result<RdWr, CrsnError> {
    let pos = tok.pos().clone();
    let wr = parse_wr(tok, pcx)?;
    if !wr.is_readable() {
        return Err(CrsnError::Parse("Argument is not readable!".into(), pos));
    }
    Ok(RdWr::new(wr.0))
}

pub fn parse_wr_offset(tok: Sexp, pcx: &ParserContext) -> Result<(Wr, u32), CrsnError> {
    match tok {
        Sexp::Atom(Atom::S(s), pos) => {
            let (s, offset) = parse_disp_offs(&s, &pos)?;
            Ok((Wr::new(WrData::try_from(parse_data_disp_from_str(s, &pos, pcx)?).err_pos(&pos)?), offset))
        }
        _ => Ok((parse_wr(tok, pcx)?, 0))
    }
}

pub fn parse_rd_offset(tok: Sexp, pcx: &ParserContext) -> Result<(Rd, u32), CrsnError> {
    match tok {
        Sexp::Atom(Atom::S(s), pos) => {
            let (s, offset) = parse_disp_offs(&s, &pos)?;
            Ok((Rd::new(RdData::try_from(parse_data_disp_from_str(s, &pos, pcx)?).err_pos(&pos)?), offset))
        }
        _ => Ok((parse_rd(tok, pcx)?, 0))
    }
}

pub fn parse_rdwr_offset(tok: Sexp, pcx: &ParserContext) -> Result<(RdWr, u32), CrsnError> {
    match tok {
        Sexp::Atom(Atom::S(s), pos) => {
            let (s, offset) = parse_disp_offs(&s, &pos)?;
            let wr = WrData::try_from(parse_data_disp_from_str(s, &pos, pcx)?).err_pos(&pos)?;

            if !wr.is_readable() {
                return Err(CrsnError::Parse("Argument is not readable!".into(), pos));
            }
            Ok((RdWr::new(wr), offset))
        }
        _ => Ok((parse_rdwr(tok, pcx)?, 0))
    }
}

fn parse_disp_offs<'a>(s : &'a str, pos: &SourcePosition) -> Result<(&'a str, u32), CrsnError> {
    if s.contains(':') {
        let mut split = s.split(':');

        let disp = split.next().unwrap();
        let num : u32 = split.next().unwrap().parse().err_pos(pos)?;
        Ok((disp, num))
    } else {
        Ok((&s, 0))
    }
}