ww/cmd/w6c/wwi.c

/*
 * wwi.c — `.wwi` export-data producer (w6c -I): a re-parseable ww-prototype
 * rendering of a package's EXPORTED surface. Since the sep-compile flip
 * (epic #22) this is the LIVE import path — the driver runs one `w6c -c -I`
 * per package and feeds each dep's `.wwi` to its importers as import scope;
 * the combined.ww amalgamator is gone.
 *
 * Specs: .ai/rob-M2-spec.md (producer) + .ai/drew-M2-checkexported.md
 * (check_exported_type). Two load-bearing choices follow them:
 *
 *   - Unparse walks the AST type-expr subtree (the N_T* nodes), NOT the
 *     checked Type* — tinfo collapses nominal pkg.Name identity (MEMORY
 *     tinfo_lossy_nominal). The AST preserves surface syntax + names.
 *   - check_exported_type rides the producer entry (flag-gated), so M2
 *     stays dead on the normal `.s` path. It rejects exactly one thing:
 *     an exported signature naming a non-exported nominal type.
 *
 * A `.wwi` is ONE package's interface. The M2 gate feeds w6c a driver-
 * combined unit (imports concatenated ahead of the target), so the emit
 * filters to PRIMARY decls (imported == 0) — in the real per-package
 * compile every decl is primary, so the filter is a no-op there. Output
 * is a pure function of the exported API: package line, byte-sorted
 * imports, byte-sorted decls (no map/hash iteration order).
 */
#include "gc.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

static int
wwi_primary(Node *n)
{
	/* imported == 1 marks a decl reached through a `//ww:module <path>`
	 * boundary (an imported module's concatenated section). */
	return n && n->imported == 0;
}

/* check_exported_type (drew): resolve an N_TNAME to its type sym WITHOUT
 * the side effects of resolve_typename (no on-demand resolve, no double
 * "unknown type" error). A primitive/keyword resolves to no SK_TYPE →
 * leaf. By the time the producer runs, check_file has finished and
 * c->cur == c->top.
 */
static Sym *
wwi_typesym(Checker *c, const char *nm)
{
	if (nm == NULL) return NULL;
	Sym *s = scope_lookup_type(c->cur, NULL, nm);
	if (s == NULL) {
		const char *dot = strrchr(nm, '.');
		if (dot)
			s = scope_lookup_type(c->cur, NULL, dot + 1);
	}
	if (s && s->kind == SK_TYPE)
		return s;
	return NULL;
}

static int
wwi_check_type(Checker *c, Pos loc, Node *t)
{
	int bad = 0;
	if (t == NULL)
		return 0;
	switch (t->kind) {
	case N_TNAME: {
		Sym *s = wwi_typesym(c, t->str);
		/* Sym.exported is vestigial (the checker never sets it); the
		 * nominal's export status lives on its decl node, parser-set.
		 * No file-guard twin to wwstage's: cstage resolves `nomem` and
		 * the primitives through lookup_builtin (no scope SK_TYPE), so
		 * scope_lookup_type returns NULL for them and they never reach
		 * this reject — harec's STORAGE_NOMEM leaf-arm, by construction. */
		if (s && s->decl && s->decl->kind == N_TYPEDECL
		    && s->decl->export == 0) {
			errorf(loc, "exported declaration references "
			    "unexported type '%s'", t->str);
			bad = 1;
		}
		break;
	}
	case N_TPTR:
	case N_TSLICE:
	case N_TBANG:
	case N_TCHAN:
		bad |= wwi_check_type(c, loc, t->lhs);
		break;
	case N_TARRAY:
		/* element only; the length is a const-expr, not a type. */
		bad |= wwi_check_type(c, loc, t->lhs);
		break;
	case N_TFN:
		for (Node *p = t->list; p; p = p->next)
			bad |= wwi_check_type(c, loc, p->lhs);
		bad |= wwi_check_type(c, loc, t->lhs);
		break;
	case N_TSTRUCT:
		for (Node *f = t->list; f; f = f->next)
			bad |= wwi_check_type(c, loc, f->lhs);
		break;
	case N_TTAGGED:
	case N_TTUPLE:
		for (Node *e = t->list; e; e = e->next)
			bad |= wwi_check_type(c, loc, e);
		break;
	case N_TENUM:
		/* the inline enum body is the definition, not a reference;
		 * recurse only its storage type (members are values). */
		bad |= wwi_check_type(c, loc, t->lhs);
		break;
	default:
		/* a non-type node in type position should not occur — leaf. */
		break;
	}
	return bad;
}

static int
wwi_check_decl(Checker *c, Node *d)
{
	int bad = 0;
	switch (d->kind) {
	case N_FNDECL:
		for (Node *p = d->list; p; p = p->next)
			bad |= wwi_check_type(c, d->pos, p->lhs);
		bad |= wwi_check_type(c, d->pos, d->lhs);	/* ret */
		break;
	case N_TYPEDECL:
		bad |= wwi_check_type(c, d->pos, d->lhs);
		break;
	case N_DEF:
		/* the declared type; the rhs const value is not a type. */
		bad |= wwi_check_type(c, d->pos, d->lhs);
		break;
	case N_LET:
		bad |= wwi_check_type(c, d->pos, d->lhs);
		break;
	default:
		break;
	}
	return bad;
}

/* type-expr unparse per rob §2.2; const-expr per rob §2.4. */

static void wwi_expr(FILE *of, Node *e);
static void wwi_type(FILE *of, Node *t);

static void
wwi_quote(FILE *of, const char *s, u64 n)
{
	fputc('"', of);
	for (u64 i = 0; i < n; i++) {
		unsigned char c = (unsigned char)s[i];
		switch (c) {
		case '"':  fputs("\\\"", of); break;
		case '\\': fputs("\\\\", of); break;
		case '\n': fputs("\\n", of); break;
		case '\t': fputs("\\t", of); break;
		default:
			if (c < 0x20)	fprintf(of, "\\x%02x", c);
			else		fputc(c, of);
		}
	}
	fputc('"', of);
}

static void
wwi_rune(FILE *of, u64 cp)
{
	/* #50: the lexer now reads \u/\U, so wide codepoints round-trip as
	 * those escapes (write-twin of lex.c lexunicode). Codepoints <=0xff
	 * keep the \xHH spelling they already round-tripped as. */
	if (cp > 0xffff) {
		fprintf(of, "'\\U%08x'", (unsigned)cp);
		return;
	}
	if (cp > 0xff) {
		fprintf(of, "'\\u%04x'", (unsigned)cp);
		return;
	}
	unsigned char c = (unsigned char)cp;
	fputc('\'', of);
	switch (c) {
	case '\'': fputs("\\'", of); break;
	case '\\': fputs("\\\\", of); break;
	case '\n': fputs("\\n", of); break;
	case '\t': fputs("\\t", of); break;
	default:
		if (c < 0x20 || c >= 0x7f)	fprintf(of, "\\x%02x", c);
		else				fputc(c, of);
	}
	fputc('\'', of);
}

static void
wwi_param(FILE *of, Node *p)
{
	if (p->str && strcmp(p->str, "...") == 0) {	/* C-style FFI `...` */
		fputs("...", of);
		return;
	}
	if (p->str && p->str[0]) {
		fputs(p->str, of);
		fputs(": ", of);
	}
	wwi_type(of, p->lhs);
	if (p->op == TK_ELLIPSIS)			/* Hare `T...` variadic */
		fputs("...", of);
}

static void
wwi_type(FILE *of, Node *t)
{
	if (t == NULL) {	/* absent return type spells void */
		fputs("void", of);
		return;
	}
	switch (t->kind) {
	case N_TNAME:
		fputs(t->str ? t->str : "void", of);
		break;
	case N_TPTR:
		fputc('*', of);
		wwi_type(of, t->lhs);
		break;
	case N_TSLICE:
		fputs("[]", of);
		wwi_type(of, t->lhs);
		break;
	case N_TARRAY:
		fputc('[', of);
		if (t->rhs)	wwi_expr(of, t->rhs);
		else		fputc('_', of);
		fputc(']', of);
		wwi_type(of, t->lhs);
		break;
	case N_TBANG:
		fputc('!', of);
		wwi_type(of, t->lhs);
		break;
	case N_TCHAN:
		fputs("chan ", of);
		wwi_type(of, t->lhs);
		break;
	case N_TFN:
		fputs("fn(", of);
		for (Node *p = t->list; p; p = p->next) {
			if (p != t->list) fputs(", ", of);
			wwi_param(of, p);
		}
		fputs(") ", of);
		wwi_type(of, t->lhs);
		break;
	case N_TSTRUCT:
		/* re-emit the `@packed` attr so the flag round-trips
		 * through sep-compile (harec unparse/type.ha:122-126). */
		fputs(t->packed ? "struct @packed { " : "struct { ", of);
		for (Node *f = t->list; f; f = f->next) {
			if (f != t->list) fputs(", ", of);
			if (f->str && f->str[0]) {
				fputs(f->str, of);
				fputs(": ", of);
			}
			wwi_type(of, f->lhs);
		}
		fputs(" }", of);
		break;
	case N_TTUPLE:
		fputc('(', of);
		for (Node *e = t->list; e; e = e->next) {
			if (e != t->list) fputs(", ", of);
			wwi_type(of, e);
		}
		fputc(')', of);
		break;
	case N_TTAGGED:
		fputc('(', of);
		for (Node *e = t->list; e; e = e->next) {
			if (e != t->list) fputs(" | ", of);
			/* #95: re-emit the spread marker purely syntactically;
			 * the consumer's type-store flattens (ref/hare/hare/
			 * unparse/type.ha:290-300). */
			if (e->op == TK_ELLIPSIS)
				fputs("...", of);
			wwi_type(of, e);
		}
		fputc(')', of);
		break;
	case N_TENUM:
		fputs("enum ", of);
		if (t->lhs) {
			wwi_type(of, t->lhs);
			fputc(' ', of);
		}
		fputs("{ ", of);
		for (Node *m = t->list; m; m = m->next) {
			if (m != t->list) fputs(", ", of);
			fputs(m->str, of);
			if (m->lhs) {
				fputs(" = ", of);
				wwi_expr(of, m->lhs);
			}
		}
		fputs(" }", of);
		break;
	default:
		fatal("wwi: unhandled type node kind %d", t->kind);
	}
}

static void
wwi_expr(FILE *of, Node *e)
{
	if (e == NULL)
		return;
	switch (e->kind) {
	case N_INTLIT:
		fprintf(of, "%llu", (unsigned long long)e->uval);
		if (e->tsuffix) fputs(e->tsuffix, of);
		break;
	case N_IDENT:
	case N_TNAME:
		fputs(e->str, of);
		break;
	case N_DOT:
		wwi_expr(of, e->lhs);
		fputc('.', of);
		fputs(e->str, of);
		break;
	case N_TRUE:	fputs("true", of); break;
	case N_FALSE:	fputs("false", of); break;
	case N_NIL:	fputs("nil", of); break;
	case N_STRLIT:	wwi_quote(of, e->str, e->strlen); break;
	/* A bare rune literal in a const-expr (types::RUNE_MIN = '\0'); casts
	 * never reach here — the checker folds a const cast to an integer
	 * literal before the producer runs (RUNE_MAX `0x10ffff: rune` emits
	 * as the plain int 1114111). */
	case N_RUNELIT:	wwi_rune(of, e->uval); break;
	case N_BIN:
		wwi_expr(of, e->lhs);
		fprintf(of, " %s ", tokname(e->op));
		wwi_expr(of, e->rhs);
		break;
	case N_UN:
		fputs(tokname(e->op), of);
		wwi_expr(of, e->lhs);
		break;
	default:
		fatal("wwi: unhandled const-expr node kind %d", e->kind);
	}
}

/* Only codegen/link-relevant attributes round-trip into the `.wwi`. Today
 * that is exactly @symbol (the FFI link-symbol override, read back at cgen
 * fficollect — dropping it makes sep-compile emit `CALL malloc` for a
 * `@symbol("rt_malloc")` fn). Named for the class so @align/@offset would
 * slot in here IF ww ever grows field-layout attributes — it has none
 * today (task #47 report). @test never reaches a `.wwi` (test fns are not
 * export-marked), so it needs no exclusion arm. */
static int
wwi_attr_relevant(const char *nm)
{
	return nm && strcmp(nm, "symbol") == 0;
}

static void
wwi_attrs(FILE *of, Node *d)
{
	for (Node *a = d->attr; a; a = a->next) {
		if (a->kind != N_ATTR || !wwi_attr_relevant(a->str))
			continue;
		fputc('@', of);
		fputs(a->str, of);
		if (a->list) {
			fputc('(', of);
			for (Node *arg = a->list; arg; arg = arg->next) {
				if (arg != a->list) fputs(", ", of);
				wwi_expr(of, arg);
			}
			fputc(')', of);
		}
		fputc(' ', of);
	}
}

static void
wwi_decl(FILE *of, Node *d)
{
	switch (d->kind) {
	case N_FNDECL:
		wwi_attrs(of, d);
		fputs("export fn ", of);
		fputs(d->str, of);
		fputc('(', of);
		for (Node *p = d->list; p; p = p->next) {
			if (p != d->list) fputs(", ", of);
			wwi_param(of, p);
		}
		fputs(") ", of);
		wwi_type(of, d->lhs);
		fputs(";\n", of);
		break;
	case N_TYPEDECL:
		fputs("export type ", of);
		fputs(d->str, of);
		fputs(" = ", of);
		wwi_type(of, d->lhs);
		fputs(";\n", of);
		break;
	case N_DEF:
		fputs("export def ", of);
		fputs(d->str, of);
		fputs(": ", of);
		wwi_type(of, d->lhs);
		/* An aggregate initializer (N_STRUCTLIT/N_ARRLIT) is a DATA-
		 * global (#52): emit a value-LESS prototype `export def X: T;`.
		 * The defining package's own .o emits the struct/array DATA; the
		 * importer registers the def by TYPE only (the DATA-suppression
		 * if(sep_mode && imported)continue at cgen.c is already in place)
		 * and field/element reads become external LEAQ refs the linker
		 * fills. Boundary (rule 7): this gives up importer-side const-
		 * fold of an aggregate def's FIELDS — a no-op, ww never folds
		 * struct-literal field access, and an aggregate-def field
		 * demanded in a const-fold context stays a LOUD error (task #71). */
		if (d->rhs && (d->rhs->kind == N_STRUCTLIT
		    || d->rhs->kind == N_ARRLIT)) {
			fputs(";\n", of);
		} else {
			fputs(" = ", of);
			wwi_expr(of, d->rhs);
			fputs(";\n", of);
		}
		break;
	case N_LET:
		fputs("export let ", of);
		fputs(d->str, of);
		fputs(": ", of);
		if (d->lhs == NULL)
			fatal("wwi: exported let '%s' has no declared type",
			    d->str);
		wwi_type(of, d->lhs);
		fputs(";\n", of);
		break;
	default:
		break;
	}
}

/* deterministic ordering per rob §3. */

struct declent { Node *d; int idx; };
struct useent  { const char *path; int idx; };

static int
declcmp(const void *a, const void *b)
{
	const struct declent *x = a, *y = b;
	int r = strcmp(x->d->str, y->d->str);	/* symbol name ONLY */
	if (r != 0) return r;
	return x->idx - y->idx;			/* total order, stable */
}

static int
usecmp(const void *a, const void *b)
{
	const struct useent *x = a, *y = b;
	int r = strcmp(x->path, y->path);
	if (r != 0) return r;
	return x->idx - y->idx;
}

static int
wwi_is_decl(Node *d)
{
	return d->kind == N_FNDECL || d->kind == N_TYPEDECL
	    || d->kind == N_DEF || d->kind == N_LET;
}

int
wwi_emit(Checker *c, FILE *of, Node *file)
{
	/* §5: check_exported_type FIRST, before any byte is written — a
	 * producer without it can emit a dangling `.wwi`. */
	int bad = 0;
	for (Node *d = file->list; d; d = d->next) {
		if (!wwi_primary(d) || !d->export || !wwi_is_decl(d))
			continue;
		bad |= wwi_check_decl(c, d);
	}
	if (bad)
		return 1;

	/* package line: leaf of the first primary decl's module tag. */
	const char *pkg = "main";
	int found = 0;
	for (Node *d = file->list; d; d = d->next) {
		if (wwi_primary(d) && d->module && d->module[0]) {
			const char *dot = strrchr(d->module, '.');
			pkg = dot ? dot + 1 : d->module;
			found = 1;
			break;
		}
	}
	/* #11: a decl-less / export-less primary body carries no
	 * module-tagged decl, so the scan above finds nothing; fall back to
	 * the primary module identity stamped on the N_FILE node at parse
	 * time. A real root `package main` arrives via a bare module-reset
	 * and leaves file->module NULL, so it stays "main". The detector is
	 * scan-miss (`!found`), NOT pkg=="main": a body whose first tagged
	 * decl legitimately leafs to "main" must keep that, and must match
	 * selfhost's `found` flag byte-for-byte (rule 10). */
	if (!found && file->module && file->module[0]) {
		const char *dot = strrchr(file->module, '.');
		pkg = dot ? dot + 1 : file->module;
	}
	fprintf(of, "package %s;\n", pkg);

	/* imports — primary N_USE, byte-sorted by import path. */
	int nuse = 0;
	for (Node *u = file->list; u; u = u->next)
		if (u->kind == N_USE && wwi_primary(u))
			nuse++;
	if (nuse > 0) {
		struct useent *us = malloc((size_t)nuse * sizeof *us);
		int k = 0;
		for (Node *u = file->list; u; u = u->next) {
			if (u->kind != N_USE || !wwi_primary(u))
				continue;
			us[k].path = u->usepath ? u->usepath : u->str;
			us[k].idx = k;
			k++;
		}
		qsort(us, (size_t)nuse, sizeof *us, usecmp);
		for (int i = 0; i < nuse; i++)
			fprintf(of, "import %s;\n", us[i].path);
		free(us);
	}

	/* decls — exported primary, byte-sorted by symbol name. */
	int ndecl = 0;
	for (Node *d = file->list; d; d = d->next)
		if (wwi_primary(d) && d->export && wwi_is_decl(d))
			ndecl++;
	if (ndecl > 0) {
		struct declent *ds = malloc((size_t)ndecl * sizeof *ds);
		int k = 0;
		for (Node *d = file->list; d; d = d->next) {
			if (!wwi_primary(d) || !d->export || !wwi_is_decl(d))
				continue;
			ds[k].d = d;
			ds[k].idx = k;
			k++;
		}
		qsort(ds, (size_t)ndecl, sizeof *ds, declcmp);
		for (int i = 0; i < ndecl; i++)
			wwi_decl(of, ds[i].d);
		free(ds);
	}
	return 0;
}