ww/lib/regex/regex_test.ww

// regex_test — exercises the lib/regex fold-1 data model (the type
// model + finish()) and the fold-2a compile() literal core. Run with
// `out/bin/ww run lib/regex/regex_test.ww`.
//
// Fold 2a ports compile()'s lit/any/match arms only; exec lives in
// later folds, so the compile_* cases pin the emitted inst PROGRAM
// (shape + payloads via indexed match-extraction), not matching.
// charclass_map's fn-ptr table is deferred behind the array→slice
// element-coercion checker gap (see regex.ww), so this test does not
// exercise the POSIX-class predicate dispatch yet — it pins variant
// discrimination (including the nominally-distinct same-underlying
// inst_split/inst_jump/inst_groupstart `size` aliases and the
// inst_any/inst_skip/inst_groupend `void` aliases), payload extraction,
// the regex/capture struct shapes, and finish(). Same
// signalled-then-fail()-with-+10 pattern as the rest of the stdlib
// run-tests; the non-zero exit pinpoints the failing case.
//
// Struct literals below name the type UNQUALIFIED (`inst_charset { … }`,
// not `regex.inst_charset { … }`): the parser rejects a module-qualified
// name in struct-literal position (#29), and the imported type
// is in scope unqualified.
package regex;

import regex;
import os;
import strings;

let signalled: i32 = 0;
fn fail() void = { os.exit(signalled + 10); };

// inst_lit / inst_match carry distinguishable payloads (rune / bool).
@test fn lit_and_match() void = {
	let a: regex.inst = ('a': regex.inst_lit);
	match (a) {
	case let l: regex.inst_lit => { if ((l: rune) != 'a') { fail(); }; };
	case => fail();
	};

	let m: regex.inst = (true: regex.inst_match);
	match (m) {
	case let b: regex.inst_match => { if (!(b: bool)) { fail(); }; };
	case => fail();
	};
};

// The three `size`-aliased variants are nominally distinct: a value
// built as inst_split must match inst_split, never inst_jump /
// inst_groupstart, despite identical underlying storage.
@test fn size_aliases_distinct() void = {
	let sp: regex.inst = ((5: size): regex.inst_split);
	match (sp) {
	case let s: regex.inst_split => { if ((s: size) != (5: size)) { fail(); }; };
	case let j: regex.inst_jump => fail();
	case let g: regex.inst_groupstart => fail();
	case => fail();
	};

	let jp: regex.inst = ((9: size): regex.inst_jump);
	match (jp) {
	case let j: regex.inst_jump => { if ((j: size) != (9: size)) { fail(); }; };
	case let s: regex.inst_split => fail();
	case => fail();
	};

	let gs: regex.inst = ((2: size): regex.inst_groupstart);
	match (gs) {
	case let g: regex.inst_groupstart => { if ((g: size) != (2: size)) { fail(); }; };
	case let s: regex.inst_split => fail();
	case => fail();
	};
};

// The `void`-aliased variants are likewise nominally distinct.
@test fn void_aliases_distinct() void = {
	let av: regex.inst_any;
	let an: regex.inst = av;
	match (an) {
	case let a: regex.inst_any => void;
	case let k: regex.inst_skip => fail();
	case let e: regex.inst_groupend => fail();
	case => fail();
	};

	let sv: regex.inst_skip;
	let sk: regex.inst = sv;
	match (sk) {
	case let k: regex.inst_skip => void;
	case let a: regex.inst_any => fail();
	case => fail();
	};

	let gv: regex.inst_groupend;
	let ge: regex.inst = gv;
	match (ge) {
	case let e: regex.inst_groupend => void;
	case let a: regex.inst_any => fail();
	case let k: regex.inst_skip => fail();
	case => fail();
	};
};

// inst_charset carries a struct payload; its fields survive the union
// round-trip.
@test fn charset_payload() void = {
	let c: regex.inst = (inst_charset { idx = 3, is_positive = true });
	match (c) {
	case let cs: regex.inst_charset => {
		if (cs.idx != (3: size)) { fail(); };
		if (!cs.is_positive) { fail(); };
	};
	case => fail();
	};
};

// inst_repeat round-trips through the inst union with its plain `size`
// fields intact. Matching the nested (void | size) min/max bounds back
// out is DEFERRED: `match` on a tagged-union-typed struct field
// diverges cs≠ww (#26 — the wwstage frames it wider), so
// asserting the bounds here would seed a rule-10-divergent fixture.
@test fn repeat_payload() void = {
	let r: regex.inst = (inst_repeat {
		id = 1, origin = 4, min = (2: size), max = void,
	});
	match (r) {
	case let rp: regex.inst_repeat => {
		if (rp.id != (1: size)) { fail(); };
		if (rp.origin != (4: size)) { fail(); };
	};
	case => fail();
	};
};

// The regex/capture structs hold their fields; finish() is a no-op
// (no-free runtime) and must accept a built regex.
@test fn struct_shapes_and_finish() void = {
	let cap: regex.capture = capture {
		content = "abc",
		start = 0,
		start_bytesize = 0,
		end = 3,
		end_bytesize = 3,
	};
	if (cap.content.len != 3) { fail(); };
	if (cap.end != (3: size)) { fail(); };

	// regex's insts/charsets ([]inst / []charset) are left empty here:
	// fold 1 ports no compile() to populate them, an empty `[]` literal
	// is unspellable as a typed slice (#25 — array→slice
	// element-coercion gap), and a struct-literal slice-field store
	// drops len/cap (#24). Declaring the regex zeroes both
	// slice headers to {0,0,0}; only n_reps is set explicitly.
	let re: regex.regex;
	re.n_reps = 0;
	if (re.n_reps != (0: size)) { fail(); };
	if (re.insts.len != 0) { fail(); };
	regex.finish(&re);
};

// compile("abc") emits the 5-inst literal program: the leading
// unanchored inst_skip (regex.ha:261-263), one inst_lit per rune, the
// epilogue inst_match(false) (ha:475-477). compile()'s
// (regex | error | nomem) return is the first >24B tagged payload in
// the tree — the receive shapes here double as #38 sret consumers
// (typed-let + match here; scrutinee-direct in compile_empty_program).
@test fn compile_literal_program() void = {
	let c: (regex.regex | regex.error | nomem) = regex.compile("abc");
	match (c) {
	case let re: regex.regex => {
		if (re.insts.len != 5) { fail(); };
		match (re.insts[0]) {
		case let k: regex.inst_skip => void;
		case => fail();
		};
		match (re.insts[1]) {
		case let l: regex.inst_lit => { if ((l: rune) != 'a') { fail(); }; };
		case => fail();
		};
		match (re.insts[2]) {
		case let l: regex.inst_lit => { if ((l: rune) != 'b') { fail(); }; };
		case => fail();
		};
		match (re.insts[3]) {
		case let l: regex.inst_lit => { if ((l: rune) != 'c') { fail(); }; };
		case => fail();
		};
		match (re.insts[4]) {
		case let m: regex.inst_match => { if ((m: bool)) { fail(); }; };
		case => fail();
		};
		if (re.charsets.len != 0) { fail(); };
		if (re.n_reps != (0: size)) { fail(); };
		regex.finish(&re);
	};
	case => fail();
	};
};

// '.' compiles to inst_any between the literals (regex.ha:460-461):
// [skip, lit 'a', any, lit 'c', match(false)].
@test fn compile_any_program() void = {
	let c: (regex.regex | regex.error | nomem) = regex.compile("a.c");
	match (c) {
	case let re: regex.regex => {
		if (re.insts.len != 5) { fail(); };
		match (re.insts[0]) {
		case let k: regex.inst_skip => void;
		case => fail();
		};
		match (re.insts[1]) {
		case let l: regex.inst_lit => { if ((l: rune) != 'a') { fail(); }; };
		case => fail();
		};
		match (re.insts[2]) {
		case let a: regex.inst_any => void;
		case => fail();
		};
		match (re.insts[3]) {
		case let l: regex.inst_lit => { if ((l: rune) != 'c') { fail(); }; };
		case => fail();
		};
		match (re.insts[4]) {
		case let m: regex.inst_match => { if ((m: bool)) { fail(); }; };
		case => fail();
		};
		regex.finish(&re);
	};
	case => fail();
	};
};

// compile("") is exactly [inst_match(false)]: the leading skip must
// not fire on immediate done (regex.ha:261 gates on `next is rune`),
// and the epilogue guard must fire on the empty program.
@test fn compile_empty_program() void = {
	match (regex.compile("")) {
	case let re: regex.regex => {
		if (re.insts.len != 1) { fail(); };
		match (re.insts[0]) {
		case let m: regex.inst_match => { if ((m: bool)) { fail(); }; };
		case => fail();
		};
		regex.finish(&re);
	};
	case => fail();
	};
};

// Every deferred metacharacter is a LOUD error carrying the exact
// fold-boundary text — falling through to the literal default would
// silently compile a wrong program, and any OTHER error text would
// mean an arm was half-ported. One pattern per deferred arm so the
// fold that ports an arm consciously deletes its row. '^' leads its
// pattern: the r_idx==0 skip gate (regex.ha:261) must compose with
// the loud arm, not bypass it.
@test fn compile_metachar_loud() void = {
	let pats: [11]str = [
		"a\\", "^a", "a$", "a[", "a(", "a)",
		"a|", "a{", "a?", "a*", "a+",
	];
	let i: i32 = 0;
	for (i < len(pats)) {
		match (regex.compile(pats[i])) {
		case let e: regex.error => {
			if (strings.compare((e: str),
				"regex: metacharacter not yet ported") != 0) {
				fail();
			};
		};
		case => fail();
		};
		i += 1;
	};
};

export fn main() i32 = {
	signalled = 1; lit_and_match();
	signalled = 2; size_aliases_distinct();
	signalled = 3; void_aliases_distinct();
	signalled = 4; charset_payload();
	signalled = 5; repeat_payload();
	signalled = 6; struct_shapes_and_finish();
	signalled = 7; compile_literal_program();
	signalled = 8; compile_any_program();
	signalled = 9; compile_empty_program();
	signalled = 10; compile_metachar_loud();
	return 0;
};