compiler/

op_code.rs

use std::collections::HashMap;

use byteorder;
use byteorder::{BigEndian, ByteOrder, WriteBytesExt};

use strum::{EnumCount, EnumIter, FromRepr};

// why not type, see https://stackoverflow.com/a/35569079/1713757
#[derive(Hash, Eq, Debug, Clone, PartialEq, PartialOrd)]
pub struct Instructions {
    pub data: Vec<u8>,
}

pub struct OpcodeDefinition {
    pub(crate) name: &'static str,
    operand_width: Vec<i32>,
}

impl OpcodeDefinition {
    pub fn name(&self) -> &'static str {
        self.name
    }

    pub fn operand_widths(&self) -> &[i32] {
        &self.operand_width
    }
}

#[repr(u8)]
#[derive(Debug, Hash, Eq, Clone, Copy, PartialEq, EnumCount, EnumIter, FromRepr)]
pub enum Opcode {
    OpConst,
    OpAdd,
    OpPop,
    OpSub,
    OpMul,
    OpDiv,
    OpTrue,
    OpFalse,
    OpEqual,
    OpNotEqual,
    OpGreaterThan,
    OpMinus,
    OpBang,
    OpJumpNotTruthy,
    OpJump,
    OpNull,
    OpGetGlobal,
    OpSetGlobal,
    OpArray,
    OpHash,
    OpIndex,
    OpCall,
    OpReturnValue,
    OpReturn,
    OpGetLocal,
    OpSetLocal,
    OpGetBuiltin,
    OpClosure,
    OpGetFree,
    OpCurrentClosure,
    OpClass,
    OpMethod,
    OpGetProperty,
    OpSetProperty,
    OpNew,
    // Appended to preserve the byte values of all existing opcodes.
    OpLessThan,
}

lazy_static! {
    pub static ref DEFINITIONS: HashMap<Opcode, OpcodeDefinition> = {
        let mut m = HashMap::new();
        m.insert(
            Opcode::OpConst,
            OpcodeDefinition {
                name: "OpConst",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpAdd,
            OpcodeDefinition {
                name: "OpAdd",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpPop,
            OpcodeDefinition {
                name: "OpPop",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpSub,
            OpcodeDefinition {
                name: "OpSub",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpMul,
            OpcodeDefinition {
                name: "OpMul",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpDiv,
            OpcodeDefinition {
                name: "OpDiv",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpTrue,
            OpcodeDefinition {
                name: "OpTrue",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpFalse,
            OpcodeDefinition {
                name: "OpFalse",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpEqual,
            OpcodeDefinition {
                name: "OpEqual",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpNotEqual,
            OpcodeDefinition {
                name: "OpNotEqual",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpGreaterThan,
            OpcodeDefinition {
                name: "OpGreatThan",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpLessThan,
            OpcodeDefinition {
                name: "OpLessThan",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpMinus,
            OpcodeDefinition {
                name: "OpMinus",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpBang,
            OpcodeDefinition {
                name: "OpBang",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpJumpNotTruthy,
            OpcodeDefinition {
                name: "OpJumpNotTruthy",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpJump,
            OpcodeDefinition {
                name: "OpJump",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpNull,
            OpcodeDefinition {
                name: "OpNull",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpGetGlobal,
            OpcodeDefinition {
                name: "OpGetGlobal",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpSetGlobal,
            OpcodeDefinition {
                name: "OpSetGlobal",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpArray,
            OpcodeDefinition {
                name: "OpArray",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpHash,
            OpcodeDefinition {
                name: "OpHash",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpIndex,
            OpcodeDefinition {
                name: "OpIndex",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpCall,
            OpcodeDefinition {
                name: "OpCall",
                operand_width: vec![1],
            },
        );
        m.insert(
            Opcode::OpReturn,
            OpcodeDefinition {
                name: "OpReturn",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpReturnValue,
            OpcodeDefinition {
                name: "OpReturnValue",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpGetLocal,
            OpcodeDefinition {
                name: "OpGetLocal",
                operand_width: vec![1],
            },
        );
        m.insert(
            Opcode::OpSetLocal,
            OpcodeDefinition {
                name: "OpSetLocal",
                operand_width: vec![1],
            },
        );
        m.insert(
            Opcode::OpGetBuiltin,
            OpcodeDefinition {
                name: "OpGetBuiltin",
                operand_width: vec![1],
            },
        );
        m.insert(
            Opcode::OpClosure,
            OpcodeDefinition {
                name: "OpClosure",
                operand_width: vec![2, 1],
            },
        );
        m.insert(
            Opcode::OpGetFree,
            OpcodeDefinition {
                name: "OpGetFree",
                operand_width: vec![1],
            },
        );
        m.insert(
            Opcode::OpCurrentClosure,
            OpcodeDefinition {
                name: "OpCurrentClosure",
                operand_width: vec![],
            },
        );
        m.insert(
            Opcode::OpClass,
            OpcodeDefinition {
                name: "OpClass",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpMethod,
            OpcodeDefinition {
                name: "OpMethod",
                operand_width: vec![2, 1],
            },
        );
        m.insert(
            Opcode::OpGetProperty,
            OpcodeDefinition {
                name: "OpGetProperty",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpSetProperty,
            OpcodeDefinition {
                name: "OpSetProperty",
                operand_width: vec![2],
            },
        );
        m.insert(
            Opcode::OpNew,
            OpcodeDefinition {
                name: "OpNew",
                operand_width: vec![1],
            },
        );
        return m;
    };
}

/// # Panics
///
/// If an operand does not fit its opcode's fixed width. `Compiler` rejects
/// those with a `CompileError` before reaching here, so a panic means the
/// caller built the instruction by hand. The alternative — truncating with
/// `as u8` / `as u16` — is a silent miscompile, so this is checked in release
/// too rather than left to `debug_assert!`.
pub fn make_instructions(op: Opcode, operands: &[usize]) -> Instructions {
    let mut instructions = Vec::new();
    instructions.push(op as u8);
    let widths = &DEFINITIONS.get(&op).unwrap().operand_width;

    for (o, w) in operands.iter().zip(widths) {
        match w {
            2 => {
                assert!(
                    *o <= u16::MAX as usize,
                    "{:?} operand {} does not fit in 2 bytes; the compiler must \
                     reject it before emitting rather than truncate it here",
                    op,
                    o
                );
                instructions.write_u16::<BigEndian>(*o as u16).unwrap();
            }
            1 => {
                assert!(
                    *o <= u8::MAX as usize,
                    "{:?} operand {} does not fit in 1 byte; the compiler must \
                     reject it before emitting rather than truncate it here",
                    op,
                    o
                );
                instructions.write_u8(*o as u8).unwrap();
            }
            _ => {
                panic!("unsupported operand width {}", w)
            }
        }
    }

    return Instructions {
        data: instructions,
    };
}

pub fn read_operands(def: &OpcodeDefinition, ins: &[u8]) -> (Vec<usize>, usize) {
    let mut operands = Vec::with_capacity(def.operand_width.len());
    let mut offset = 0;

    for w in &def.operand_width {
        match w {
            2 => {
                operands.push(BigEndian::read_u16(&ins[offset..offset + 2]) as usize);
                offset += 2;
            }
            1 => {
                operands.push(ins[offset] as usize);
                offset += 1;
            }
            0 => {}
            _ => {
                panic!("unsupported operand width {} for read", w)
            }
        }
    }

    return (operands, offset);
}

pub fn concat_instructions(expected: &Vec<Instructions>) -> Instructions {
    let mut out = Instructions {
        data: vec![],
    };

    for instruction in expected {
        out = out.merge_instructions(instruction)
    }

    return out;
}

impl Instructions {
    // prettify bytecodes
    pub fn string(&self) -> String {
        let mut ret = String::new();
        let mut i = 0;
        while i < self.data.len() {
            let op: u8 = self.data[i];
            let Some(opcode) = Opcode::from_repr(op) else {
                ret.push_str(&format!("{:04} <unknown opcode 0x{:02x}>\n", i, op));
                i += 1;
                continue;
            };

            let definition = DEFINITIONS.get(&opcode).unwrap();
            let width: usize = definition.operand_width.iter().map(|w| *w as usize).sum();
            if i + 1 + width > self.data.len() {
                ret.push_str(&format!("{:04} {} <truncated operands>\n", i, definition.name));
                break;
            }
            let (operands, read_size) = read_operands(definition, &self.data[i + 1..]);
            ret.push_str(&format!("{:04} {}\n", i, Self::fmt_instructions(definition, &operands)));
            i = i + 1 + read_size;
        }

        return ret;
    }

    fn fmt_instructions(def: &OpcodeDefinition, operands: &[usize]) -> String {
        match def.operand_width.len() {
            2 => format!("{} {} {}", def.name, operands[0], operands[1]),
            1 => format!("{} {}", def.name, operands[0]),
            0 => def.name.to_string(),
            _ => {
                panic!("unsupported operand width {}", def.operand_width.len());
            }
        }
    }

    pub fn merge_instructions(&self, other: &Instructions) -> Instructions {
        let ins = [self, other];
        // Maybe extend_from_slice, but I have not make it work
        // https://stackoverflow.com/a/69578632/1713757
        return Instructions {
            data: ins
                .iter()
                .fold(vec![], |sum, &i| [sum.as_slice(), i.data.as_slice()].concat()),
        };
    }
}