ww/lib/shlex/shlextest.ww

// shlextest — exercises lib/shlex. Run with
// `out/bin/ww run lib/shlex/shlextest.ww`.
//
// Two cohorts, grouped Hare-style — one @test fn per cohort, table-
// driven inside via per-arity helpers:
//
//   • split rows go through [[check1]] / [[check2]] / [[check3]] /
//     [[checkerr]]. Inputs are lifted from ref/hare/shlex/+test.ha
//     @test fn split() (the de-facto spec for this port). Per-row
//     arity is fixed (1, 2, or 3 expected tokens across the Hare
//     cases), so we ship per-arity helpers rather than a full
//     variadic check that would obscure the row data.
//
//   • quote rows go through [[checkquote]] — uniform (input, expected)
//     shape against a memio.dynamic sink, mirroring Hare's escape.ha
//     testquote table.
//
// Failure path: each @test fn bumps `signalled` to its slot index,
// the helpers do `exit(signalled + 10)` on miscompare so the harness
// reports `WEXITSTATUS = 11..N` pointing at the failing scenario.
// Same convention as fnmatchtest / logtest.

package shlex;

import shlex;
import io;
import memio;

// Direct rt_syscall binding rather than `use os;` — os exports
// read/write/close, which collide with io.read/write/close under the
// driver's flat-scope concat. Mirrors fnmatchtest / logtest / fmttest
// / bufiotest.
@symbol("rt_syscall") fn syscall1ww(num: i64, a: i64) i64;
fn doexit(code: i32) void = {
	syscall1ww(60i64, code: i64);
};

let signalled: i32 = 0;

fn fail() void = { doexit(signalled + 10); };

fn streq(a: str, b: str) bool = {
	if (a.len != b.len) { return false; };
	let i: i32 = 0;
	for (i < a.len) {
		if (a[i] != b[i]) { return false; };
		i += 1;
	};
	return true;
};

// checkN — split `in` and assert the result is a slice of length N
// matching the named expected tokens. Leaks the result (test process
// is short-lived; same precedent as fnmatchtest).

fn check1(in: str, e0: str) void = {
	let r = shlex.split(in);
	match (r) {
	case shlex.syntaxerr => { fail(); };
	case let s: []str => {
		if (s.len != 1) { fail(); };
		if (!streq(s[0], e0)) { fail(); };
	};
	};
};

fn check2(in: str, e0: str, e1: str) void = {
	let r = shlex.split(in);
	match (r) {
	case shlex.syntaxerr => { fail(); };
	case let s: []str => {
		if (s.len != 2) { fail(); };
		if (!streq(s[0], e0)) { fail(); };
		if (!streq(s[1], e1)) { fail(); };
	};
	};
};

fn check3(in: str, e0: str, e1: str, e2: str) void = {
	let r = shlex.split(in);
	match (r) {
	case shlex.syntaxerr => { fail(); };
	case let s: []str => {
		if (s.len != 3) { fail(); };
		if (!streq(s[0], e0)) { fail(); };
		if (!streq(s[1], e1)) { fail(); };
		if (!streq(s[2], e2)) { fail(); };
	};
	};
};

fn checkerr(in: str) void = {
	let r = shlex.split(in);
	match (r) {
	case shlex.syntaxerr => {};
	case let s: []str => { fail(); };
	};
};

fn checkempty(in: str) void = {
	let r = shlex.split(in);
	match (r) {
	case shlex.syntaxerr => { fail(); };
	case let s: []str => {
		if (s.len != 0) { fail(); };
	};
	};
};

fn checkquote(in: str, expected: str) void = {
	let st: memio.stream = memio.dynamic();
	let snk: io.stream = &st.vt;

	let r = shlex.quote(snk, in);
	let n: size = 0;
	match (r) {
	case let v: size => { n = v; };
	case let _e: io.error => { fail(); };
	};
	if (n != expected.len: size) { fail(); };

	let view: str = memio.string(&st);
	if (!streq(view, expected)) { fail(); };

	let _c: (void | io.error) = io.close(snk);
};

// ---- split: Hare's @test fn split() table --------------------------
//
// 9 success rows + 3 syntaxerr rows ported VERBATIM from
// ref/hare/shlex/+test.ha; plus one ww-specific edge (empty input →
// empty []str) confirmed by drew.
//
// Local @test fns are `test_*`-prefixed because `use shlex;` flat-
// concats shlex's exported names (split / quote / quotestr / strerror)
// into the fixture's namespace, and bare `fn split() ...` would
// duplicate-define them. Retires when task #17 (cgen mod-mangles fn
// labels) lands.

@test fn test_split() void = {
	check1("hello\\ world", "hello world");
	check1("'hello\\ world'", "hello\\ world");
	check1("\"hello\\\\world\"", "hello\\world");
	// "hello "'"'"world"'"' → hello "world"
	check1("\"hello \"'\"'\"world\"'\"'", "hello \"world\"");
	check3("hello '' world", "hello", "", "world");
	check2("Empty ''", "Empty", "");
	check2("   Leading spaces", "Leading", "spaces");
	check3("with\\ backslashes 'single quoted' \"double quoted\"",
	    "with backslashes", "single quoted", "double quoted");
	check2("'multiple spaces'      42", "multiple spaces", "42");
	// pin the '\t'/'\n' arms of split's whitespace test.
	check3("a\tb\nc", "a", "b", "c");

	// Invalid
	checkerr("\"dangling double quote");
	checkerr("'dangling single quote");
	checkerr("unterminated\\ backslash \\");

	// Empty input → empty []str (ww edge confirmed by drew).
	checkempty("");
};

// ---- quote: Hare's testquote rows + the empty-input edge ----------
//
// 4 rows from ref/hare/shlex/+test.ha @test fn quote(). The empty-
// input row (→ `''`) is implementation-specific (Hare's testquote
// doesn't cover it) but is documented behaviour per shlex.ww's
// quote() header — exercised here so the contract is load-bearing.

@test fn test_quote() void = {
	checkquote("hello", "hello");
	checkquote("hello world", "'hello world'");
	checkquote("'hello' \"world\"", "''\"'\"'hello'\"'\"' \"world\"'");
	checkquote("hello\\world", "'hello\\world'");
	checkquote("", "''");
	// pin issafe's special-char set: all safe → emitted raw, unquoted.
	checkquote("@%+=:,./-", "@%+=:,./-");
};

// ---- quotestr ------------------------------------------------------

@test fn test_quotestr() void = {
	let r: str = shlex.quotestr("hello world");
	if (!streq(r, "'hello world'")) { fail(); };
	// leak r — short-lived test process, same precedent as fnmatchtest.
};

// ---- strerror ------------------------------------------------------

@test fn test_strerror() void = {
	let e: shlex.syntaxerr;
	let s: str = shlex.strerror(e);
	if (!streq(s, "Invalid shell syntax")) { fail(); };
};

export fn main() i32 = {
	signalled = 1; test_split();
	signalled = 2; test_quote();
	signalled = 3; test_quotestr();
	signalled = 4; test_strerror();
	return 0;
};