ww/lib/ww/typ.ww

// lib/ww/typ.ww — port of cmd/wcc/type.c.
//
// Status: full structural port. The C version uses module-globals for
// the primitive types (tyvoid, tyi32, …); ww doesn't have writable
// global storage yet, so we bundle the primitives into a `tctx` that
// the checker passes around explicitly. typesinit fills the tctx
// once per program.

package ww;

import os;

// ---- TypeKind ---------------------------------------------------------
// Numeric values must stay aligned with cmd/wcc/ww.h TypeKind so the
// next diff signal (typed-AST printer / cgen) can compare across the
// two implementations.

// Mirror of the C `TypeKind` enum in cmd/wcc/ww.h. Numeric values
// are explicit and must stay in sync — the selfhost selfcheck and
// typed-AST printers depend on matching numeric layout.
type tykind = enum i32 {
	TY_NONE     = 0,
	TY_VOID     = 1,
	TY_BOOL     = 2,
	TY_RUNE     = 3,
	TY_I8       = 4,
	TY_I16      = 5,
	TY_I32      = 6,
	TY_I64      = 7,
	TY_U8       = 8,
	TY_U16      = 9,
	TY_U32      = 10,
	TY_U64      = 11,
	TY_UINT     = 12,
	TY_INT      = 13,
	TY_UINTPTR  = 14,
	TY_F32      = 15,
	TY_F64      = 16,
	TY_STR      = 17,
	TY_PTR      = 18,
	TY_SLICE    = 19,
	TY_ARRAY    = 20,
	TY_STRUCT   = 21,
	TY_FN       = 22,
	TY_CHAN     = 23,
	TY_NAMED    = 24,
	TY_TUPLE    = 25,
	TY_TAGGED   = 26,
	TY_ERR      = 27,
	TY_NEVER    = 28,
	TY_UNTYPED_INT   = 29,
	TY_UNTYPED_FLOAT = 30,
	TY_UNTYPED_STR   = 31,
	TY_UNTYPED_RUNE  = 32,
	TY_UNTYPED_BOOL  = 33,
	TY_UNTYPED_NIL   = 34,
	// Tail-appended values keep prior TY_* stable for the byte-diff
	// against cmd/wcc/ww.h.
	TY_ENUM     = 35,
};

// ---- tinfo / tfield / tparam -----------------------------------------

type tfield = struct {
	name:   str,
	type_:  *tinfo,
	offset: u64,
	tnext:  *tfield,
};

type tparam = struct {
	name:   str,
	type_:  *tinfo,
	// #61a: per-variant `!T` error mark for TY_TAGGED variants.
	// cstage-MIRROR divergence: harec carries no per-variant flag —
	// it models `!T` as a distinct STORAGE_ERROR type node
	// (ref/harec/include/types.h:144, src/types.c:151-159
	// type_is_error). wwstage tinfo has no iserror field
	// (check.ww TTAGGED arm), so the bit rides the shared param
	// struct instead, matching cstage Type.iserror semantics.
	// Faithful STORAGE_ERROR-node port filed as #62.
	iserror: bool,
	tnext:  *tparam,
};

// #57 A.6.3i-phase-1: tuple positional element. Distinct from tfield
// (named, struct member) per harec's split at ref/harec/include/types.h
// :109-115 (struct_field) vs :122-126 (type_tuple) — tuple positionals
// carry no name (positional only) and a separate next-link. Rule-12
// sea-of-stars mirrors Hare's structural choice; the empty-name idiom
// from #50's TTAGGED-on-tparam would conflate two semantic axes
// (variants can be named; positionals never can).
type ttupleelem = struct {
	type_:  *tinfo,
	offset: u64,
	tnext:  *ttupleelem,
};

type tinfo = struct {
	kind: tykind,
	size:     u64,
	align:    u64,
	sub:      *tinfo,	// ptr/slice/array/chan element
	alen:     u64,
	fields:   *tfield,
	params:   *tparam,
	tupleelems: *ttupleelem,	// #57 A.6.3i-phase-1: TY_TUPLE
				// positional chain (harec types.h:122-126
				// `struct type_tuple`). Distinct slot from
				// .fields so struct-member vs tuple-
				// positional stay axis-separated.
	ret:      *tinfo,
	variadic: i32,
	nullable: i32,	// #61 A.3: TY_TAGGED `(*T | void)` fold collapses to
				// 8B ptr slot (null is the void variant). Mirrors
				// cstage Type.nullable (cmd/wcc/ww.h:430-433).
	name:     str,
	under:    *tinfo,
	slotsize: u64,	// #61 A.5: stack-slot SSoT split from `size`.
				// `size` stays natural (Hare-faithful);
				// `slotsize` carries the slot-padded width
				// cgen's let/struct-field layout demands.
				// For primitives/ptr/slice/chan/fn/str/tagged
				// `slotsize == size`; struct + tuple + array
				// of struct diverge — see check.ww tinfo-
				// fornode + cgenutil.ww registerstruct.
				// Pad-to-8 of narrow primitives in let slots
				// still lives at slotsize()'s read site;
				// graduating it here would break `[N]i32`
				// stride (4*N stays natural).
};

// #61 audit §1.8 / Rob+Drew convergence 2026-05-20: memoizes
// tinfofornode lookups keyed by AST pointer. Linked-list shape mirrors
// other wwstage-side caches (cgen.aliases, cgen.structs) — sea-of-stars
// over hash-table cleverness, and Sym/Scope already pay the FNV cost
// for the resolver pass.
type tinfocacheent = struct {
	key:    *node,
	val:    *tinfo,
	cnext:  *tinfocacheent,
};

// ---- tctx — the box of primitive types -------------------------------

type tctx = struct {
	tyvoid:  *tinfo,
	tybool:  *tinfo,
	tyrune:  *tinfo,
	tyi8:    *tinfo,
	tyi16:   *tinfo,
	tyi32:   *tinfo,
	tyi64:   *tinfo,
	tyu8:    *tinfo,
	tyu16:   *tinfo,
	tyu32:   *tinfo,
	tyu64:   *tinfo,
	tyint:   *tinfo,
	tyuint:  *tinfo,
	tyuintptr: *tinfo,
	tyf32:   *tinfo,
	tyf64:   *tinfo,
	tystr:   *tinfo,
	tyerr:   *tinfo,
	tynever: *tinfo,
	tyuntypedint:   *tinfo,
	tyuntypedfloat: *tinfo,
	tyuntypedstr:   *tinfo,
	tyuntypedrune:  *tinfo,
	tyuntypedbool:  *tinfo,
	tyuntypednil:   *tinfo,
	tinfocache:    *tinfocacheent,
};

// ---- constructors -----------------------------------------------------

export fn newtype(k: tykind) *tinfo = {
	let t: *tinfo = alloc(tinfo{kind=k, size=0u64, align=0u64, sub=nil, alen=0u64, fields=nil, params=nil, tupleelems=nil, ret=nil, variadic=0, nullable=0, name="", under=nil, slotsize=0u64})!;
	return t;
};

fn prim(k: tykind, nm: str, sz: u64, al: u64) *tinfo = {
	let t: *tinfo = newtype(k);
	t.name = nm;
	t.size = sz;
	if (al > 0u64) { t.align = al; } else { t.align = sz; };
	t.slotsize = sz;
	return t;
};

export fn typesinit(c: *tctx) void = {
	c.tyvoid   = prim(tykind.TY_VOID, "void",  0u64, 1u64);
	c.tybool   = prim(tykind.TY_BOOL, "bool",  1u64, 1u64);
	c.tyrune   = prim(tykind.TY_RUNE, "rune",  4u64, 4u64);
	c.tyi8     = prim(tykind.TY_I8,   "i8",    1u64, 1u64);
	c.tyi16    = prim(tykind.TY_I16,  "i16",   2u64, 2u64);
	c.tyi32    = prim(tykind.TY_I32,  "i32",   4u64, 4u64);
	c.tyi64    = prim(tykind.TY_I64,  "i64",   8u64, 8u64);
	c.tyu8     = prim(tykind.TY_U8,   "u8",    1u64, 1u64);
	c.tyu16    = prim(tykind.TY_U16,  "u16",   2u64, 2u64);
	c.tyu32    = prim(tykind.TY_U32,  "u32",   4u64, 4u64);
	c.tyu64    = prim(tykind.TY_U64,  "u64",   8u64, 8u64);
	c.tyint    = prim(tykind.TY_INT,  "int",   8u64, 8u64);
	c.tyuint   = prim(tykind.TY_UINT, "uint",  8u64, 8u64);
	c.tyuintptr= prim(tykind.TY_UINTPTR, "uintptr", 8u64, 8u64);
	c.tyf32    = prim(tykind.TY_F32,  "f32",   4u64, 4u64);
	c.tyf64    = prim(tykind.TY_F64,  "f64",   8u64, 8u64);
	c.tystr    = prim(tykind.TY_STR,  "str",  16u64, 8u64); // sizelint-ok: SSoT for tystr (#64)
	c.tyerr    = prim(tykind.TY_ERR,  "<err>", 0u64, 1u64);
	c.tynever  = prim(tykind.TY_NEVER, "never", 0u64, 1u64);

	c.tyuntypedint   = prim(tykind.TY_UNTYPED_INT,   "untyped_int",   0u64, 1u64);
	c.tyuntypedfloat = prim(tykind.TY_UNTYPED_FLOAT, "untyped_float", 0u64, 1u64);
	c.tyuntypedstr   = prim(tykind.TY_UNTYPED_STR,   "untyped_str",   0u64, 1u64);
	c.tyuntypedrune  = prim(tykind.TY_UNTYPED_RUNE,  "untyped_rune",  0u64, 1u64);
	c.tyuntypedbool  = prim(tykind.TY_UNTYPED_BOOL,  "untyped_bool",  0u64, 1u64);
	c.tyuntypednil   = prim(tykind.TY_UNTYPED_NIL,   "untyped_nil",   0u64, 1u64);
};

export fn typeptr(sub: *tinfo) *tinfo = {
	let t: *tinfo = newtype(tykind.TY_PTR);
	t.sub = sub;
	t.size = 8u64;
	t.align = 8u64;
	t.slotsize = 8u64;
	return t;
};

export fn typeslice(sub: *tinfo) *tinfo = {
	let t: *tinfo = newtype(tykind.TY_SLICE);
	t.sub = sub;
	t.size = 24u64;		// sizelint-ok: SSoT for slice header (#64)
	t.align = 8u64;
	t.slotsize = 24u64;	// sizelint-ok: SSoT for slice slotsize (#64)
	return t;
};

export fn typearray(sub: *tinfo, n: u64) *tinfo = {
	let t: *tinfo = newtype(tykind.TY_ARRAY);
	t.sub = sub;
	t.alen = n;
	if (sub != nil) {
		t.size = sub.size * n;
		t.align = sub.align;
		// #61 A.5: ti.slotsize = stride * elen using the element's
		// slot-padded width. Primitives have slotsize == size so
		// `[N]i32` stride stays 4 (natural); structs have padded
		// slotsize so `[N]Triplet` stride lifts to 16.
		t.slotsize = sub.slotsize * n;
	} else {
		t.align = 1u64;
	};
	return t;
};

export fn typechan(sub: *tinfo) *tinfo = {
	let t: *tinfo = newtype(tykind.TY_CHAN);
	t.sub = sub;
	t.size = 8u64;
	t.align = 8u64;
	t.slotsize = 8u64;
	return t;
};

export fn typenamed(name: str, under: *tinfo) *tinfo = {
	let t: *tinfo = newtype(tykind.TY_NAMED);
	t.name = name;
	t.under = under;
	if (under != nil) {
		t.size = under.size;
		t.align = under.align;
		t.slotsize = under.slotsize;
	};
	return t;
};

// #61 audit §1.8 — A.1 infrastructure: tinfocache lookup/bind. Keyed
// by AST node-pointer so two different N_TNAME("i32") nodes get
// independent entries that both resolve to c.tyi32. Used by
// tinfofornode in check.ww; cgen still reads sizes via primtypesize
// until A.2+ graduates each walker family.
export fn tinfocachelookup(c: *tctx, key: *node) *tinfo = {
	let e: *tinfocacheent = c.tinfocache;
	for (e != nil) {
		if (e.key == key) { return e.val; };
		e = e.cnext;
	};
	return nil;
};

export fn tinfocachebind(c: *tctx, key: *node, val: *tinfo) void = {
	let e: *tinfocacheent = alloc(tinfocacheent{key=key, val=val, cnext=c.tinfocache})!;
	c.tinfocache = e;
};

// ---- predicates -------------------------------------------------------

export fn typeisint(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_I8)  { return true; };
	if (k == tykind.TY_I16) { return true; };
	if (k == tykind.TY_I32) { return true; };
	if (k == tykind.TY_I64) { return true; };
	if (k == tykind.TY_U8)  { return true; };
	if (k == tykind.TY_U16) { return true; };
	if (k == tykind.TY_U32) { return true; };
	if (k == tykind.TY_U64) { return true; };
	if (k == tykind.TY_INT) { return true; };
	if (k == tykind.TY_UINT){ return true; };
	if (k == tykind.TY_UINTPTR) { return true; };
	if (k == tykind.TY_RUNE){ return true; };
	if (k == tykind.TY_UNTYPED_INT)  { return true; };
	if (k == tykind.TY_UNTYPED_RUNE) { return true; };
	if (k == tykind.TY_ENUM)         { return typeisint(t.sub); };
	if (k == tykind.TY_NAMED) { return typeisint(t.under); };
	return false;
};

export fn typeisfloat(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_F32) { return true; };
	if (k == tykind.TY_F64) { return true; };
	if (k == tykind.TY_UNTYPED_FLOAT) { return true; };
	if (k == tykind.TY_NAMED) { return typeisfloat(t.under); };
	return false;
};

export fn typeisnum(t: *tinfo) bool = {
	if (typeisint(t)) { return true; };
	return typeisfloat(t);
};

// TY_RUNE is unsigned: Unicode codepoint (0..0x10FFFF) zero-extends on
// sub-word load (MOVL, not MOVSXD). TY_ENUM recurses on .sub so a
// `type k = enum u32 {…}` reads as unsigned. Cite cstage type.c:178
// `type_isunsigned`; rule 10 keeps wwstage aligned down to cstage.
export fn typeisunsigned(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_U8)  { return true; };
	if (k == tykind.TY_U16) { return true; };
	if (k == tykind.TY_U32) { return true; };
	if (k == tykind.TY_U64) { return true; };
	if (k == tykind.TY_UINT){ return true; };
	if (k == tykind.TY_UINTPTR) { return true; };
	if (k == tykind.TY_RUNE){ return true; };
	if (k == tykind.TY_NAMED) { return typeisunsigned(t.under); };
	if (k == tykind.TY_ENUM)  { return typeisunsigned(t.sub); };
	return false;
};

// typeissigned — does this type need sign-extension on a sub-word
// (1/2/4B) load? Mirrors cstage cgen.c:240 `fld_issigned`. Cgen-facing
// predicate (TY_BOOL is unsigned for storage purposes — 0/1 → MOVZBQ),
// so it doesn't simply mirror `!typeisunsigned`. Pair-of-`is*`
// convention follows ref/hare/types/ helpers.
export fn typeissigned(t: *tinfo) bool = {
	if (t == nil) { return false; };
	if (t.kind == tykind.TY_BOOL) { return false; };
	if (typeisunsigned(t)) { return false; };
	return typeisint(t);
};

// typeisstr — TY_STR (and TY_UNTYPED_STR for literals pre-default).
// Cite cstage cgen.c:159 `type_isstr` — single TY_NAMED peel, accepts
// the same untyped form. ww walks the .under chain so alias-of-alias
// (`type s2 = s1; type s1 = str;`) lands the same way.
export fn typeisstr(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_STR) { return true; };
	if (k == tykind.TY_UNTYPED_STR) { return true; };
	if (k == tykind.TY_NAMED) { return typeisstr(t.under); };
	return false;
};

// typeisslice — TY_SLICE. Cite cstage cgen.c:174 `type_isslice`.
export fn typeisslice(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_SLICE) { return true; };
	if (k == tykind.TY_NAMED) { return typeisslice(t.under); };
	return false;
};

// typeistagged — TY_TAGGED (alias-aware). Cite cstage cgen.c:516
// `type_istagged` — same single-peel shape. The node-keyed wwstage
// helper this replaces also unwrapped a leading N_TBANG so
// `type error = !(invalid | overflow);` registered as tagged. Post-
// A.6.2 the TBANG unwrap is handled by tinfofornode (check.ww:1145-
// 1152 returns the inner tinfo unchanged) so we recover the cstage
// semantics with the bare kind check + NAMED chase.
export fn typeistagged(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_TAGGED) { return true; };
	if (k == tykind.TY_NAMED) { return typeistagged(t.under); };
	return false;
};

// typeisf32 — narrower-than-typeisfloat: only TY_F32 (after alias
// chase). Cite cstage cgen.c:188 `type_isf32`. Used to pick MOVSS vs
// MOVSD and the SS-variant arithmetic / cast opcodes.
export fn typeisf32(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_F32) { return true; };
	if (k == tykind.TY_NAMED) { return typeisf32(t.under); };
	return false;
};

// typeisnullable — TY_TAGGED with the `(*T | void)` one-word fold.
// Cite cstage cgen.c:396 `type_isnullable`. The .nullable flag is
// stamped by tinfofornode (check.ww:1309-1318) when the two-variant
// shape matches.
export fn typeisnullable(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_TAGGED) { return t.nullable != 0; };
	if (k == tykind.TY_NAMED) { return typeisnullable(t.under); };
	return false;
};

// typeis8byteprim — does this type take exactly one 8-byte stack
// slot (ptr / fn / chan / 64-bit int / scalar primitive padded up to
// 8 / `[N]T` whose natural width is 8) rather than a wider aggregate?
// Mirrors the ladder cstage's cgen.c N_LET zero-init takes on `sz==8`
// (cmd/wcc/check.c sizing + cgen.c N_LET). The node-keyed wwstage
// helper this replaces predates tinfo and AST-walked TBANG / TNAME
// alias chains; tinfofornode now collapses TBANG (check.ww:1145) and
// TY_NAMED.under carries the chain, so the tinfo walk handles every
// shape the AST walker did.
export fn typeis8byteprim(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_PTR)    { return true; };
	if (k == tykind.TY_FN)     { return true; };
	if (k == tykind.TY_CHAN)   { return true; };
	if (k == tykind.TY_SLICE)  { return false; };
	if (k == tykind.TY_TUPLE)  { return false; };
	if (k == tykind.TY_TAGGED) { return false; };
	if (k == tykind.TY_STR)    { return false; };
	if (k == tykind.TY_STRUCT) { return false; };
	if (k == tykind.TY_ARRAY)  { return t.size == 8u64; };
	if (k == tykind.TY_NAMED)  { return typeis8byteprim(t.under); };
	// Remaining: primitives (i8/u8/.../i64/u64/bool/rune/f32/f64/
	// int/uint/uintptr) and TY_VOID. All slot-pad to 8 and zero-init
	// in cstage's `sz==8` branch.
	return true;
};

export fn typeisuntyped(t: *tinfo) bool = {
	if (t == nil) { return false; };
	let k: tykind = t.kind;
	if (k == tykind.TY_UNTYPED_INT)   { return true; };
	if (k == tykind.TY_UNTYPED_FLOAT) { return true; };
	if (k == tykind.TY_UNTYPED_STR)   { return true; };
	if (k == tykind.TY_UNTYPED_RUNE)  { return true; };
	if (k == tykind.TY_UNTYPED_BOOL)  { return true; };
	if (k == tykind.TY_UNTYPED_NIL)   { return true; };
	return false;
};

// typeeq — structural equality. Named types compare nominally.
export fn typeeq(a: *tinfo, b: *tinfo) bool = {
	if (a == b) { return true; };
	if (a == nil) { return false; };
	if (b == nil) { return false; };
	if (a.kind != b.kind) { return false; };
	let k: tykind = a.kind;
	if (k == tykind.TY_PTR)   { return typeeq(a.sub, b.sub); };
	if (k == tykind.TY_SLICE) { return typeeq(a.sub, b.sub); };
	if (k == tykind.TY_CHAN)  { return typeeq(a.sub, b.sub); };
	if (k == tykind.TY_ARRAY) {
		if (a.alen != b.alen) { return false; };
		return typeeq(a.sub, b.sub);
	};
	if (k == tykind.TY_FN) {
		if (a.variadic != b.variadic) { return false; };
		if (!typeeq(a.ret, b.ret))   { return false; };
		let pa: *tparam = a.params;
		let pb: *tparam = b.params;
		for (true) {
			if (pa == nil) { if (pb == nil) { return true; }; return false; };
			if (pb == nil) { return false; };
			if (!typeeq(pa.type_, pb.type_)) { return false; };
			pa = pa.tnext;
			pb = pb.tnext;
		};
		return true;
	};
	if (k == tykind.TY_STRUCT) {
		let fa: *tfield = a.fields;
		let fb: *tfield = b.fields;
		for (true) {
			if (fa == nil) { if (fb == nil) { return true; }; return false; };
			if (fb == nil) { return false; };
			let na: str = fa.name;
			let nb: str = fb.name;
			if (na.len != nb.len) { return false; };
			let i: i32 = 0;
			for (i < na.len) {
				if (na[i] != nb[i]) { return false; };
				i += 1;
			};
			if (!typeeq(fa.type_, fb.type_)) { return false; };
			fa = fa.tnext;
			fb = fb.tnext;
		};
		return true;
	};
	if (k == tykind.TY_NAMED) { return false; };	// nominal: only same ptr
	if (k == tykind.TY_TUPLE) {
		let pa: *tparam = a.params;
		let pb: *tparam = b.params;
		for (true) {
			if (pa == nil) { if (pb == nil) { return true; }; return false; };
			if (pb == nil) { return false; };
			if (!typeeq(pa.type_, pb.type_)) { return false; };
			pa = pa.tnext;
			pb = pb.tnext;
		};
		return true;
	};
	return true;	// primitives match by kind alone
};