/*
 * 840_zeroinit_run — task #16. A bare `let x: T;` (no initializer) must
 * zero-fill its slot (Go zero-value, rob's ruling). Before the fix, cgen
 * emitted the zero-fill ONLY for 8B-primitive slots and >8B composites; a
 * SUB-8 aggregate (`let c: [3]u8;` = 3 bytes, a 3-byte struct, …) matched
 * neither arm and fell through to NOTHING, so the slot read whatever the
 * stack held.
 *
 * THE BUG (cstage == wwstage, BOTH wrong — shared gap, NOT rule-10):
 *   ken's bytes verdict (.ai/ken-bytes16-verdict.md) traced lib/bytes'
 *   green-but-broken 967 to ltrim_cases' `let c: [3]u8;` reading a prior
 *   deep-frame sibling's leftover bytes. The masking was a stack-zero
 *   accident: a bare let on a FRESH frame happens to read 0, so the bug
 *   only fires when a deep-framed fn ran first. byte-id was BLIND (#263):
 *   both stages emitted the identical no-store sequence.
 *
 * REPRO SHAPE (ken's bytes-independent minimal repro, generalised):
 *   dirty() writes a big local ([64]u8 = 222) to soil the stack region;
 *   probe() then declares the bare let as its FIRST local — reusing
 *   dirty()'s slot — and reads it. A correct zero-fill returns 0; the
 *   pre-fix garbage returned (222 * width) & 0xff (154 for [3]u8).
 *
 * THE FIX (#16, both stages): widen the no-rhs zero-fill gate from
 *   `sz > 8` to `sz > 0` (cgen.c N_LET) and add the symmetric `zsz > 0`
 *   run arm (cgenstmt.ww cglet), so 1..7-byte slots zero through the same
 *   MOVL/MOVB tail. sz == 8 keeps its immediate MOVQ $0; sz == 0 (`[0]T`)
 *   needs no stores.
 *
 * EACH ROW CARRIES BOTH DIMENSIONS (802 model):
 *   (a) cstage `ww build` + run, asserting exit 0 — pins that the bare let
 *       reads zero even after a dirtied frame (runtime correctness the
 *       byte-id gate cannot see).
 *   (b) w6c vs w6c_ww `.s` cmp — FAILS if the stages diverge (rule-10).
 *
 * GATE POLARITY: must stay GREEN. A nonzero exit means a bare-let slot
 * regressed to reading stack garbage; a byte-id FAIL means the stages
 * diverged on the zero-fill emission.
 */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/stat.h>
#include <sys/wait.h>

static int
runwait(const char *cmd)
{
	int rc = system(cmd);
	if (rc == -1) return -1;
	if (WIFEXITED(rc)) return WEXITSTATUS(rc);
	return -1;
}

struct row { const char *label; const char *src; int want_exit; };

/* Every row shares the dirty()→probe() spine: dirty soils the frame,
 * probe declares the bare let FIRST and reads it. want_exit is 0 (all
 * reads zero). The discriminating rows are the SUB-8 aggregates ([3]u8,
 * [5]u8, [7]u8, the 3-byte struct) — the exact gap #16 closed; the
 * scalar / [20]u8 / str / slice rows are controls (already zeroed pre-#16
 * via the sz==8 and sz>8 arms) pinning no-regression. */
static const struct row rows[] = {
	/* the exact ken repro: [3]u8 sub-8 array. Pre-fix returned 154. */
	{ "arr3_u8",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = {\n"
	  "  let c: [3]u8;\n"
	  "  return (c[0]: i32) + (c[1]: i32) + (c[2]: i32);\n"
	  "};\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* sub-8, width 5 — exercises the MOVL+MOVB tail split (4 + 1). */
	{ "arr5_u8",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = {\n"
	  "  let c: [5]u8; let acc: i32 = 0; let j: i32 = 0;\n"
	  "  for (j < 5) { acc += (c[j]: i32); j += 1; };\n"
	  "  return acc;\n"
	  "};\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* sub-8, width 7 — the widest sub-8 extent (MOVL + MOVW + MOVB is
	 * unreachable under the run's 4/1 tail, so this is MOVL + 3×MOVB). */
	{ "arr7_u8",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = {\n"
	  "  let c: [7]u8; let acc: i32 = 0; let j: i32 = 0;\n"
	  "  for (j < 7) { acc += (c[j]: i32); j += 1; };\n"
	  "  return acc;\n"
	  "};\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* sub-8 STRUCT (3 bytes) — the non-array half of the gap. Read each
	 * field; a garbage slot would sum nonzero. */
	{ "struct3_u8",
	  "package main;\n"
	  "type S = struct { a: u8, b: u8, c: u8 };\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = {\n"
	  "  let s: S;\n"
	  "  return (s.a: i32) + (s.b: i32) + (s.c: i32);\n"
	  "};\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* CONTROL: scalar i32 (sz==8) — zeroed pre-#16 by the MOVQ $0 arm. */
	{ "scalar_i32",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = { let x: i32; return x; };\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* CONTROL: [20]u8 (>8) — zeroed pre-#16 by the #84 sz>8 run. */
	{ "arr20_u8",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = {\n"
	  "  let c: [20]u8; let acc: i32 = 0; let j: i32 = 0;\n"
	  "  for (j < 20) { acc += (c[j]: i32); j += 1; };\n"
	  "  return acc;\n"
	  "};\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* CONTROL: bare str header (24B) — rob's declmod `let empty: str;`
	 * shape. .len must read 0 (zeroed pre-#16 by the sz>8 run). */
	{ "str_hdr",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = { let empty: str; return empty.len; };\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	/* CONTROL: bare slice header (24B) — .len must read 0. */
	{ "slice_hdr",
	  "package main;\n"
	  "fn dirty() void = {\n"
	  "  let big: [64]u8; let i: i32 = 0;\n"
	  "  for (i < 64) { big[i] = 222u8; i += 1; };\n"
	  "};\n"
	  "fn probe() i32 = { let xs: []i32; return xs.len; };\n"
	  "export fn main() i32 = { dirty(); return probe(); };\n", 0 },
	{ NULL, NULL, 0 }
};

static int
slurp_eq(const char *a, const char *b)
{
	FILE *fa = fopen(a, "rb");
	FILE *fb = fopen(b, "rb");
	if (!fa || !fb) { if (fa) fclose(fa); if (fb) fclose(fb); return -1; }
	int rc = 0;
	for (;;) {
		int ca = fgetc(fa);
		int cb = fgetc(fb);
		if (ca != cb) { rc = -1; break; }
		if (ca == EOF) break;
	}
	fclose(fa); fclose(fb);
	return rc;
}

int
main(void)
{
	const char *bin = getenv("BIN");
	if (!bin) bin = "out/bin";
	char absbin[1024];
	if (bin[0] != '/') {
		char cwd[1024];
		if (getcwd(cwd, sizeof cwd) == NULL) return 1;
		snprintf(absbin, sizeof absbin, "%s/%s", cwd, bin);
		bin = absbin;
	}

	char w6c[1100], w6c_ww[1100];
	snprintf(w6c, sizeof w6c, "%s/w6c", bin);
	snprintf(w6c_ww, sizeof w6c_ww, "%s/w6c_ww", bin);
	if (access(w6c_ww, X_OK) != 0) {
		fprintf(stderr, "zeroinit: w6c_ww missing — cannot run the "
		    "cs==ww byte-id gate\n");
		return 1;
	}

	int n = 0, fail = 0;
	for (int i = 0; rows[i].src; i++, n++) {
		char src[64];
		snprintf(src, sizeof src, "/tmp/wwzi_%d_%d.ww", getpid(), i);
		FILE *f = fopen(src, "wb");
		if (f == NULL) { fail++; continue; }
		fputs(rows[i].src, f);
		fclose(f);

		/* (a) cstage build + run in a scratch dir. */
		char tmpdir[64];
		snprintf(tmpdir, sizeof tmpdir, "/tmp/wwzi_%d_d_%d", getpid(), i);
		mkdir(tmpdir, 0755);

		char cmd[2048];
		snprintf(cmd, sizeof cmd, "cd %s && %s/ww build %s",
		    tmpdir, bin, src);
		if (runwait(cmd) != 0) {
			fprintf(stderr, "row[%s]: cstage build failed\n",
			    rows[i].label);
			fail++;
			unlink(src); rmdir(tmpdir);
			continue;
		}

		char outbin[128];
		const char *base = strrchr(src, '/');
		base = base ? base + 1 : src;
		snprintf(outbin, sizeof outbin, "%s/%s", tmpdir, base);
		char *dot = strrchr(outbin, '.');
		if (dot && strcmp(dot, ".ww") == 0) *dot = '\0';

		int got = runwait(outbin);
		if (got != rows[i].want_exit) {
			fprintf(stderr, "row[%s]: cstage exit %d, want %d "
			    "(bare let read stack garbage?)\n",
			    rows[i].label, got, rows[i].want_exit);
			fail++;
		}
		unlink(outbin); rmdir(tmpdir);

		/* (b) cs==ww byte-id gate. */
		char cs_s[64], ws_s[64];
		snprintf(cs_s, sizeof cs_s, "/tmp/wwzi_%d_%d_cs.s", getpid(), i);
		snprintf(ws_s, sizeof ws_s, "/tmp/wwzi_%d_%d_ww.s", getpid(), i);

		snprintf(cmd, sizeof cmd, "%s -o %s %s 2>/dev/null", w6c, cs_s, src);
		if (runwait(cmd) != 0) {
			fprintf(stderr, "row[%s]: w6c failed\n", rows[i].label);
			fail++; unlink(src); continue;
		}
		snprintf(cmd, sizeof cmd, "%s -o %s %s 2>/dev/null",
		    w6c_ww, ws_s, src);
		if (runwait(cmd) != 0) {
			fprintf(stderr, "row[%s]: w6c_ww failed\n", rows[i].label);
			fail++; unlink(src); unlink(cs_s); continue;
		}
		if (slurp_eq(cs_s, ws_s) != 0) {
			fprintf(stderr, "row[%s]: cstage/wwstage .s DIFFER "
			    "(rule-10 byte-id violation)\n", rows[i].label);
			fail++;
		}
		unlink(src); unlink(cs_s); unlink(ws_s);
	}

	if (fail) {
		fprintf(stderr, "%d/%d zeroinit tests failed\n", fail, n);
		return 1;
	}
	printf("zeroinit: %d/%d ok (cstage run + cs==ww byte-id)\n", n, n);
	return 0;
}