jsrt/compiler/src/emit/mod.rs

//! Emits C: one fn for each walk that the program calls, one struct for each
//! cell (see `docs/model.md`, section 5). Cells and walks that C cannot tell
//! apart share one struct and one fn (see `classes()`)
//!
//! A value's C representation is a function of its type (`rep()`), one for
//! each kind of type. An object is a pointer to a struct of its cell. Each
//! object starts with a header that holds the number of its cell, and the
//! fields that can be absent start as absent
//!
//! The runtime (`lib/rt/`) gives the primitive types, the string ops, the
//! `any_t` ops and the allocator

use std::{
    cell::RefCell,
    collections::{BTreeSet, HashMap, HashSet},
    fmt::Write,
};

use crate::{
    infer::{
        HOST, HOST_COMPLETE, InferResult, Place, TypeCache,
        cells::{Cell, Code, LENGTH_FIELD, ObjKind, read_value},
        types::Type,
    },
    libqjs::parse::{Atom, AtomRef, ImmOperand, JsValue, opcodes::OpCode},
    native::{self, NATIVES, ProtoOf, atom_name, js_string_text},
    regify::{BasicBlock, RegFn, RegInstr},
    transform::{FIRST_ARG, PROTO_FIELD, THIS_ARG},
};

mod calls;
mod classes;
mod conv;
mod instrs;
mod objects;
mod reps;

use classes::*;
use reps::*;

/// The number of a class of cells in C: the number in the header of their
/// objects, and the name of their struct
type CellNum = u32;

pub fn emit(fns: &[RegFn], cache: &TypeCache, atoms: &[Atom]) -> String {
    let (nums, cells, walk_rep) = classes(fns, cache);

    let mut e = Emitter {
        fns,
        cache,
        atoms,
        cells,
        nums,
        walk_rep,
        queued: vec![false; cache.walks.len()],
        queue: vec![],
        helpers: HashSet::new(),
        sums: RefCell::new(BTreeSet::new()),
        helper_protos: String::new(),
        helper_defs: String::new(),
    };

    // Every struct is declared first, so that a field can point to any of
    // them
    let mut structs = String::new();

    for id in 0..e.cells.len() as CellNum {
        let _ = writeln!(structs, "typedef struct {0} {0};", struct_name(id));
    }

    for id in 0..e.cells.len() as CellNum {
        e.struct_def(id, &mut structs);
    }

    // Which cells are fns, by number, for the runtime's `typeof`
    let fn_cells: Vec<&str> = e.cells.iter().map(|cell| if cell.code().is_some() { "1" } else { "0" }).collect();
    let mut statics = format!("const bool jsrt_fn_cells[] = {{ {} }};\n\n", fn_cells.join(", "));

    // A built-in fn is one static object
    let mut init = String::new();

    for cell in cache.builtins() {
        let id = e.num(cell);
        let name = struct_name(id);
        let _ = writeln!(statics, "static {name} {};", builtin_name(id));
        let _ = writeln!(init, "    init_{name}(&{}, {id});", builtin_name(id));
    }

    for kind in ProtoOf::ALL {
        let ty = &cache.protos[kind as usize];

        if *ty != Type::Unknown {
            let _ = writeln!(statics, "static {} {};", e.c_type(e.rep(ty)), proto_global(kind));
        }
    }

    if cache.host_complete != Type::Unknown {
        let _ = writeln!(statics, "static {} {HOST_COMPLETE_GLOBAL};", e.c_type(e.rep(&cache.host_complete)));
    }

    // The walks that the code calls, from the host's calls. The host calls the
    // entry with nothing: its env and `this` are undefined
    let entry = cache.find(fns.len() - 1, HOST, &[], fns);
    let entry_args = vec!["(undefined_t){0}"; cache.walks[entry].req.arg_types.len()].join(", ");
    e.use_walk(entry);
    let complete = e.host_complete();

    let mut fn_protos = String::new();
    let mut bodies = String::new();

    while let Some(w) = e.queue.pop() {
        let res = cache.walks[w].result.as_ref().unwrap();
        let _ = writeln!(fn_protos, "{};", e.signature(w, res));
        e.emit_fn(w, res, &mut bodies);
    }

    // The program is one file, so it also holds the runtime. The runtime goes
    // first (see `lib/rt/jsrt.h`)
    let mut out = String::from(
        "#define JSRT_IMPLEMENTATION\n#include \"jsrt.h\"\n#include <math.h>\n#include <stdlib.h>\n\n// Natives\n",
    );

    for n in NATIVES {
        let params: Vec<String> = n.args.iter().map(|t| e.c_type(e.rep(t))).collect();
        let _ = writeln!(out, "{} {}({});", e.c_type(e.rep(&n.ret)), n.name, params.join(", "));
    }

    let _ = write!(
        out,
        "\n{}\n{structs}\n{statics}\n{fn_protos}\n{}\n{}{bodies}",
        sum_defs(&e.sums.borrow()),
        e.helper_protos,
        e.helper_defs
    );
    // What the host calls (see `lib/rt/jsrt.h`)
    let _ = writeln!(out, "void jsrt_start(void) {{\n{init}    {}({entry_args});\n}}\n", e.fn_name(entry));
    let _ = writeln!(out, "void jsrt_complete(int32_t h) {{\n{complete}}}");
    out
}

struct Emitter<'a> {
    fns: &'a [RegFn],
    cache: &'a TypeCache,
    atoms: &'a [Atom],
    cells: Vec<Cell>,       // a cell of each class, by number
    nums: HashMap<Cell, CellNum>, // the class of each cell
    walk_rep: Vec<usize>,   // by walk: the walk whose fn it uses
    queued: Vec<bool>,      // the walks that the code calls, by walk
    queue: Vec<usize>,      // the walks that the code calls, which are not written yet
    helpers: HashSet<String>, // the helpers made so far, by name
    sums: RefCell<BTreeSet<Kinds>>, // the tagged unions that the C uses
    helper_protos: String,
    helper_defs: String,
}

impl<'a> Emitter<'a> {
    fn num(&self, cell: &Cell) -> CellNum {
        self.nums[cell]
    }

    /// The number of the cell of an `Obj` type
    fn num_of(&self, ty: &Type) -> CellNum {
        match ty {
            Type::Obj(cell) => self.num(cell),
            _ => unreachable!(),
        }
    }

    /// Walk `w` is called, so the walk whose fn it uses is written
    fn use_walk(&mut self, w: usize) {
        let w = self.walk_rep[w];

        if !self.queued[w] {
            self.queued[w] = true;
            self.queue.push(w);
        }
    }

    fn signature(&self, w: usize, res: &InferResult) -> String {
        let params: Vec<String> = (self.cache.walks[w].req.arg_types.iter().enumerate())
            .map(|(i, ty)| format!("{} p{i}", self.c_type(self.rep(ty))))
            .collect();

        let ret = self.c_type(self.rep(&res.ret_type));
        format!("static {ret} {}({})", self.fn_name(w), params.join(", "))
    }

    fn emit_fn(&mut self, w: usize, res: &InferResult, out: &mut String) {
        let fn_id = self.cache.walks[w].req.fn_id;
        let f = &self.fns[fn_id];
        let mut body = String::new();

        // Args are values `var_count + i` (see regify)
        for (i, ty) in self.cache.walks[w].req.arg_types.iter().enumerate() {
            let arg = f.var_count + i;
            self.assign(&mut body, &format!("v{arg}"), &format!("p{i}"), self.rep(ty), self.rep(&res.val_types[arg]));
        }

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

            if !res.reachable[b] {
                let _ = writeln!(body, "    abort();");
                continue;
            }

            let code = &f.code[block.start..block.end];

            let (instrs, last) = match code.split_last() {
                Some((last, rest)) if is_terminator(last.op) => (rest, Some(last)),
                _ => (code, None),
            };

            for (i, instr) in instrs.iter().enumerate() {
                self.emit_instr(fn_id, res, block.start + i, instr, &mut body);
            }

            // Postfix copies carry values into the next block's stack slots
            for copy in &block.postfix {
                let (src, dst) = (copy.args[0], copy.dsts[0]);
                let (from, to) = (self.rep(&res.val_types[src]), self.rep(&res.val_types[dst]));
                self.assign(&mut body, &format!("v{dst}"), &format!("v{src}"), from, to);
            }

            let goto = |i: usize| format!("goto L{};", block.succ[i]);

            match last.map(|l| l.op) {
                Some(OpCode::IfFalse | OpCode::IfFalse8 | OpCode::IfTrue | OpCode::IfTrue8) => {
                    let last = last.unwrap();
                    let cond = self.truthy(&format!("v{}", last.args[0]), &res.val_types[last.args[0]]);
                    let not = if matches!(last.op, OpCode::IfFalse | OpCode::IfFalse8) { "!" } else { "" };
                    let _ = writeln!(body, "    if ({not}{cond}) {} else {}", goto(0), goto(1));
                }
                Some(OpCode::Goto | OpCode::Goto8 | OpCode::Goto16) => {
                    let _ = writeln!(body, "    {}", goto(0));
                }
                Some(_) => self.emit_instr(fn_id, res, block.end - 1, last.unwrap(), &mut body),
                None if !block.succ.is_empty() => {
                    let _ = writeln!(body, "    {}", goto(0));
                }
                None => {}
            }
        }

        let _ = writeln!(out, "{} {{", self.signature(w, res));

        for (v, ty) in res.val_types.iter().enumerate() {
            let _ = writeln!(out, "    {} v{v};", self.c_type(self.rep(ty)));
        }

        let _ = writeln!(out, "{body}}}\n");
    }

    /// A new object of cell `id`, on the heap
    fn alloc(&self, id: CellNum) -> String {
        let name = struct_name(id);
        format!("init_{name}(jsrt_alloc(sizeof({name})), {id})")
    }

    /// The C fn of walk `w`: the fn of the walk of its class
    fn fn_name(&self, w: usize) -> String {
        let w = self.walk_rep[w];
        format!("f{}_{w}", self.cache.walks[w].req.fn_id)
    }
}

/// `s` as a C string literal. Every byte that is not plain printable ASCII is
/// an octal escape, so UTF-8 passes through unchanged
fn c_str(s: &str) -> String {
    let mut lit = String::from("\"");

    for b in s.bytes() {
        match b {
            b'"' | b'\\' => lit.extend(['\\', b as char]),
            0x20..0x7f => lit.push(b as char),
            _ => lit += &format!("\\{b:03o}"),
        }
    }

    lit + "\""
}

fn is_num(rep: Rep) -> bool {
    matches!(rep, Rep::Int | Rep::Double)
}

/// The struct of the objects of cell `id`
fn struct_name(id: CellNum) -> String {
    format!("S{id}")
}

/// The static object of the built-in fn of cell `id`
fn builtin_name(id: CellNum) -> String {
    format!("b{id}")
}

/// The global that holds what the host calls when an IO op completes
const HOST_COMPLETE_GLOBAL: &str = "jsrt_host_complete";

fn proto_global(kind: ProtoOf) -> String {
    format!("proto_{kind:?}").to_lowercase()
}

fn is_terminator(op: OpCode) -> bool {
    use OpCode::*;
    matches!(
        op,
        Goto | Goto8 | Goto16 | IfFalse | IfFalse8 | IfTrue | IfTrue8 | Return | ReturnAsync | ReturnUndef
    )
}