ww/test/wcc/738_module_decl.c

/*
 * Pins three invariants from the module-system rewrite:
 *   1. The parser ACCEPTS `package foo;` as the first non-comment item.
 *   2. The parser ACCEPTS multiple `package X;` decls (concatenated
 *      multi-file streams from the driver) and stamps subsequent
 *      decls with whichever package is "current".
 *   3. Bare comments + a `package foo;` is still legal — the keyword
 *      may follow leading comments.
 *
 * The relaxed-or-strict missing-package check was softened to a
 * silent default (curmod=NULL) so legacy fragment-driven tests
 * still parse. That softening is documented at parsefile() in
 * cmd/wcc/parse.c.
 */
#include "ww.h"
#include <stdio.h>
#include <string.h>
#include <stdlib.h>

static int
parses_clean(const char *src)
{
	Arena *a = newarena();
	Lex l;
	Parser p;
	lexinit(&l, a, "<test>", src, strlen(src));
	parserinit(&p, a, &l);
	Node *n = parsefile(&p);
	int ok = (n != NULL && p.errs == 0 && l.errs == 0);
	freearena(a);
	return ok;
}

static const char *first_decl_module(Node *file) {
	if (file == NULL || file->list == NULL) return NULL;
	return file->list->module;
}

static int
check_module_stamps(const char *src, const char *want_first, const char *want_last)
{
	Arena *a = newarena();
	Lex l;
	Parser p;
	lexinit(&l, a, "<test>", src, strlen(src));
	parserinit(&p, a, &l);
	Node *n = parsefile(&p);
	int ok = (n != NULL && p.errs == 0 && l.errs == 0);
	if (ok) {
		const char *first = first_decl_module(n);
		Node *last = n->list;
		while (last && last->next) last = last->next;
		const char *lastmod = last ? last->module : NULL;
		int match_first = (first == NULL && want_first == NULL)
		    || (first != NULL && want_first != NULL
		        && strcmp(first, want_first) == 0);
		int match_last = (lastmod == NULL && want_last == NULL)
		    || (lastmod != NULL && want_last != NULL
		        && strcmp(lastmod, want_last) == 0);
		ok = match_first && match_last;
	}
	freearena(a);
	return ok;
}

enum packagemode {
	PACKAGE_FULL,
	PACKAGE_IMPORTS,
	PACKAGE_HEADER,
};

static const char *
packagemodename(enum packagemode mode)
{
	switch (mode) {
	case PACKAGE_FULL: return "full";
	case PACKAGE_IMPORTS: return "imports";
	case PACKAGE_HEADER: return "header";
	}
	return "unknown";
}

static Node *
parsepackage(Parser *p, enum packagemode mode)
{
	switch (mode) {
	case PACKAGE_FULL: return parsefile(p);
	case PACKAGE_IMPORTS: return parseimports(p);
	case PACKAGE_HEADER: return parsepackageheader(p);
	}
	return NULL;
}

/*
 * The three package-clause parsers intentionally expose different AST
 * boundaries.  Header/imports outer nodes stay on the package keyword; the
 * full parser's outer file node retains its historical 1:1 root position.
 * Imports-only parsing keeps its package marker on the keyword, while full
 * parsing moves only a blank package marker to the underscore so the checker
 * can diagnose the name token.  An ordinary package marker never moves.
 */
static int
check_package_shape(enum packagemode mode, const char *name,
    int markerline, int markercol)
{
	Arena *a = newarena();
	Lex l;
	Parser p;
	char src[160];
	snprintf(src, sizeof src,
	    "// leading comment\npackage %s;\nimport alpha;\n"
	    "fn x() void = {};\n", name);
	lexinit(&l, a, "shape.ww", src, strlen(src));
	parserinit(&p, a, &l);
	Node *file = parsepackage(&p, mode);
	Node *marker = file ? file->body : NULL;
	int wantmarker = mode != PACKAGE_HEADER;
	int outerline = mode == PACKAGE_FULL ? 1 : 2;
	int ok = file != NULL && file->kind == N_FILE
	    && p.errs == 0 && l.errs == 0
	    && file->pkgname != NULL && strcmp(file->pkgname, name) == 0
	    && file->pos.line == outerline && file->pos.col == 1
	    && ((mode == PACKAGE_FULL)
		|| (file->module != NULL && strcmp(file->module, name) == 0))
	    && ((!wantmarker && marker == NULL)
		|| (wantmarker && marker != NULL && marker->next == NULL
		    && marker->kind == N_FILE
		    && marker->pkgname != NULL
		    && strcmp(marker->pkgname, name) == 0
		    && marker->pos.line == markerline
		    && marker->pos.col == markercol));
	if (!ok) {
		fprintf(stderr,
		    "package shape mismatch: mode=%s name=%s "
		    "errs=%d/%d outer=%d:%d marker=%d:%d\n",
		    packagemodename(mode), name, p.errs, l.errs,
		    file ? file->pos.line : 0, file ? file->pos.col : 0,
		    marker ? marker->pos.line : 0, marker ? marker->pos.col : 0);
	}
	freearena(a);
	return ok;
}

static int
rejects_malformed_package(enum packagemode mode)
{
	Arena *a = newarena();
	Lex l;
	Parser p;
	const char *src = "package ;\nfn x() void = {};\n";
	char *diag = NULL;
	size_t diaglen = 0;
	FILE *prev = errout;
	FILE *capture = open_memstream(&diag, &diaglen);
	if (capture != NULL)
		errout = capture;
	lexinit(&l, a, "malformed.ww", src, strlen(src));
	parserinit(&p, a, &l);
	Node *file = parsepackage(&p, mode);
	int ok = file != NULL && p.errs + l.errs > 0;
	if (capture != NULL) {
		fclose(capture);
		errout = prev;
	}
	if (!ok)
		fprintf(stderr, "malformed package accepted in %s mode\n",
		    packagemodename(mode));
	free(diag);
	freearena(a);
	return ok;
}

/*
 * N_USE keeps the first import-spec token (alias when present) separately
 * from the first dotted-path token.  That distinction is checker-visible for
 * import errors, so pin it in each parser boundary rather than inferring the
 * path position later from the binding spelling.
 */
static int
check_import_path_position(enum packagemode mode, const char *spec,
    int speccol, int pathcol, const char *alias, const char *binding,
    int blank)
{
	Arena *a = newarena();
	Lex l;
	Parser p;
	char src[192];
	snprintf(src, sizeof src,
	    "package foo;\nimport %s;\nfn x() void = {};\n", spec);
	lexinit(&l, a, "usepath.ww", src, strlen(src));
	parserinit(&p, a, &l);
	Node *file = parsepackage(&p, mode);
	Node *u = file ? file->list : NULL;
	while (u != NULL && u->kind != N_USE)
		u = u->next;
	int ok = file != NULL && p.errs == 0 && l.errs == 0
	    && u != NULL
	    && u->pos.file != NULL && strcmp(u->pos.file, "usepath.ww") == 0
	    && u->pos.line == 2 && u->pos.col == 1
	    && u->usefile != NULL && strcmp(u->usefile, "usepath.ww") == 0
	    && u->useline == 2 && u->usecol == speccol
	    && u->usepathfile != NULL
	    && strcmp(u->usepathfile, "usepath.ww") == 0
	    && u->usepathline == 2 && u->usepathcol == pathcol
	    && u->usesource != NULL
	    && strcmp(u->usesource, "alpha.beta") == 0
	    && u->usepath != NULL && strcmp(u->usepath, "alpha.beta") == 0
	    && ((alias == NULL && u->usealias == NULL)
		|| (alias != NULL && u->usealias != NULL
		    && strcmp(u->usealias, alias) == 0))
	    && ((binding == NULL && u->str == NULL)
		|| (binding != NULL && u->str != NULL
		    && strcmp(u->str, binding) == 0))
	    && u->useblank == blank;
	if (!ok) {
		fprintf(stderr,
		    "import path position mismatch: mode=%s spec=%s "
		    "errs=%d/%d specpos=%d:%d pathpos=%d:%d\n",
		    packagemodename(mode), spec, p.errs, l.errs,
		    u ? u->useline : 0, u ? u->usecol : 0,
		    u ? u->usepathline : 0, u ? u->usepathcol : 0);
	}
	freearena(a);
	return ok;
}

int
main(void)
{
	int pass = 0, fail = 0;

	if (parses_clean("package foo;\nfn x() void = {};\n")) pass++;
	else { fprintf(stderr, "738[1] basic package accept FAILED\n"); fail++; }

	if (parses_clean("// header\n// more\npackage foo;\nfn x() void = {};\n")) pass++;
	else { fprintf(stderr, "738[2] package after comments FAILED\n"); fail++; }

	if (check_module_stamps(
	    "package foo;\nfn x() void = {};\nfn y() void = {};\n",
	    "foo", "foo")) pass++;
	else { fprintf(stderr, "738[3] decl module stamping FAILED\n"); fail++; }

	/* Mirrors driver-emitted multi-file concatenated streams. */
	if (check_module_stamps(
	    "package foo;\nfn a() void = {};\n"
	    "package bar;\nfn b() void = {};\n",
	    "foo", "bar")) pass++;
	else { fprintf(stderr, "738[4] mid-stream package switch FAILED\n"); fail++; }

	if (parses_clean(
	    "package foo;\nimport encoding.utf8;\nfn x() void = {};\n")) pass++;
	else { fprintf(stderr, "738[5] dotted import accept FAILED\n"); fail++; }

	/* Dotted import stores only the leaf identifier on N_USE.str
	 * (post-task-#22 — the driver translates the full dotted path to
	 * a directory walk; the checker only needs the package bareword
	 * for n_use → decl disambiguation). */
	{
		Arena *a = newarena();
		Lex l;
		Parser p;
		const char *src = "package foo;\nimport encoding.utf8;\n";
		lexinit(&l, a, "<test>", src, strlen(src));
		parserinit(&p, a, &l);
		Node *n = parsefile(&p);
		int ok = 0;
		if (n != NULL && p.errs == 0 && l.errs == 0 && n->list != NULL) {
			Node *u = n->list;
			ok = (u->kind == N_USE
			    && u->str != NULL
			    && strcmp(u->str, "utf8") == 0);
		}
		freearena(a);
		if (ok) pass++;
		else { fprintf(stderr, "738[6] dotted import leaf-store FAILED\n"); fail++; }
	}

	/* `package _;` is syntax in every loader/compiler parser mode. */
	if (check_package_shape(PACKAGE_HEADER, "_", 0, 0)) pass++;
	else { fprintf(stderr, "738[7] blank package header shape FAILED\n"); fail++; }

	if (check_package_shape(PACKAGE_IMPORTS, "_", 2, 1)) pass++;
	else { fprintf(stderr, "738[8] blank imports-only shape FAILED\n"); fail++; }

	if (check_package_shape(PACKAGE_FULL, "_", 2, 9)) pass++;
	else { fprintf(stderr, "738[9] blank full-parser shape FAILED\n"); fail++; }

	/* Nonblank package-marker positions remain on the package keyword. */
	if (check_package_shape(PACKAGE_HEADER, "foo", 0, 0)) pass++;
	else { fprintf(stderr, "738[10] named package header shape FAILED\n"); fail++; }

	if (check_package_shape(PACKAGE_IMPORTS, "foo", 2, 1)) pass++;
	else { fprintf(stderr, "738[11] named imports-only shape FAILED\n"); fail++; }

	if (check_package_shape(PACKAGE_FULL, "foo", 2, 1)) pass++;
	else { fprintf(stderr, "738[12] named full-parser shape FAILED\n"); fail++; }

	/* Extending the name slot to `_` must not accept a missing name. */
	if (rejects_malformed_package(PACKAGE_HEADER)) pass++;
	else { fprintf(stderr, "738[13] malformed package header FAILED\n"); fail++; }

	if (rejects_malformed_package(PACKAGE_IMPORTS)) pass++;
	else { fprintf(stderr, "738[14] malformed imports-only package FAILED\n"); fail++; }

	if (rejects_malformed_package(PACKAGE_FULL)) pass++;
	else { fprintf(stderr, "738[15] malformed full-parser package FAILED\n"); fail++; }

	/* Default imports use the first dotted-path identifier for both slots. */
	for (enum packagemode mode = PACKAGE_FULL; mode <= PACKAGE_HEADER; mode++) {
		if (check_import_path_position(mode, "alpha.beta", 8, 8,
		    NULL, "beta", 0)) pass++;
		else {
			fprintf(stderr, "738 default import path position FAILED (%s)\n",
			    packagemodename(mode));
			fail++;
		}
	}

	/* Explicit aliases keep the binding/spec token at col 8, path at 14. */
	for (enum packagemode mode = PACKAGE_FULL; mode <= PACKAGE_HEADER; mode++) {
		if (check_import_path_position(mode, "local alpha.beta", 8, 14,
		    "local", "local", 0)) pass++;
		else {
			fprintf(stderr, "738 explicit import path position FAILED (%s)\n",
			    packagemodename(mode));
			fail++;
		}
	}

	/* A blank alias likewise keeps `_` at col 8 and the path at col 10. */
	for (enum packagemode mode = PACKAGE_FULL; mode <= PACKAGE_HEADER; mode++) {
		if (check_import_path_position(mode, "_ alpha.beta", 8, 10,
		    NULL, NULL, 1)) pass++;
		else {
			fprintf(stderr, "738 blank import path position FAILED (%s)\n",
			    packagemodename(mode));
			fail++;
		}
	}

	printf("738_module_decl: %d pass, %d fail\n", pass, fail);
	return fail == 0 ? 0 : 1;
}