From 2f788305b44c844336b867f37a2997d6bab70329 Mon Sep 17 00:00:00 2001
From: Hojun-Cho <ghwns6743@gmail.com>
Date: Mon, 22 Jun 2026 01:48:38 +0900
Subject: [PATCH] test: migrate 110_uniesc rows to in-language @test (test/lang
 pilot)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Proof-of-path for the test-arch reframe: pure-behavior rows become
in-language @test row-tables run by `ww test`, reserving C drivers for
the two irreducible jobs (byte-id + bootstrap). 110_uniesc's 6 value
rows become [N]struct row-tables in a new test/lang/ tree; the trailing
.wwi round-trip stays in C as external-observer residue. Additive — the
C rows are kept; de-dup and make-rewiring land at fold 6 (task #12).
Green on both stages.
---
 test/lang/uniesc_test.ww | 76 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 76 insertions(+)
 create mode 100644 test/lang/uniesc_test.ww

diff --git a/test/lang/uniesc_test.ww b/test/lang/uniesc_test.ww
new file mode 100644
index 00000000..b95055a5
--- /dev/null
+++ b/test/lang/uniesc_test.ww
@@ -0,0 +1,76 @@
+// uniesc_test — in-language behavior rows for \u / \U / \x escapes,
+// migrated from test/wcc/110_uniesc_run.c (task #12 proof-of-path). The
+// compiler-under-test compiles THIS file, so a miscompile of an escape
+// decode fails the assert in-language rather than via an external runner.
+//
+// Each row's literal is the exact escape form from the C source (not the
+// raw multibyte character): the feature under test is the escape-decode
+// path (ref/hare/hare/lex/lex.ha:347 lex_unicode + appendrune in
+// lex_string), distinct from source-UTF-8 decoding. A rune literal carries
+// the raw codepoint; a string literal UTF-8-encodes it into 1-4 bytes; \x
+// >=0x80 shares the codepoint path so it widens to its multibyte form.
+//
+// Carve-out: the C file's trailing .wwi round-trip case (an exported
+// wide-rune `def` re-serialized byte-exact) is an external-observer
+// residue and stays in C — not migrated here.
+
+package uniesc_test;
+
+import strings;
+
+// One rune row: the escape literal must decode to codepoint `cp`.
+type runecase = struct {
+	lit: rune,
+	cp: u32,
+};
+
+// rune_u_2byte (é -> U+00E9) and rune_U_emoji (\U0001F600 -> U+1F600,
+// the >0xFFFF case that forced the int-cast spelling pre-#50).
+@test fn rune_escapes() void = {
+	let rows: [2]runecase = [
+		runecase { lit = '\u00e9', cp = 233u32 },
+		runecase { lit = '\U0001F600', cp = 128512u32 },
+	];
+	let i: i32 = 0;
+	for (i < len(rows)) {
+		assert((rows[i].lit: u32) == rows[i].cp);
+		i += 1;
+	};
+};
+
+// One string row: the escape literal must UTF-8-encode to `n` bytes
+// b0..b3 (unused tail bytes are 0 and guarded by the `n` check).
+type strcase = struct {
+	lit: str,
+	n: i32,
+	b0: u8,
+	b1: u8,
+	b2: u8,
+	b3: u8,
+};
+
+// str_u_2byte (é -> C3 A9), str_u_3byte (€ -> E2 82 AC),
+// str_U_4byte (\U0001F600 -> F0 9F 98 80), str_x_widens (\xe9 -> C3 A9,
+// the >=0x80 \x byte that now widens to its 2-byte UTF-8 form).
+@test fn string_escapes() void = {
+	let rows: [4]strcase = [
+		strcase { lit = "\u00e9", n = 2,
+			b0 = 0xc3u8, b1 = 0xa9u8, b2 = 0u8, b3 = 0u8 },
+		strcase { lit = "\u20ac", n = 3,
+			b0 = 0xe2u8, b1 = 0x82u8, b2 = 0xacu8, b3 = 0u8 },
+		strcase { lit = "\U0001F600", n = 4,
+			b0 = 0xf0u8, b1 = 0x9fu8, b2 = 0x98u8, b3 = 0x80u8 },
+		strcase { lit = "\xe9", n = 2,
+			b0 = 0xc3u8, b1 = 0xa9u8, b2 = 0u8, b3 = 0u8 },
+	];
+	let i: i32 = 0;
+	for (i < len(rows)) {
+		let b: []u8 = strings.toutf8(rows[i].lit);
+		assert(len(b) == rows[i].n);
+		assert(b[0] == rows[i].b0);
+		if (rows[i].n > 1) { assert(b[1] == rows[i].b1); };
+		if (rows[i].n > 2) { assert(b[2] == rows[i].b2); };
+		if (rows[i].n > 3) { assert(b[3] == rows[i].b3); };
+		i += 1;
+	};
+};