jsrt/compiler/src/emit.rs

//! C code generation.
//!
//! One struct per object shape, and one function per inferred
//! function-and-argument pair. The body is walked the same way inference walks
//! it — block by block, instruction by instruction — with one difference:
//! inference skips an instruction that defines nothing, but jumps and returns
//! define nothing and still have to be emitted.
//!
//! A function value is compile-time only: every call site knows the callee, so
//! calls go straight to the monomorphized function by name. The function is
//! never stored, so a closure's struct is its captured environment alone.

use std::collections::HashMap;
use std::fmt::Write;

use crate::infer::types::Type;
use crate::infer::{InferRequest, InferResult, TypeCache};
use crate::libqjs::parse::{Atom, AtomRef, ImmOperand, JsValue, atom_to_str, opcodes::OpCode};
use crate::model::{EXTRA_ARGS, FIELD_PROTO};
use crate::regify::{BasicBlock, RegFn, RegInstr};

/// The C name of each monomorphized variant, keyed by its function and argument
/// types, so a call site and a definition agree.
type Names = HashMap<(usize, Vec<Type>), String>;

pub fn emit(cache: &TypeCache, reg_fns: &[RegFn], atoms: &[Atom]) -> String {
    let mut out = String::new();

    out.push_str("// Generated by the jsrt compiler. Do not edit.\n\n");
    out.push_str("#include <stdbool.h>\n#include <stdint.h>\n\n");
    out.push_str("// Placeholders, until the lattice can say more than this.\n");
    out.push_str("typedef struct { char _; } undefined_t;\n");
    out.push_str("typedef struct { char _; } any_t;\n");
    out.push_str("typedef struct { char _; } empty_t;\n\n");

    let names = mangle(cache, reg_fns, atoms);

    emit_shapes(cache, atoms, &mut out);

    out.push_str("// ── prototypes ──\n");
    for (req, res) in sorted(cache) {
        let _ = writeln!(out, "{};", signature(req, res, cache, &names));
    }
    out.push('\n');

    for (req, res) in sorted(cache) {
        emit_fn(req, res, reg_fns, cache, atoms, &names, &mut out);
    }

    out
}

/// Results in a stable order, so the same program emits the same file.
fn sorted(cache: &TypeCache) -> Vec<(&InferRequest, &InferResult)> {
    let mut rs: Vec<_> = cache.results().collect();
    rs.sort_by_key(|(req, _)| (req.fn_id, format!("{:?}", req.arg_types)));
    rs
}

/// One C name per function-and-argument pair: the name, its id, and a per-
/// function variant index.
fn mangle(cache: &TypeCache, reg_fns: &[RegFn], atoms: &[Atom]) -> Names {
    let mut names = HashMap::new();
    let mut seen: HashMap<usize, usize> = HashMap::new();

    for (req, _) in sorted(cache) {
        let variant = seen.entry(req.fn_id).or_insert(0);
        let base = sanitize(&atom_to_str(&reg_fns[req.fn_id].name, atoms));
        let key = (req.fn_id, req.arg_types.clone());
        names.insert(key, format!("{}_{}_{}", base, req.fn_id, variant));
        *variant += 1;
    }

    names
}

fn name_of(fn_id: usize, arg_types: &[Type], names: &Names) -> String {
    names[&(fn_id, arg_types.to_vec())].clone()
}

fn signature(req: &InferRequest, res: &InferResult, cache: &TypeCache, names: &Names) -> String {
    let params: Vec<String> = req
        .arg_types
        .iter()
        .enumerate()
        .map(|(i, ty)| format!("{} a{i}", c_type(ty, cache)))
        .collect();

    format!(
        "{} {}({})",
        c_type(&res.ret_type, cache),
        name_of(req.fn_id, &req.arg_types, names),
        params.join(", ")
    )
}

fn emit_fn(
    req: &InferRequest,
    res: &InferResult,
    reg_fns: &[RegFn],
    cache: &TypeCache,
    atoms: &[Atom],
    names: &Names,
    out: &mut String,
) {
    let f = &reg_fns[req.fn_id];

    let _ = writeln!(out, "// {}", atom_to_str(&f.name, atoms));
    let _ = writeln!(out, "{} {{", signature(req, res, cache, names));

    // Every value gets a slot. Locals are not renamed, so a local and the
    // values flowing through it share one.
    for (v, ty) in res.val_types.iter().enumerate() {
        let _ = writeln!(out, "    {} v{v};", c_type(ty, cache));
    }
    if !res.val_types.is_empty() {
        out.push('\n');
    }

    for (bi, block) in f.blocks.iter().enumerate() {
        let _ = writeln!(out, "L{bi}:;");

        // Stack values that outlive the block are copied into the successor's
        // slots, and that has to happen before control leaves the block — so
        // the terminator is held back until the copies are out.
        let body = &f.code[block.start..block.end];
        let (body, terminator) = match body.last() {
            Some(last) if is_terminator(last.op) => (&body[..body.len() - 1], Some(last)),
            _ => (body, None),
        };

        for instr in body {
            emit_instr(instr, block, res, f, reg_fns, cache, atoms, names, out);
        }

        for instr in &block.postfix {
            emit_instr(instr, block, res, f, reg_fns, cache, atoms, names, out);
        }

        if let Some(instr) = terminator {
            emit_instr(instr, block, res, f, reg_fns, cache, atoms, names, out);
        }
    }

    out.push_str("}\n\n");
}

#[allow(clippy::too_many_arguments)]
fn emit_instr(
    instr: &RegInstr,
    block: &BasicBlock,
    res: &InferResult,
    f: &RegFn,
    reg_fns: &[RegFn],
    cache: &TypeCache,
    atoms: &[Atom],
    names: &Names,
    out: &mut String,
) {
    use OpCode::*;

    let a = |i: usize| format!("v{}", instr.args[i]);

    // A binary operand sits under its partner on the stack, so the left-hand
    // side of the source expression is the *later* arg.
    let bin = |op: &str| format!("{} {} {}", a(1), op, a(0));

    let set = |expr: String, out: &mut String| {
        for d in &instr.dsts {
            let _ = writeln!(out, "    v{d} = {expr};");
        }
    };

    match instr.op {
        Nop | Drop => {}

        PushMinus1 => set("-1".into(), out),
        Push0 => set("0".into(), out),
        Push1 => set("1".into(), out),
        Push2 => set("2".into(), out),
        Push3 => set("3".into(), out),
        Push4 => set("4".into(), out),
        Push5 => set("5".into(), out),
        Push6 => set("6".into(), out),
        Push7 => set("7".into(), out),
        PushI8 | PushI16 | PushI32 => {
            let v = match instr.imm {
                ImmOperand::I8(n) => n as i64,
                ImmOperand::I16(n) => n as i64,
                ImmOperand::I32(n) => n as i64,
                _ => unreachable!(),
            };
            set(v.to_string(), out);
        }
        PushTrue => set("true".into(), out),
        PushFalse => set("false".into(), out),

        PushConst | PushConst8 => {
            let ImmOperand::Const(k) = instr.imm else {
                unreachable!()
            };
            let expr = match &f.cpool[k as usize] {
                JsValue::Int(n) => n.to_string(),
                JsValue::Float(x) => format!("{x:?}"),
                JsValue::BoolTrue => "true".into(),
                JsValue::BoolFalse => "false".into(),
                _ => unreachable!(),
            };
            set(expr, out);
        }

        Add => set(bin("+"), out),
        Sub => set(bin("-"), out),
        Mul => set(bin("*"), out),
        Mod => set(bin("%"), out),
        Div => set(format!("(double){} / (double){}", a(1), a(0)), out),
        Neg => set(format!("-{}", a(0)), out),
        Plus => set(a(0), out),
        Inc => set(format!("{} + 1", a(0)), out),
        Dec => set(format!("{} - 1", a(0)), out),

        Lt => set(bin("<"), out),
        Lte => set(bin("<="), out),
        Gt => set(bin(">"), out),
        Gte => set(bin(">="), out),
        Eq | StrictEq => set(bin("=="), out),
        Neq | StrictNeq => set(bin("!="), out),
        Not => set(format!("!{}", a(0)), out),

        Dup | SetLoc | SetName => {
            let src = instr.args[0];
            for d in &instr.dsts {
                emit_copy(*d, &res.val_types[*d], src, &res.val_types[src], atoms, out);
            }
        }
        Undefined => set("(undefined_t){0}".into(), out),

        GetArg0 => set("a0".into(), out),
        GetArg1 => set("a1".into(), out),
        GetArg2 => set("a2".into(), out),
        GetArg3 => set("a3".into(), out),
        GetArg => {
            let ImmOperand::Arg(k) = instr.imm else {
                unreachable!()
            };
            set(format!("a{k}"), out);
        }

        // A new object, or a new closure: both start with no fields. A
        // closure's function is compile-time only, so it leaves no trace here.
        Object | Fclosure | Fclosure8 => {
            let d = instr.dsts[0];
            let _ = writeln!(
                out,
                "    v{d} = ({}){{0}};",
                c_type(&res.val_types[d], cache)
            );
        }

        DefineField => {
            let ImmOperand::Atom(atom) = instr.imm else {
                unreachable!()
            };
            let obj = instr.args[1];
            let d = instr.dsts[0];

            // A write-back names the object itself, so there is nothing to
            // carry over. Otherwise the result is a different shape, and C will
            // not assign one struct type to another — so the fields that
            // survive are copied one at a time.
            if d != obj {
                for (carried, _) in res.val_types[obj].fields() {
                    if carried == atom {
                        continue;
                    }
                    let cn = field_name(carried, atoms);
                    let _ = writeln!(out, "    v{d}.{cn} = v{obj}.{cn};");
                }
            }

            let _ = writeln!(out, "    v{d}.{} = {};", field_name(atom, atoms), a(0));
        }

        GetField => {
            let ImmOperand::Atom(atom) = instr.imm else {
                unreachable!()
            };
            set(format!("{}.{}", a(0), field_name(atom, atoms)), out);
        }

        Call | Call0 | Call1 | Call2 | Call3 | TailCall | CallMethod | TailCallMethod => {
            emit_call(instr, res, reg_fns, names, out);
        }

        Return | ReturnAsync => {
            let _ = writeln!(out, "    return {};", a(0));
        }
        ReturnUndef => {
            let _ = writeln!(out, "    return ({}){{0}};", c_type(&res.ret_type, cache));
        }

        Goto | Goto8 | Goto16 => {
            let _ = writeln!(out, "    goto L{};", block.succ[0]);
        }
        IfFalse | IfFalse8 => {
            let _ = writeln!(out, "    if (!{}) goto L{};", a(0), block.succ[0]);
            if let Some(fallthrough) = block.succ.get(1) {
                let _ = writeln!(out, "    goto L{fallthrough};");
            }
        }
        IfTrue | IfTrue8 => {
            let _ = writeln!(out, "    if ({}) goto L{};", a(0), block.succ[0]);
            if let Some(fallthrough) = block.succ.get(1) {
                let _ = writeln!(out, "    goto L{fallthrough};");
            }
        }

        other => {
            let _ = writeln!(out, "    /* TODO {other:?} */");
        }
    }
}

/// Move one value into another's slot.
///
/// A local's slot is declared with the type it ends up holding, so an earlier
/// assignment to it can be a different shape. C will not assign one struct type
/// to another, so the fields the two shapes share are carried over one at a
/// time. Fields the destination does not have yet are simply not set.
fn emit_copy(
    dst: usize,
    dst_ty: &Type,
    src: usize,
    src_ty: &Type,
    atoms: &[Atom],
    out: &mut String,
) {
    if dst_ty == src_ty || !is_object(dst_ty) || !is_object(src_ty) {
        let _ = writeln!(out, "    v{dst} = v{src};");
        return;
    }

    for (name, _) in src_ty.fields() {
        if !dst_ty.fields().any(|(n, _)| n == name) {
            continue;
        }
        let n = field_name(name, atoms);
        let _ = writeln!(out, "    v{dst}.{n} = v{src}.{n};");
    }
}

/// Control leaves the block at these, so nothing may be emitted after one.
fn is_terminator(op: OpCode) -> bool {
    use OpCode::*;
    matches!(
        op,
        Goto | Goto8
            | Goto16
            | IfFalse
            | IfFalse8
            | IfTrue
            | IfTrue8
            | Return
            | ReturnUndef
            | ReturnAsync
    )
}

/// Something with a struct representation whose fields can be copied. A closure
/// counts: its representation is its environment.
fn is_object(ty: &Type) -> bool {
    matches!(
        ty,
        Type::Empty | Type::WithField { .. } | Type::Closure { .. }
    )
}

/// The frame is the one inference reads: `[this,] callee, arg0 .. argN`, with
/// `args` in pop order. The request is rebuilt exactly as inference built it, so
/// the variant is an exact lookup, and the callee is emitted by name.
fn emit_call(
    instr: &RegInstr,
    res: &InferResult,
    reg_fns: &[RegFn],
    names: &Names,
    out: &mut String,
) {
    use OpCode::*;

    let is_method = matches!(instr.op, CallMethod | TailCallMethod);
    let n = instr.args.len();
    let (callee_idx, this, arg_end) = if is_method {
        (instr.args[n - 2], Some(instr.args[n - 1]), n - 2)
    } else {
        (instr.args[n - 1], None, n - 1)
    };

    let Type::Closure { fn_id, .. } = res.val_types[callee_idx] else {
        unreachable!()
    };

    let declared = reg_fns[fn_id].arg_count - EXTRA_ARGS;
    let mut arg_types = Vec::with_capacity(declared + EXTRA_ARGS);
    let mut args = Vec::with_capacity(declared + EXTRA_ARGS);

    let mut push = |ty: Type, expr: String| {
        arg_types.push(ty);
        args.push(expr);
    };

    for i in 0..declared {
        match arg_end.checked_sub(i + 1) {
            Some(k) if i < arg_end => {
                let v = instr.args[k];
                push(res.val_types[v].clone(), format!("v{v}"));
            }
            _ => push(Type::Undefined, "(undefined_t){0}".to_string()),
        }
    }

    match this {
        Some(v) => push(res.val_types[v].clone(), format!("v{v}")),
        None => push(Type::Undefined, "(undefined_t){0}".to_string()),
    }

    push(res.val_types[callee_idx].clone(), format!("v{callee_idx}"));

    let call = format!("{}({})", name_of(fn_id, &arg_types, names), args.join(", "));

    if instr.dsts.is_empty() {
        let _ = writeln!(out, "    {call};");
    } else {
        for d in &instr.dsts {
            let _ = writeln!(out, "    v{d} = {call};");
        }
    }
}

fn emit_shapes(cache: &TypeCache, atoms: &[Atom], out: &mut String) {
    let mut shapes: Vec<_> = cache.shapes().collect();
    shapes.sort_by_key(|(_, id)| id.0);

    for (ty, id) in shapes {
        let _ = writeln!(out, "typedef struct {{");

        for (name, value) in ty.fields() {
            let _ = writeln!(
                out,
                "    {} {};",
                c_type(value, cache),
                field_name(name, atoms)
            );
        }

        let _ = writeln!(out, "}} Shape{};\n", id.0);
    }
}

fn c_type(ty: &Type, cache: &TypeCache) -> String {
    match ty {
        Type::Int => "int32_t".to_string(),
        Type::Float => "double".to_string(),
        Type::Bool => "bool".to_string(),
        Type::Undefined => "undefined_t".to_string(),
        // Unknown means the value was never inferred: dead code, or a known
        // gap. It has no real type, so it renders as undefined.
        Type::Unknown => "undefined_t".to_string(),
        Type::Any => "any_t".to_string(),
        // A closure is its environment; the function is compile-time only.
        Type::Closure { env, .. } => c_type(env, cache),
        // An object with no fields carries nothing, so it needs no struct.
        Type::Empty => "empty_t".to_string(),
        // A native is called by name, so its value carries nothing.
        Type::Native(_) => "empty_t".to_string(),
        Type::WithField { .. } => format!("Shape{}", cache.shape_id(ty).unwrap().0),
    }
}

fn field_name(name: AtomRef, atoms: &[Atom]) -> String {
    if name == FIELD_PROTO {
        return "proto".to_string();
    }

    match name {
        AtomRef::TaggedInt(n) => format!("i{n}"),
        _ => sanitize(&atom_to_str(&name, atoms)),
    }
}

fn sanitize(s: &str) -> String {
    let mut out: String = s
        .chars()
        .map(|c| if c.is_ascii_alphanumeric() { c } else { '_' })
        .collect();

    if !out.starts_with(|c: char| c.is_ascii_alphabetic() || c == '_') {
        out.insert(0, '_');
    }

    out
}