ww/lib/ww/ast.ww

// lib/ww/ast.ww — port of cmd/wcc/ast.c (Node defs + printer).
//
// Status: AST printer is fully ported. Constructor `newnode` is here.
// The parser (parse.ww) is currently minimal — see its file header.
//
// Calling-convention shim: same as tok/lex — `node` is too big to pass
// by value (8 *node pointers + 2 strs + a few ints), so callers always
// hand around `*node`. Only `newnode` allocates and returns a *node.

package ww;

import os;
import strconv;
import tok;

// ---- Nkind ------------------------------------------------------------
//
// Mirror of cmd/wcc/ww.h Nkind. Values must stay numerically equal so
// the AST diff probe in 990_selfhost works.

// Mirror of the C `Nkind` enum in cmd/wcc/ww.h. Numeric values are
// explicit and must stay in sync — the 990_selfhost test diffs
// astprint against the C side byte-for-byte. Tail-appended entries
// (TYPETEST onward) preserve every prior N_* value.
type nkind = enum i32 {
	N_NONE     = 0,

	N_INTLIT   = 1,
	N_FLOATLIT = 2,
	N_STRLIT   = 3,
	N_RUNELIT  = 4,
	N_TRUE     = 5,
	N_FALSE    = 6,
	N_NIL      = 7,
	N_IDENT    = 8,

	N_BIN       = 9,
	N_UN        = 10,
	N_CALL      = 11,
	N_INDEX     = 12,
	N_DOT       = 13,
	N_CAST      = 14,
	N_STRUCTLIT = 15,
	N_ARRLIT    = 16,
	N_FIELD     = 17,
	N_ASSIGN    = 18,
	N_ALLOC     = 19,
	N_FREE      = 20,
	N_RECV      = 21,
	N_SLICE     = 22,
	N_SPREAD    = 23,

	N_BLOCK    = 24,
	N_EXPRSTMT = 25,
	N_LET      = 26,
	N_RETURN   = 27,
	N_IF       = 28,
	N_FOR      = 29,
	N_FORRANGE = 30,
	N_DEFER    = 31,
	N_BREAK    = 32,
	N_CONTINUE = 33,
	N_SWITCH   = 34,
	N_CASE     = 35,

	N_FILE     = 36,
	N_USE      = 37,
	N_DEF      = 38,
	N_TYPEDECL = 39,
	N_FNDECL   = 40,
	N_PARAM    = 41,

	N_TNAME    = 42,
	N_TPTR     = 43,
	N_TSLICE   = 44,
	N_TARRAY   = 45,
	N_TFN      = 46,
	N_TSTRUCT  = 47,
	N_TFIELD   = 48,
	N_TCHAN    = 49,

	N_ATTR     = 50,
	N_TTUPLE   = 51,
	N_TTAGGED  = 52,
	N_TUPLE    = 53,
	N_MATCH    = 54,
	N_MCASE    = 55,
	N_TRYPROP  = 56,
	N_TRYUNW   = 57,
	N_MLET     = 58,
	N_MASSIGN  = 59,

	N_TYPETEST    = 60,
	N_TYPEASSERT  = 61,
	N_VOIDLIT     = 62,
	N_TBANG       = 63,
	N_YIELD       = 64,
	N_TENUM       = 65,
	N_TENUMMEMBER = 66,

	// N_TPARAM — chain wrapper for N_TTUPLE.list elements. Mirror of
	// cstage's Tparam (cmd/wcc/check.c:1437-1451) lifted to the AST so
	// `exprtype` can return shared element-type nodes (sym.decl.lhs,
	// struct field's .lhs, another N_TTUPLE's .list element) without
	// corrupting source ASTs by reusing their .next. .lhs holds the
	// element type AST (possibly shared); .next chains within the
	// parent N_TTUPLE.list. Cstage keeps Tparam at the Type-layer; ww
	// has no separate type layer for tuple chains, so the wrapper sits
	// at the AST layer. Other node fields are unused. Never appears
	// outside an N_TTUPLE.list; astprint unwraps transparently to keep
	// the 990 -a byte-diff against cstage.
	N_TPARAM      = 67,

	N_LAST     = 68,
};

// ---- Node -------------------------------------------------------------

type node = struct {
	kind:    nkind,
	file:    str,
	line:    i32,
	col:     i32,
	op:      tkind,	// for nkind.N_BIN / nkind.N_UN / nkind.N_ASSIGN
	str:     str,
	uval:    u64,
	fval:    f64,
	lhs:     *node,
	rhs:     *node,
	cond:    *node,
	body:    *node,
	els:     *node,
	list:    *node,
	next:    *node,
	attr:    *node,
	exported: i32,	// bool — `export` keyword present
	type_:   *void,	// filled in by checker; type.ww treats it as *tinfo
	tsuffix: str,	// typed numeric literal suffix ("i32", "u64", ...)
	nmod:    str,	// originating module from `// MODULE: foo`; "" if none
};

export fn newnode(k: nkind, file: str, line: i32, col: i32) *node = {
	// fval cast-init: 990's wwdump TK_FLOAT diff requires this file
	// to tokenise identically through C and ww (lex.ww:382 has the
	// same workaround for the cstage %g-formats vs ww-skips divergence).
	let n: *node = alloc(node{kind=k, file=file, line=line, col=col, op=tkind.TK_NONE, str="", uval=0u64, fval=0: f64, lhs=nil, rhs=nil, cond=nil, body=nil, els=nil, list=nil, next=nil, attr=nil, exported=0, type_=nil, tsuffix="", nmod=""})!;
	return n;
};

// ---- printer ----------------------------------------------------------

fn nkname(k: nkind) str = {
	if (k == nkind.N_NONE)      { return "none"; };
	if (k == nkind.N_INTLIT)    { return "int"; };
	if (k == nkind.N_FLOATLIT)  { return "float"; };
	if (k == nkind.N_STRLIT)    { return "str"; };
	if (k == nkind.N_RUNELIT)   { return "rune"; };
	if (k == nkind.N_TRUE)      { return "true"; };
	if (k == nkind.N_FALSE)     { return "false"; };
	if (k == nkind.N_NIL)       { return "nil"; };
	if (k == nkind.N_IDENT)     { return "id"; };
	if (k == nkind.N_BIN)       { return "bin"; };
	if (k == nkind.N_UN)        { return "un"; };
	if (k == nkind.N_CALL)      { return "call"; };
	if (k == nkind.N_INDEX)     { return "index"; };
	if (k == nkind.N_DOT)       { return "dot"; };
	if (k == nkind.N_CAST)      { return "cast"; };
	if (k == nkind.N_STRUCTLIT) { return "structlit"; };
	if (k == nkind.N_ARRLIT)    { return "arrlit"; };
	if (k == nkind.N_FIELD)     { return "field"; };
	if (k == nkind.N_ASSIGN)    { return "assign"; };
	if (k == nkind.N_ALLOC)     { return "alloc"; };
	if (k == nkind.N_FREE)      { return "free"; };
	if (k == nkind.N_RECV)      { return "recv"; };
	if (k == nkind.N_SLICE)     { return "slice"; };
	if (k == nkind.N_SPREAD)    { return "spread"; };
	if (k == nkind.N_BLOCK)     { return "block"; };
	if (k == nkind.N_EXPRSTMT)  { return "exprstmt"; };
	if (k == nkind.N_LET)       { return "let"; };
	if (k == nkind.N_RETURN)    { return "return"; };
	if (k == nkind.N_IF)        { return "if"; };
	if (k == nkind.N_FOR)       { return "for"; };
	if (k == nkind.N_FORRANGE)  { return "forrange"; };
	if (k == nkind.N_DEFER)     { return "defer"; };
	if (k == nkind.N_BREAK)     { return "break"; };
	if (k == nkind.N_CONTINUE)  { return "continue"; };
	if (k == nkind.N_SWITCH)    { return "switch"; };
	if (k == nkind.N_CASE)      { return "case"; };
	if (k == nkind.N_FILE)      { return "file"; };
	if (k == nkind.N_USE)       { return "use"; };
	if (k == nkind.N_DEF)       { return "def"; };
	if (k == nkind.N_TYPEDECL)  { return "typedecl"; };
	if (k == nkind.N_FNDECL)    { return "fn"; };
	if (k == nkind.N_PARAM)     { return "param"; };
	if (k == nkind.N_TNAME)     { return "tname"; };
	if (k == nkind.N_TPTR)      { return "tptr"; };
	if (k == nkind.N_TSLICE)    { return "tslice"; };
	if (k == nkind.N_TARRAY)    { return "tarray"; };
	if (k == nkind.N_TFN)       { return "tfn"; };
	if (k == nkind.N_TSTRUCT)   { return "tstruct"; };
	if (k == nkind.N_TFIELD)    { return "tfield"; };
	if (k == nkind.N_TCHAN)     { return "tchan"; };
	if (k == nkind.N_ATTR)      { return "attr"; };
	if (k == nkind.N_TTUPLE)    { return "ttuple"; };
	if (k == nkind.N_TTAGGED)   { return "ttagged"; };
	if (k == nkind.N_TUPLE)     { return "tuple"; };
	if (k == nkind.N_MATCH)     { return "match"; };
	if (k == nkind.N_MCASE)     { return "mcase"; };
	if (k == nkind.N_TRYPROP)   { return "tryprop"; };
	if (k == nkind.N_TRYUNW)    { return "tryunw"; };
	if (k == nkind.N_MLET)      { return "mlet"; };
	if (k == nkind.N_MASSIGN)   { return "massign"; };
	if (k == nkind.N_TYPETEST)   { return "typetest"; };
	if (k == nkind.N_TYPEASSERT) { return "typeassert"; };
	if (k == nkind.N_VOIDLIT)   { return "voidlit"; };
	if (k == nkind.N_TBANG)     { return "tbang"; };
	if (k == nkind.N_YIELD)     { return "yield"; };
	if (k == nkind.N_TENUM)       { return "tenum"; };
	if (k == nkind.N_TENUMMEMBER) { return "tenummember"; };
	if (k == nkind.N_TPARAM)    { return "tparam"; };
	if (k == nkind.N_LAST)      { return "last"; };
	return "?";
};

fn ind(fd: i32, d: i32) void = {
	let i: i32 = 0;
	for (i < d) {
		os.write(fd, "  ".ptr, 2u64);
		i += 1;
	};
};

fn putc1(fd: i32, b: u8) void = {
	let buf: [1]u8;
	buf[0] = b;
	os.write(fd, buf.ptr, 1u64);
};

fn putq(fd: i32, s: str) void = {
	putc1(fd, 34u8);	// '"'
	let i: i32 = 0;
	for (i < s.len) {
		let c: u8 = s[i];
		if (c == 34u8) {		// '"'
			os.write(fd, "\\\"".ptr, 2u64);
		} else { if (c == 92u8) {	// '\\'
			os.write(fd, "\\\\".ptr, 2u64);
		} else { if (c == 10u8) {	// '\n'
			os.write(fd, "\\n".ptr, 2u64);
		} else { if (c == 9u8) {	// '\t'
			os.write(fd, "\\t".ptr, 2u64);
		} else { if (c < 32u8) {
			let hi: u8 = c >> 4u8;
			let lo: u8 = c & 15u8;
			let h: u8 = 0u8;
			let l: u8 = 0u8;
			if (hi < 10u8) { h = hi + 48u8; } else { h = (hi - 10u8) + 97u8; };
			if (lo < 10u8) { l = lo + 48u8; } else { l = (lo - 10u8) + 97u8; };
			let buf: [4]u8;
			buf[0] = 92u8;
			buf[1] = 120u8;
			buf[2] = h;
			buf[3] = l;
			os.write(fd, buf.ptr, 4u64);
		} else {
			putc1(fd, c);
		};};};};};
		i += 1;
	};
	putc1(fd, 34u8);
};

fn pr(fd: i32, n: *node, d: i32) void = {
	if (n == nil) {
		ind(fd, d);
		os.write(fd, "()\n".ptr, 3u64);
		return;
	};
	// N_TPARAM wraps an N_TTUPLE.list element so exprtype can return
	// shared element-type nodes without corrupting their .next chain.
	// Cstage has no AST-level wrapper, so unwrap here to keep the 990
	// -a byte-diff with cstage's astprint.
	if (n.kind == nkind.N_TPARAM) {
		pr(fd, n.lhs, d);
		return;
	};
	ind(fd, d);
	putc1(fd, 40u8);	// '('
	let nm: str = nkname(n.kind);
	os.write(fd, nm.ptr, nm.len: u64);

	if (n.kind == nkind.N_INTLIT) {
		putc1(fd, 32u8);
		let s: str = strconv.u64tos(n.uval, strconv.base.DEC);
		os.write(fd, s.ptr, s.len: u64);
	} else { if (n.kind == nkind.N_RUNELIT) {
		putc1(fd, 32u8);
		let s: str = strconv.u64tos(n.uval, strconv.base.DEC);
		os.write(fd, s.ptr, s.len: u64);
	} else { if (
		n.kind == nkind.N_STRLIT  ||
		n.kind == nkind.N_IDENT   ||
		n.kind == nkind.N_USE     ||
		n.kind == nkind.N_DOT     ||
		n.kind == nkind.N_DEF     ||
		n.kind == nkind.N_TYPEDECL ||
		n.kind == nkind.N_FNDECL  ||
		n.kind == nkind.N_PARAM   ||
		n.kind == nkind.N_LET     ||
		n.kind == nkind.N_TNAME   ||
		n.kind == nkind.N_TFIELD  ||
		n.kind == nkind.N_TENUMMEMBER ||
		n.kind == nkind.N_FIELD   ||
		n.kind == nkind.N_ATTR
	) {
		// Match C ast.c: print the str field whenever it's non-nil,
		// even if its length is zero (e.g. an empty STRLIT prints
		// `(str ""`).
		let s: str = n.str;
		if (s.ptr != nil) {
			putc1(fd, 32u8);
			putq(fd, s);
		};
	} else { if (
		n.kind == nkind.N_BIN ||
		n.kind == nkind.N_UN  ||
		n.kind == nkind.N_ASSIGN
	) {
		putc1(fd, 32u8);
		let on: str = tokname(n.op);
		os.write(fd, on.ptr, on.len: u64);
	};};};};

	if (n.kind == nkind.N_FNDECL) {
		if (n.exported != 0) { os.write(fd, " export".ptr, 7u64); };
	};
	if (n.kind == nkind.N_DEF) {
		if (n.exported != 0) { os.write(fd, " export".ptr, 7u64); };
	};
	if (n.kind == nkind.N_TYPEDECL) {
		if (n.exported != 0) { os.write(fd, " export".ptr, 7u64); };
	};
	putc1(fd, 10u8);	// '\n'

	if (n.attr != nil) {
		ind(fd, d + 1);
		os.write(fd, "(@\n".ptr, 3u64);
		let m: *node = n.attr;
		for (m != nil) {
			pr(fd, m, d + 2);
			m = m.next;
		};
		ind(fd, d + 1);
		os.write(fd, ")\n".ptr, 2u64);
	};
	if (n.lhs  != nil) { pr(fd, n.lhs,  d + 1); };
	if (n.rhs  != nil) { pr(fd, n.rhs,  d + 1); };
	if (n.cond != nil) { pr(fd, n.cond, d + 1); };
	if (n.body != nil) { pr(fd, n.body, d + 1); };
	if (n.els  != nil) { pr(fd, n.els,  d + 1); };
	if (n.list != nil) {
		ind(fd, d + 1);
		os.write(fd, "(list\n".ptr, 6u64);
		let m: *node = n.list;
		for (m != nil) {
			pr(fd, m, d + 2);
			m = m.next;
		};
		ind(fd, d + 1);
		os.write(fd, ")\n".ptr, 2u64);
	};
	ind(fd, d);
	os.write(fd, ")\n".ptr, 2u64);
};

export fn astprint(fd: i32, n: *node) void = {
	pr(fd, n, 0);
};