ww/cmd/wcc/ww.h

/*
 * ww.h — central header for libwcc.a (the ww frontend library).
 *
 * Plan 9 in spirit. This file mirrors cc/cc.h's role: one shared
 * header that declares everything every translation unit in the
 * frontend cares about.
 *
 * Phases add to this file (lexer/parser/checker), they do not branch
 * a sibling header. There is one frontend; there is one ww.h.
 */
#ifndef WW_H
#define WW_H

#include <stddef.h>
#include <stdint.h>
#include <stdarg.h>
#include <stdio.h>

/* version banner — printed by `ww -V` */
#define WW_VERSION	"0.0"

/* short integer aliases, Plan 9 / Hare-flavoured */
typedef int8_t		i8;
typedef int16_t		i16;
typedef int32_t		i32;
typedef int64_t		i64;
typedef uint8_t		u8;
typedef uint16_t	u16;
typedef uint32_t	u32;
typedef uint64_t	u64;

/* forward decls — concrete shapes appear in their phases. */
typedef struct Tok	Tok;
typedef struct Lex	Lex;
typedef struct Node	Node;
typedef struct Sym	Sym;
typedef struct Type	Type;
typedef struct Scope	Scope;
typedef struct Arena	Arena;

/* mem.c — bump arena (no free; reset/destroy at end of phase) */
struct Arena {
	u8	*buf;	/* base of current chunk */
	u64	off;	/* bytes used in current chunk */
	u64	cap;	/* capacity of current chunk */
	struct Arena *next;	/* older chunks (linked list, head = current) */
	u64	total;	/* across all chunks, debug only */
};

Arena	*newarena(void);
void	*amalloc(Arena*, u64);		/* zeroed, aligned to 16 */
char	*astrdup(Arena*, const char*);
char	*astrndup(Arena*, const char*, u64);
char	*aprintf(Arena*, const char*, ...);
void	freearena(Arena*);

/* err.c — diagnostics. Phase 0 has only fatal/warn; later phases add
 * source-location-bearing variants. */
typedef struct Pos	Pos;
struct Pos {
	const char *file;
	i32	line;
	i32	col;
};

extern Pos	noPos;
extern int	nerrors;
extern int	nwarnings;
extern FILE	*errout;

void	fatal(const char*, ...) __attribute__((noreturn, format(printf, 1, 2)));
void	errorf(Pos, const char*, ...) __attribute__((format(printf, 2, 3)));
void	warnf(Pos, const char*, ...) __attribute__((format(printf, 2, 3)));

/* tiny helpers */
#define nelem(a)	((sizeof(a) / sizeof((a)[0])))

/* ---- lexer (lex.c, tok.c) ----------------------------------------- */
typedef enum {
	/* zero is "no token" so memset-zero structs read sane */
	TK_NONE = 0,

	/* trivial */
	TK_EOF,
	TK_ERR,
	TK_IDENT,
	TK_INT,
	TK_FLOAT,
	TK_RUNE,
	TK_STR,

	/* keywords — stay grouped, used by tok.c kwtab */
	TK_FN,
	TK_LET,
	TK_DEF,
	TK_IF,
	TK_ELSE,
	TK_FOR,
	TK_SWITCH,
	TK_CASE,
	TK_RETURN,
	TK_USE,
	TK_TYPE,
	TK_STRUCT,
	TK_DEFER,
	TK_BREAK,
	TK_CONTINUE,
	TK_EXPORT,
	TK_PROC,
	TK_CHAN,
	TK_NIL,
	TK_TRUE,
	TK_FALSE,
	TK_AS,		/* Hare-style type assertion: e as T */
	TK_STATIC,	/* Hare-style storage-class qualifier */
	TK_MATCH,	/* match expression head */
	TK_CONST,	/* const binding */
	TK_UNDER,	/* bare '_' discard */

	/* punct + operators */
	TK_LPAREN,	/* ( */
	TK_RPAREN,	/* ) */
	TK_LBRACE,	/* { */
	TK_RBRACE,	/* } */
	TK_LBRACK,	/* [ */
	TK_RBRACK,	/* ] */
	TK_COMMA,	/* , */
	TK_SEMI,	/* ; */
	TK_COLON,	/* : */
	TK_DOT,		/* . */
	TK_ELLIPSIS,	/* ... */
	TK_DOTDOT,	/* .. (range op) */
	TK_AT,		/* @ */
	TK_QUESTION,	/* ? */

	TK_ASSIGN,	/* = */
	TK_PLUSEQ,	/* += */
	TK_MINUSEQ,	/* -= */
	TK_STAREQ,	/* *= */
	TK_SLASHEQ,	/* /= */
	TK_PERCENTEQ,	/* %= */
	TK_AMPEQ,	/* &= */
	TK_PIPEEQ,	/* |= */
	TK_CARETEQ,	/* ^= */
	TK_LSHIFTEQ,	/* <<= */
	TK_RSHIFTEQ,	/* >>= */

	TK_PLUS,	/* + */
	TK_MINUS,	/* - */
	TK_STAR,	/* * */
	TK_SLASH,	/* / */
	TK_PERCENT,	/* % */
	TK_AMP,		/* & */
	TK_PIPE,	/* | */
	TK_CARET,	/* ^ */
	TK_TILDE,	/* ~ */
	TK_LSHIFT,	/* << */
	TK_RSHIFT,	/* >> */

	TK_EQ,		/* == */
	TK_NEQ,		/* != */
	TK_LT,		/* < */
	TK_LE,		/* <= */
	TK_GT,		/* > */
	TK_GE,		/* >= */

	TK_AND,		/* && */
	TK_OR,		/* || */
	TK_NOT,		/* ! */

	TK_LARROW,	/* <- (chan recv) */
	TK_ARROW,	/* -> (reserved) */
	TK_FATARROW,	/* => (match arms) */

	/* Appended after TK_FATARROW (not grouped with the keyword block)
	 * to keep the numeric value of every existing kind unchanged —
	 * the selfhost wwdump-diff test (990) is byte-sensitive. */
	TK_IS,		/* Hare-style type test: e is T */
	TK_VOID,	/* `void` — both a type name and a zero-size value */
	TK_YIELD,	/* `yield expr;` — value-return from a match arm */
	TK_ENUM,	/* Hare-style `enum [storage] { ... }` type form */
	TK_MODULE,	/* `module foo;` — directory-as-module declaration */

	TK_LAST		/* sentinel for tables */
} Tkind;

struct Tok {
	Tkind	kind;
	Pos	pos;
	const char *text;	/* lexeme (arena-owned, NUL-terminated) */
	u64	tlen;		/* byte length of lexeme (sans NUL) */
	/* numeric values pre-parsed; string/rune unescaped */
	union {
		u64	uval;	/* TK_INT, TK_RUNE */
		double	fval;	/* TK_FLOAT */
	} v;
	/* for typed numeric literals: "i32", "u8", "f64", ... or NULL. */
	const char *tsuffix;
};

struct Lex {
	const char *file;
	const char *src;	/* full source, NUL-terminated */
	u64	srclen;
	u64	pos;		/* current byte offset */
	i32	line;
	i32	col;
	Arena	*a;		/* token-text arena */
	int	errs;
};

void	lexinit(Lex*, Arena*, const char *file, const char *src, u64 len);
Tok	lexnext(Lex*);
const char *tokname(Tkind);	/* canonical spelling, e.g. "fn", "+=" */
void	tokprint(FILE*, Tok);	/* one line, "%s:%d:%d: %s %q" */
Tkind	kwlookup(const char *s, u64 n);	/* TK_NONE if not a keyword */

/* ---- AST (ast.c, parse.c) ----------------------------------------- */
typedef enum {
	N_NONE = 0,

	/* literals */
	N_INTLIT,
	N_FLOATLIT,
	N_STRLIT,
	N_RUNELIT,
	N_TRUE,
	N_FALSE,
	N_NIL,
	N_IDENT,

	/* expressions */
	N_BIN,		/* op, lhs, rhs */
	N_UN,		/* op, lhs */
	N_CALL,		/* lhs=callee, list=args */
	N_INDEX,	/* lhs=base, rhs=index */
	N_DOT,		/* lhs=base, str=field */
	N_CAST,		/* lhs=expr, rhs=type-expr */
	N_STRUCTLIT,	/* lhs=type-expr, list=N_FIELD */
	N_ARRLIT,	/* list=elements (for [a,b,...]) */
	N_FIELD,	/* str=name, lhs=value */
	N_ASSIGN,	/* op, lhs, rhs */
	N_ALLOC,	/* lhs=expr, rhs=size-or-null */
	N_FREE,		/* lhs=expr */
	N_RECV,		/* lhs (chan recv: <-c) */
	N_SLICE,	/* lhs=base, rhs=lo or NULL, cond=hi or NULL */
	N_SPREAD,	/* lhs (variadic spread in arg position: e...) */

	/* statements */
	N_BLOCK,	/* list=stmts */
	N_EXPRSTMT,	/* lhs=expr */
	N_LET,		/* str=name, lhs=type-expr|NULL, rhs=init|NULL */
	N_RETURN,	/* lhs=expr|NULL */
	N_IF,		/* cond, body, els */
	N_FOR,		/* lhs=init, cond, rhs=post, body */
	N_FORRANGE,	/* str=elem name, lhs=slice expr, body=block */
	N_DEFER,	/* lhs=expr */
	N_BREAK,
	N_CONTINUE,
	N_SWITCH,	/* lhs=scrutinee, list=cases */
	N_CASE,		/* list=exprs (empty=default), body */

	/* declarations */
	N_FILE,		/* list=top decls */
	N_USE,		/* str=path */
	N_DEF,		/* str=name, lhs=type|NULL, rhs=init */
	N_TYPEDECL,	/* str=name, lhs=type-expr */
	N_FNDECL,	/* str=name, list=params, lhs=ret-type, body|NULL */
	N_PARAM,	/* str=name, lhs=type-expr */

	/* type expressions */
	N_TNAME,	/* str */
	N_TPTR,		/* lhs=inner */
	N_TSLICE,	/* lhs=inner */
	N_TARRAY,	/* lhs=element, rhs=len-expr */
	N_TFN,		/* list=params, lhs=ret */
	N_TSTRUCT,	/* list=fields */
	N_TFIELD,	/* str=name, lhs=type */
	N_TCHAN,	/* lhs=element */

	/* attribute on a decl */
	N_ATTR,		/* str=name, list=args */

	/* multi-value (tuple) plumbing */
	N_TTUPLE,	/* type expr: (T1, T2, ...). list = element type exprs */
	N_TTAGGED,	/* type expr: (T1 | T2 | ...). list = variant type exprs */
	N_TUPLE,	/* expr: (e1, e2, ...). list = element exprs */
	N_MATCH,	/* match (lhs) { list of cases }; cases are N_MCASE */
	N_MCASE,	/* str=binding name (or NULL), lhs=variant type expr or NULL, body */
	N_TRYPROP,	/* lhs? — propagate error variant */
	N_TRYUNW,	/* lhs! — abort on error variant */
	N_MLET,		/* let a, b = expr;  list = N_LET stubs (str, lhs=type), rhs = expr */
	N_MASSIGN,	/* a, b = expr;  list = lvalue exprs, rhs = expr */

	/* Appended after N_MASSIGN — keeps the numeric value of every
	 * existing kind unchanged, so the selfhost AST dump still diffs
	 * byte-for-byte. lhs=value, rhs=variant type expr. */
	N_TYPETEST,	/* lhs is T  →  bool */
	N_TYPEASSERT,	/* lhs as T  →  T (abort if tag mismatch) */
	N_VOIDLIT,	/* `void` as expression — zero-size void value */
	N_TBANG,	/* `!T` — error-flagged type. lhs = inner type. */
	N_YIELD,	/* `yield expr;` — set the enclosing match's value-
			 * return and jump to its end label. lhs = value. */
	N_TENUM,	/* `enum [storage] { ... }` type form.
			 * lhs = storage type expr or NULL (default i32);
			 * list = chain of N_TENUMMEMBER. */
	N_TENUMMEMBER,	/* enum member. str=name, lhs=value expr or NULL
			 * (auto-increment when omitted). */

	N_LAST
} Nkind;

struct Node {
	Nkind		kind;
	Pos		pos;
	Tkind		op;	/* for N_BIN/N_UN/N_ASSIGN */
	const char	*str;	/* identifier/literal/name/path */
	u64		strlen;
	u64		uval;	/* int/rune lit */
	double		fval;	/* float lit */
	Node		*lhs;
	Node		*rhs;
	Node		*cond;
	Node		*body;
	Node		*els;
	Node		*list;	/* head of singly-linked sibling chain */
	Node		*next;	/* sibling link inside `list` */
	Node		*attr;	/* @attribute chain (N_ATTR list) */
	int		export;
	Type		*type;	/* filled in by checker */
	const char	*tsuffix;	/* typed numeric literal suffix */
	const char	*module;	/* `// MODULE: foo` directive at the
					 * decl's section in combined.ww.
					 * NULL for nested nodes; only top-
					 * level decls (fn/def/type/let)
					 * carry it. */
};

Node	*newnode(Arena*, Nkind, Pos);
void	astprint(FILE*, Node*);	/* s-expr, deterministic, one-line per node */

typedef struct Parser Parser;
struct Parser {
	Lex	*l;
	Arena	*a;
	Tok	cur;
	Tok	la;	/* one-token lookahead buffer */
	int	hasla;
	int	errs;
	int	nocast;	/* in case-selector ctx, ':' is a separator */
	const char *curmod;	/* most-recent `module foo;` declaration —
				 * stamped onto each top-level decl that
				 * follows. */
};

void	parserinit(Parser*, Arena*, Lex*);
Node	*parsefile(Parser*);
Node	*parseexpr_top(Parser*);	/* for testing: parse one expression */

/* ---- types (type.c) ----------------------------------------------- */
typedef enum {
	TY_NONE = 0,
	TY_VOID,
	TY_BOOL,
	TY_RUNE,
	TY_I8, TY_I16, TY_I32, TY_I64,
	TY_U8, TY_U16, TY_U32, TY_U64,
	TY_UINT, TY_INT,
	TY_UINTPTR,
	TY_F32, TY_F64,
	TY_STR,
	TY_PTR,
	TY_SLICE,
	TY_ARRAY,
	TY_STRUCT,
	TY_FN,
	TY_CHAN,
	TY_NAMED,
	TY_TUPLE,
	TY_TAGGED,	/* (T1 | T2 | ...) — Hare-style sum type */
	TY_ERR,
	TY_NEVER,	/* bottom: assignable to anything; size 0 */
	/* untyped constants (not surfaced to users; checker-internal) */
	TY_UNTYPED_INT,
	TY_UNTYPED_FLOAT,
	TY_UNTYPED_STR,
	TY_UNTYPED_RUNE,
	TY_UNTYPED_BOOL,
	TY_UNTYPED_NIL,
	/* Appended at the tail to keep existing TY_* values stable —
	 * lib/ww/typ.ww mirrors them as explicit `def` numbers. */
	TY_ENUM		/* `enum [storage] { ... }`. sub=storage,
			 * fields=member list (Tfield.offset = u64 value). */
} TypeKind;

typedef struct Tfield Tfield;
struct Tfield {
	const char *name;
	Type	*type;
	u64	offset;
	Tfield	*next;
};

typedef struct Tparam Tparam;
struct Tparam {
	const char *name;
	Type	*type;
	Tparam	*next;
	int	variadic;	/* Hare-style `T...` — `type` is []T,
				 * call site gathers / forwards. Distinct
				 * from Type.variadic (C-style FFI `...`). */
};

struct Type {
	TypeKind	kind;
	u64		size;
	u64		align;
	Type		*sub;	/* ptr/slice/array/chan element */
	u64		alen;	/* array length */
	Tfield		*fields;/* struct */
	Tparam		*params;/* fn */
	Type		*ret;	/* fn */
	int		variadic;
	const char	*name;	/* named alias / debug */
	Type		*under;	/* underlying resolved type for NAMED */
	int		iserror;/* Hare-style `!T` error mark; propagates
				 * through NAMED aliases. Variants with
				 * iserror=1 are the propagation target of
				 * the `?` operator. */
	int		nullable;/* TY_TAGGED with exactly `(*T | void)` —
				 * stored as a single 8-byte pointer; null
				 * is the void variant. Mirrors Hare's
				 * `(*T | null)` folding. */
};

extern Type	*ty_void, *ty_bool, *ty_rune;
extern Type	*ty_i8, *ty_i16, *ty_i32, *ty_i64;
extern Type	*ty_u8, *ty_u16, *ty_u32, *ty_u64;
extern Type	*ty_int, *ty_uint, *ty_uintptr;
extern Type	*ty_f32, *ty_f64, *ty_str;
extern Type	*ty_err;
extern Type	*ty_never;
extern Type	*ty_untyped_int, *ty_untyped_float, *ty_untyped_str;
extern Type	*ty_untyped_rune, *ty_untyped_bool, *ty_untyped_nil;

void	typesinit(Arena*);
Type	*newtype(Arena*, TypeKind);
Type	*type_ptr(Arena*, Type *sub);
Type	*type_slice(Arena*, Type *sub);
Type	*type_array(Arena*, Type *sub, u64 len);
Type	*type_chan(Arena*, Type *sub);
Type	*type_named(Arena*, const char *name, Type *under);
const char *type_name(Arena*, Type*);	/* arena'd debug string */
int	type_eq(Type *a, Type *b);		/* structural equality */
int	type_isint(Type *t);
int	type_isfloat(Type *t);
int	type_isnum(Type *t);
int	type_isunsigned(Type *t);
int	type_isuntyped(Type *t);
int	type_assignable(Type *dst, Type *src);
Type	*type_default(Type *t);		/* untyped → default concrete */

/* ---- symbols (sym.c) ---------------------------------------------- */
typedef enum {
	SK_NONE = 0,
	SK_VAR,
	SK_PARAM,
	SK_DEF,
	SK_TYPE,
	SK_FN,
	SK_USE,
	SK_FIELD	/* not stored in scope; used by check internally */
} Skind;

struct Sym {
	const char	*name;
	Skind		kind;
	Type		*type;
	Node		*decl;
	int		exported;
	int		is_const;	/* const-bound (assignment rejected) */
	int		use_alias;	/* also bound as a `use` module name. Set
					 * when a `use foo;` directive collides
					 * with a same-named SK_TYPE/SK_FN/etc.
					 * Lets resolve_typename treat `foo.x`
					 * as module-qualified even though the
					 * primary kind isn't SK_USE. */
	const char	*mod;		/* importing module's bareword for
					 * symbols originating in a `use`-
					 * imported module. NULL for primary
					 * (root) compilation unit symbols.
					 * Used by scope_lookup_in_module to
					 * disambiguate same-leaf-name types
					 * coming from different imports. */
	Sym		*next;		/* iteration */
	Sym		*hashnext;	/* bucket chain */
	Scope		*scope;
};

struct Scope {
	Scope	*parent;
	Sym	*first, *last;
	Sym	**buckets;
	u64	nbuckets;
	Arena	*a;
};

Scope	*newscope(Arena*, Scope *parent);
Sym	*scope_define(Scope*, const char *name, Skind, Type*, Node *decl);
Sym	*scope_define_in_module(Scope*, const char *name, const char *mod,
		Skind, Type*, Node *decl);
Sym	*scope_lookup(Scope*, const char *name);	/* walk up parents */
Sym	*scope_lookup_local(Scope*, const char *name);
Sym	*scope_lookup_in_module(Scope*, const char *mod, const char *name);
Sym	*scope_lookup_prefer(Scope*, const char *mod, const char *name);

/* ---- checker (check.c) -------------------------------------------- */
typedef struct Checker Checker;
struct Checker {
	Arena	*a;
	Scope	*top;	/* file scope */
	Scope	*cur;	/* current scope */
	Type	*ret;	/* expected return type of current fn (or NULL) */
	const char *cur_mod;	/* importing-module bareword for the decl
				 * currently being checked; NULL for primary
				 * compilation unit. Drives same-module
				 * preference in bare-leaf lookups so a bare
				 * `read` inside lib/os resolves to os.read
				 * rather than colliding io.read. */
	Node	*file;	/* current N_FILE root; used by check_module_shadow
			 * to consult the declaring source file's own `use`
			 * directives when refusing param/let names that
			 * would shadow an imported module bareword. */
	int	loops;	/* nesting count for break/continue */
	int	errs;
};

void	check_init(Checker*, Arena*);
void	check_file(Checker*, Node *file);

/* fold_int_literal — fold an integer-literal-leaf expression to its
 * u64 value. Accepts int/rune literal, true/false/nil, and a unary
 * +/-/~ over the same (any depth). No sibling-ident, no binary op.
 * Returns 1 on success; the call site decides what a miss means
 * (eval_enum_value's leaf delegation, emit_defs's DATA-row gate). */
int	fold_int_literal(Node*, u64*);

#endif /* WW_H */