ww/selfhost/cmd/wcc/check.ww

// selfhost/cmd/wcc/check.ww — minimal port of cmd/wcc/check.c.
//
// Status: name-resolution + primitive-type seeding only. Full type
// inference, conversion rules, tagged-union dispatch typing, return-
// type checking, etc. all live in cmd/wcc/check.c (937 lines) and
// will land here in subsequent commits.
//
// What this version does:
//   1. Creates a top scope and seeds it with primitive type names so
//      `i32`, `str`, `*u8` etc. resolve.
//   2. Walks the file's top-level decls (use/def/type/fn/let) and
//      installs Sym entries for each.
//   3. Recursively walks fn bodies; for every N_IDENT used as an
//      expression or as a type name, looks it up and counts the
//      resolved vs. unresolved.
//   4. Returns a summary the caller (wwdump -r) prints; the test
//      asserts unresolved == 0 on every selfhost fixture, which is
//      the floor signal that the frontend can name-resolve real ww.

use os;
use mem;
use tok;

type checker = struct {
	a:        *arena,
	tc:       *tctx,
	top:      *scope,
	cur:      *scope,
	nresolved:   i32,
	nunresolved: i32,
	errs:        i32,
	verbose:     i32,	// when non-zero, log each unresolved name
	fnret:       *node,	// enclosing fn's return type AST (for `?`)
};

// seedprimitives — install the built-in type names so `i32`, `str`,
// etc. can be looked up like ordinary symbols.
fn seedprimitives(c: *checker) void = {
	scopedefine(c.top, "void",    SK_TYPE, c.tc.tyvoid,    nil);
	scopedefine(c.top, "bool",    SK_TYPE, c.tc.tybool,    nil);
	scopedefine(c.top, "rune",    SK_TYPE, c.tc.tyrune,    nil);
	scopedefine(c.top, "i8",      SK_TYPE, c.tc.tyi8,      nil);
	scopedefine(c.top, "i16",     SK_TYPE, c.tc.tyi16,     nil);
	scopedefine(c.top, "i32",     SK_TYPE, c.tc.tyi32,     nil);
	scopedefine(c.top, "i64",     SK_TYPE, c.tc.tyi64,     nil);
	scopedefine(c.top, "u8",      SK_TYPE, c.tc.tyu8,      nil);
	scopedefine(c.top, "u16",     SK_TYPE, c.tc.tyu16,     nil);
	scopedefine(c.top, "u32",     SK_TYPE, c.tc.tyu32,     nil);
	scopedefine(c.top, "u64",     SK_TYPE, c.tc.tyu64,     nil);
	scopedefine(c.top, "int",     SK_TYPE, c.tc.tyint,     nil);
	scopedefine(c.top, "uint",    SK_TYPE, c.tc.tyuint,    nil);
	scopedefine(c.top, "uintptr", SK_TYPE, c.tc.tyuintptr, nil);
	scopedefine(c.top, "f32",     SK_TYPE, c.tc.tyf32,     nil);
	scopedefine(c.top, "f64",     SK_TYPE, c.tc.tyf64,     nil);
	scopedefine(c.top, "str",     SK_TYPE, c.tc.tystr,     nil);
	scopedefine(c.top, "never",   SK_TYPE, c.tc.tynever,   nil);
	// `nil`, `true`, `false` are keywords — handled at the lex/parser
	// level, no symbol needed.
	// `len`, `alloc`, `free` are pseudo-builtins; scopedefine them so
	// their use sites resolve. The actual semantics live in cgen.
	scopedefine(c.top, "len",   SK_FN, nil, nil);
	scopedefine(c.top, "alloc", SK_FN, nil, nil);
	scopedefine(c.top, "free",  SK_FN, nil, nil);
};

// installdecl — install the top-level decl's name into the top scope.
// We don't compute its type yet (that's the resolve pass) — just bind
// the name so forward references resolve.
fn installdecl(c: *checker, d: *node) void = {
	if (d == nil) { return; };
	let k: i32 = d.kind;
	let nm: str = d.str;
	if (k == N_USE)      { scopedefine(c.top, nm, SK_USE,  nil, d); return; };
	if (k == N_DEF)      { scopedefine(c.top, nm, SK_DEF,  nil, d); return; };
	if (k == N_TYPEDECL) { scopedefine(c.top, nm, SK_TYPE, nil, d); return; };
	if (k == N_FNDECL)   { scopedefine(c.top, nm, SK_FN,   nil, d); return; };
	if (k == N_LET)      { scopedefine(c.top, nm, SK_VAR,  nil, d); return; };
};

// resolvewalk — recursive AST walk that, for every N_IDENT and
// N_TNAME seen, looks up the name and bumps the resolved/unresolved
// counters. Local lets are installed in the current scope as soon as
// their init/type expressions have been walked (forward use of a let
// before its declaration would resolve to nothing — same semantics as
// the C checker's collect-then-resolve flow within a function).
// Also runs the typed checks (match exhaustiveness, ? subset) in
// the same pass — they need the same scope state.
fn resolvewalk(c: *checker, n: *node) void = {
	if (n == nil) { return; };
	let k: i32 = n.kind;

	// Typed checks fire on the way down so the scrutinee/operand
	// is examined before the arm bodies install new bindings.
	if (k == N_MATCH)   { check_match_exhaustive(c, n); };
	if (k == N_TRYPROP) { check_tryprop(c, n); };

	// `use IDENT;` — name is a module label, not a free ident.
	if (k == N_USE) { return; };

	if (k == N_IDENT) {
		let nm: str = n.str;
		if (nm.len > 0) {
			let s: *sym = scopelookup(c.cur, nm);
			if (s == nil) {
				c.nunresolved += 1;
				if (c.verbose != 0) {
					os.write(2, "  unresolved id: ".ptr, 17u64);
					os.write(2, nm.ptr, nm.len: u64);
					os.write(2, "\n".ptr, 1u64);
				};
			} else { c.nresolved += 1; };
		};
	};

	if (k == N_TNAME) {
		let nm: str = n.str;
		if (nm.len > 0) {
			let s: *sym = scopelookup(c.cur, nm);
			// `pkg.Type` — strip the last dot prefix and look up
			// the leaf if `pkg` is a use-imported name. Mirrors
			// cmd/wcc/check.c resolve_typename.
			if (s == nil) {
				let dot: i32 = nm.len - 1;
				for (dot >= 0) {
					if (nm[dot] == 46u8) { break; };
					dot -= 1;
				};
				if (dot > 0) {
					let head: str;
					head.ptr = nm.ptr;
					head.len = dot;
					let m: *sym = scopelookup(c.cur, head);
					if (m != nil) {
						let leaf: str;
						leaf.ptr = nm.ptr + (dot + 1): u64;
						leaf.len = nm.len - (dot + 1);
						s = scopelookup(c.cur, leaf);
					};
				};
			};
			if (s == nil) {
				c.nunresolved += 1;
				if (c.verbose != 0) {
					os.write(2, "  unresolved tname: ".ptr, 20u64);
					os.write(2, nm.ptr, nm.len: u64);
					os.write(2, "\n".ptr, 1u64);
				};
			} else { c.nresolved += 1; };
		};
	};

	// `match (e) { case let v: T => stmt; ... }` — the binding `v`
	// is declared by the case arm and visible inside its body.
	if (k == N_MCASE) {
		if (n.lhs  != nil) { resolvewalk(c, n.lhs);  };
		let nm: str = n.str;
		if (nm.len > 0) {
			scopedefine(c.cur, nm, SK_VAR, nil, n);
		};
		if (n.body != nil) { resolvewalk(c, n.body); };
		return;
	};

	if (k == N_DOT) {
		// Walk only the base; the .field name is a member, not a
		// free identifier.
		if (n.lhs != nil) { resolvewalk(c, n.lhs); };
		return;
	};

	if (k == N_FIELD) {
		if (n.lhs != nil) { resolvewalk(c, n.lhs); };
		return;
	};

	if (k == N_TFIELD) {
		if (n.lhs != nil) { resolvewalk(c, n.lhs); };
		return;
	};

	// Walk children (mirroring ast.ww's printer descent order).
	if (n.attr != nil) { resolvewalk(c, n.attr); };
	if (n.lhs  != nil) { resolvewalk(c, n.lhs);  };
	if (n.rhs  != nil) { resolvewalk(c, n.rhs);  };
	if (n.cond != nil) { resolvewalk(c, n.cond); };
	if (n.body != nil) { resolvewalk(c, n.body); };
	if (n.els  != nil) { resolvewalk(c, n.els);  };
	if (n.list != nil) {
		let m: *node = n.list;
		for (m != nil) {
			resolvewalk(c, m);
			m = m.next;
		};
	};

	// After walking children: a local `let X: T = init;` registers
	// `X` so subsequent statements can resolve it. Top-level lets
	// are installed in installdecl, so this duplicate install at
	// the file scope just no-ops (scopedefine returns nil on dup).
	if (k == N_LET) {
		let nm: str = n.str;
		if (nm.len > 0) {
			scopedefine(c.cur, nm, SK_VAR, nil, n);
		};
	};
};

// ---- type-level helpers (AST-level, no resolved tinfo) --------------
//
// The selfhost check operates on AST type expressions rather than
// resolved Type structs. These helpers mirror what cmd/wcc/check.c
// does with tinfo, but only on the subset of cases this checker
// needs to enforce: tagged-union exhaustiveness, ? subset
// propagation, and !-flag semantics.

// unwrapbang — strip an N_TBANG wrapper; leaves other nodes alone.
fn unwrapbang(n: *node) *node = {
	if (n == nil) { return nil; };
	if (n.kind == N_TBANG) { return n.lhs; };
	return n;
};

// resolvealias — if n is an N_TNAME pointing at a typedecl, return
// the typedecl's body (possibly recursively). Pass-through for any
// other node. The chain stops once we hit a non-N_TNAME node or a
// name we can't resolve.
fn resolvealias(c: *checker, n: *node) *node = {
	let cur: *node = n;
	for (cur != nil) {
		if (cur.kind != N_TNAME) { return cur; };
		let s: *sym = scopelookup(c.cur, cur.str);
		if (s == nil) { return cur; };
		if (s.skind != SK_TYPE) { return cur; };
		let body: *node = nil;
		if (s.decl != nil) { body = s.decl.lhs; };
		if (body == nil) { return cur; };
		cur = unwrapbang(body);
	};
	return n;
};

// is_tagged_type — true if `n` (after alias resolution) is an
// N_TTAGGED type expression.
fn is_tagged_type(c: *checker, n: *node) bool = {
	let u: *node = resolvealias(c, unwrapbang(n));
	if (u == nil) { return false; };
	return u.kind == N_TTAGGED;
};

// type_eq_ast — structural equality on AST type expressions, mod
// the `!` wrapper. Mirrors variant_match in cgen + check.c: NAMED
// types compare by string (the closest stand-in for pointer
// identity at the AST level); other nodes recurse by kind.
fn type_eq_ast(a: *node, b: *node) bool = {
	let aa: *node = unwrapbang(a);
	let bb: *node = unwrapbang(b);
	if (aa == nil) { return bb == nil; };
	if (bb == nil) { return false; };
	if (aa.kind != bb.kind) { return false; };
	let k: i32 = aa.kind;
	if (k == N_TNAME) { return streq(aa.str, bb.str); };
	if (k == N_TPTR)  { return type_eq_ast(aa.lhs, bb.lhs); };
	if (k == N_TSLICE){ return type_eq_ast(aa.lhs, bb.lhs); };
	if (k == N_TCHAN) { return type_eq_ast(aa.lhs, bb.lhs); };
	// Conservative: anything else (struct/fn/tagged/tuple/array)
	// fails the cheap check. Selfhost code doesn't currently rely
	// on equality at these shapes for the targeted checks.
	return false;
};

// variant_is_error — does this variant carry the `!` mark? Either
// the variant itself is N_TBANG or it's an alias whose typedecl
// body is `!T`. Mirrors C check.c's iserror-after-NAMED rule.
fn variant_is_error(c: *checker, v: *node) bool = {
	if (v == nil) { return false; };
	if (v.kind == N_TBANG) { return true; };
	if (v.kind == N_TNAME) {
		let s: *sym = scopelookup(c.cur, v.str);
		if (s != nil) {
			if (s.skind == SK_TYPE) {
				if (s.decl != nil) {
					if (s.decl.lhs != nil) {
						if (s.decl.lhs.kind == N_TBANG) {
							return true;
						};
					};
				};
			};
		};
	};
	return false;
};

// tagged_has_errflag — true iff any variant of `n` (assumed
// N_TTAGGED) is `!`-marked. Picks the explicit-flag semantics over
// the legacy "first variant = success" rule.
fn tagged_has_errflag(c: *checker, n: *node) bool = {
	let v: *node = n.list;
	for (v != nil) {
		if (variant_is_error(c, v)) { return true; };
		v = v.next;
	};
	return false;
};

// is_error_variant — under flag-aware mode (any !-marked variant),
// returns true iff `v` is `!`-marked. Under legacy mode (no flags),
// returns true iff `v` is not the first variant of `tagged`.
fn is_error_variant(c: *checker, tagged: *node, v: *node) bool = {
	if (tagged_has_errflag(c, tagged)) {
		return variant_is_error(c, v);
	};
	// Legacy: first variant of the union is success.
	if (tagged.list == v) { return false; };
	return true;
};

// scrutinee_type — resolve the type expression for a match's
// scrutinee. Handles N_IDENT (look up local/param's declared
// type) and N_DOT (struct-field access). Returns nil if we
// can't statically determine the type. Used by exhaustiveness.
fn scrutinee_type(c: *checker, e: *node) *node = {
	if (e == nil) { return nil; };
	if (e.kind == N_IDENT) {
		let s: *sym = scopelookup(c.cur, e.str);
		if (s == nil) { return nil; };
		if (s.decl == nil) { return nil; };
		// For N_LET / N_PARAM: declared type is decl.lhs.
		return s.decl.lhs;
	};
	return nil;
};

// ---- match exhaustiveness --------------------------------------------
//
// For every match arm, verify that every variant of the scrutinee's
// tagged-union type is handled by some case (or a default arm
// exists). Multi-pattern `case A | B =>` covers all alts.

fn case_covers(c: *checker, cs: *node, want: *node) bool = {
	if (cs.lhs != nil) {
		if (type_eq_ast(cs.lhs, want)) { return true; };
	};
	let alt: *node = cs.list;
	for (alt != nil) {
		if (type_eq_ast(alt, want)) { return true; };
		alt = alt.next;
	};
	return false;
};

fn err_match_variant(c: *checker, n: *node, vname: *node) void = {
	os.write(2, "match: variant not handled".ptr, 26u64);
	if (vname != nil) {
		if (vname.kind == N_TNAME) {
			os.write(2, " (".ptr, 2u64);
			os.write(2, vname.str.ptr, vname.str.len: u64);
			os.write(2, ")".ptr, 1u64);
		};
	};
	os.write(2, "\n".ptr, 1u64);
	c.errs += 1;
};

fn check_match_exhaustive(c: *checker, n: *node) void = {
	if (n == nil) { return; };
	if (n.lhs == nil) { return; };
	let st: *node = scrutinee_type(c, n.lhs);
	let u: *node = resolvealias(c, unwrapbang(st));
	if (u == nil) { return; };
	if (u.kind != N_TTAGGED) { return; };
	// Default arm absorbs anything; skip.
	let cs: *node = n.list;
	for (cs != nil) {
		if (cs.lhs == nil) { return; };	// default
		cs = cs.next;
	};
	// For each variant of u, look for a covering case.
	let v: *node = u.list;
	for (v != nil) {
		let covered: bool = false;
		let cs2: *node = n.list;
		for (cs2 != nil) {
			if (case_covers(c, cs2, v)) {
				covered = true;
				cs2 = nil;
			} else {
				cs2 = cs2.next;
			};
		};
		if (!covered) { err_match_variant(c, n, v); };
		v = v.next;
	};
};

// ---- ? subset propagation --------------------------------------------
//
// For `expr?`, the operand's error subset must be a subset of the
// enclosing fn's return-type variants. Mirrors C check.c. Operand
// is N_TRYPROP; its lhs is the value-bearing expr; we look at the
// expr's *declared* type for N_IDENT/N_CALL cases.

fn expr_type_for_tryprop(c: *checker, e: *node) *node = {
	if (e == nil) { return nil; };
	if (e.kind == N_IDENT) {
		let s: *sym = scopelookup(c.cur, e.str);
		if (s == nil) { return nil; };
		if (s.decl == nil) { return nil; };
		return s.decl.lhs;
	};
	if (e.kind == N_CALL) {
		// callee return type lookup: callee is e.lhs (N_IDENT or
		// N_DOT). We need the fn-decl's lhs (return-type AST).
		let callee: *node = e.lhs;
		if (callee == nil) { return nil; };
		let nm: str;
		nm.ptr = nil; nm.len = 0;
		if (callee.kind == N_IDENT) { nm = callee.str; };
		if (callee.kind == N_DOT)   { nm = callee.str; };
		if (nm.len == 0) { return nil; };
		let s: *sym = scopelookup(c.cur, nm);
		if (s == nil) { return nil; };
		if (s.skind != SK_FN) { return nil; };
		if (s.decl == nil) { return nil; };
		return s.decl.lhs;
	};
	return nil;
};

fn check_tryprop(c: *checker, n: *node) void = {
	if (n == nil) { return; };
	let t: *node = expr_type_for_tryprop(c, n.lhs);
	let u: *node = resolvealias(c, unwrapbang(t));
	if (u == nil) { return; };
	if (u.kind != N_TTAGGED) { return; };
	// Does the operand have any error variants?
	let has_err: bool = false;
	let v: *node = u.list;
	for (v != nil) {
		if (is_error_variant(c, u, v)) { has_err = true; };
		v = v.next;
	};
	if (!has_err) { return; };
	// Enclosing fn must return a tagged union with each operand
	// error variant present.
	let r: *node = resolvealias(c, unwrapbang(c.fnret));
	if (r == nil) {
		os.write(2, "?: enclosing fn has no tagged-union return\n".ptr, 43u64);
		c.errs += 1;
		return;
	};
	if (r.kind != N_TTAGGED) {
		os.write(2, "?: enclosing fn return is not tagged\n".ptr, 37u64);
		c.errs += 1;
		return;
	};
	let ev: *node = u.list;
	for (ev != nil) {
		if (is_error_variant(c, u, ev)) {
			let found: bool = false;
			let rv: *node = r.list;
			for (rv != nil) {
				if (type_eq_ast(rv, ev)) {
					found = true;
					rv = nil;
				} else { rv = rv.next; };
			};
			if (!found) {
				os.write(2, "?: error variant not in enclosing return\n".ptr, 41u64);
				c.errs += 1;
			};
		};
		ev = ev.next;
	};
};

// install_param — when entering a fn body, define its params in a
// fresh local scope.
fn installparams(c: *checker, params: *node) void = {
	let p: *node = params;
	for (p != nil) {
		if (p.kind == N_PARAM) {
			let nm: str = p.str;
			if (nm.len > 0) {
				scopedefine(c.cur, nm, SK_PARAM, nil, p);
			};
		};
		p = p.next;
	};
};

// resolvefnbody — open a child scope for the fn, install its params,
// then walk the body. Local lets installed by walk_stmt (a future
// extension); for the current pass we just resolve-walk without
// per-statement scopes.
fn resolvefnbody(c: *checker, fnnode: *node) void = {
	let outer: *scope = c.cur;
	c.cur = newscope(c.a, c.cur);
	installparams(c, fnnode.list);
	let prevret: *node = c.fnret;
	c.fnret = fnnode.lhs;	// return type AST, used by `?` check
	if (fnnode.body != nil) {
		resolvewalk(c, fnnode.body);
	};
	c.fnret = prevret;
	c.cur = outer;
};

export fn checkinit(c: *checker, a: *arena, tc: *tctx) void = {
	c.a = a;
	c.tc = tc;
	c.top = newscope(a, nil);
	c.cur = c.top;
	c.nresolved = 0;
	c.nunresolved = 0;
	c.errs = 0;
	c.verbose = 0;
	c.fnret = nil;
	seedprimitives(c);
};

export fn checkfile(c: *checker, file: *node) void = {
	if (file == nil) { return; };
	if (file.kind != N_FILE) { return; };

	// Pass 1: install all top-level names.
	let d: *node = file.list;
	for (d != nil) {
		installdecl(c, d);
		d = d.next;
	};

	// Pass 2: walk decl bodies/types and resolve identifiers.
	d = file.list;
	for (d != nil) {
		let k: i32 = d.kind;
		if (k == N_FNDECL) {
			if (d.lhs != nil) { resolvewalk(c, d.lhs); };	// return type
			resolvefnbody(c, d);
		} else { if (k == N_DEF) {
			if (d.lhs != nil) { resolvewalk(c, d.lhs); };
			if (d.rhs != nil) { resolvewalk(c, d.rhs); };
		} else { if (k == N_TYPEDECL) {
			if (d.lhs != nil) { resolvewalk(c, d.lhs); };
		} else { if (k == N_LET) {
			if (d.lhs != nil) { resolvewalk(c, d.lhs); };
			if (d.rhs != nil) { resolvewalk(c, d.rhs); };
		};};};};
		d = d.next;
	};

};