1 files changed, 87 insertions, 0 deletions
diff --git a/tests/drivers/fuzz_compiler.py b/tests/drivers/fuzz_compiler.py
new file mode 100644
index 0000000..512ec26
--- /dev/null
+++ b/tests/drivers/fuzz_compiler.py
@@ -0,0 +1,87 @@
+#!/usr/bin/env python3
+"""Deterministic grammar fuzzing for the parser, runtime, and ownership paths.
+
+Usage: python3 tests/drivers/fuzz_compiler.py [path/to/zy] [path/to/zym] [--js]
+"""
+import random
+import subprocess
+import sys
+
+
+args = [arg for arg in sys.argv[1:] if arg != "--js"]
+js = "--js" in sys.argv[1:]
+zy = args[0] if len(args) > 0 else "bin/zy"
+zym = args[1] if len(args) > 1 else "bin/zym"
+rng = random.Random(0x5A17)
+values = []
+library_values = []
+
+
+def expr(depth):
+    if depth == 0 or rng.randrange(4) == 0:
+        value = rng.randrange(10)
+        return str(value), value
+    left, a = expr(depth - 1)
+    right, b = expr(depth - 1)
+    op = rng.choice(("+", "-", "*"))
+    if op == "+":
+        value = a + b
+    elif op == "-":
+        value = a - b
+    else:
+        value = a * b
+    return f"({left} {op} {right})", value
+
+
+lines = ["include std;", "fn churn(n) {", " int i = 0;", " while (i < n) {", "  { int* p = new(int); *p = i; }", "  i++;", " }", " return 0;", "}", "fn stable() { return 73; }", "fn main() {", " let base = live_objects();"]
+for _ in range(128):
+    source, value = expr(4)
+    lines.append(f" put_n({source}); put_c(32);")
+    values.append(value)
+for _ in range(32):
+    a = "".join(rng.choice("abc") for _ in range(rng.randrange(11)))
+    b = "".join(rng.choice("abc") for _ in range(rng.randrange(6)))
+    lines.append(f' put_n(Text.len("{a}")); put_c(32); put_n(Text.eq("{a}", "{b}")); put_c(32);')
+    lines.append(f' put_n(Text.starts("{a}", "{b}")); put_c(32); put_n(Text.ends("{a}", "{b}")); put_c(32); put_n(Text.find("{a}", "{b}")); put_c(32);')
+    lines.append(f' put_n(Bytes.len(b"{a}")); put_c(32); put_n(Bytes.eq(b"{a}", b"{b}")); put_c(32);')
+    lines.append(f' put_n(Bytes.starts(b"{a}", b"{b}")); put_c(32); put_n(Bytes.ends(b"{a}", b"{b}")); put_c(32); put_n(Bytes.find(b"{a}", b"{b}")); put_c(32);')
+    library_values += [len(a), int(a == b), int(a.startswith(b)), int(a.endswith(b)), a.find(b), len(a.encode()), int(a == b), int(a.startswith(b)), int(a.endswith(b)), a.find(b)]
+lines += [
+    " churn(300); gc(); put_n(live_objects() - base); put_c(32);",
+    ' char* source = "fn f() { return 7; }"; let parsed = tokens(source);',
+    ' put_n(parsed[0] > 0); put_c(32); put_n(eq_w(source, "fn f() { return 7; }")); put_c(32);',
+    ' char* eval_source = "return 42;"; int e = 0;',
+    ' while (e < 16) { let q = tokens(eval_source); put_n(eval(q)); put_c(32); e++; }',
+    ' put_n(stable()); put_c(32); put_n(eq_w(eval_source, "return 42;")); put_c(10);',
+    " return 0;",
+    "}",
+]
+program = "\n".join(lines) + "\n"
+expected = " ".join(str(x) for x in values + library_values + [0, 1, 1, *([42] * 16), 73, 1]) + "\n"
+
+
+def run(binary):
+    result = subprocess.run(binary, input=program, text=True, capture_output=True, timeout=30)
+    if result.returncode or result.stdout != expected:
+        actual_words = result.stdout.split()
+        expected_words = expected.split()
+        mismatch = next((i for i, pair in enumerate(zip(actual_words, expected_words)) if pair[0] != pair[1]), min(len(actual_words), len(expected_words)))
+        raise SystemExit(
+            f"{' '.join(binary)} failed deterministic fuzz seed 0x5a17\n"
+            f"exit={result.returncode}, token mismatch at {mismatch}\n"
+            f"expected={expected_words[max(0, mismatch - 3):mismatch + 4]!r}\n"
+            f"stdout={actual_words[max(0, mismatch - 3):mismatch + 4]!r}\n"
+            f"stderr={result.stderr!r}"
+        )
+    return result.stdout
+
+
+hosts = [(zy, [zy]), (zym, [zym])]
+if js:
+    hosts.append(("javascript", ["node", "src/js/tinyvm.js"]))
+outputs = [run(command) for _, command in hosts]
+if any(output != outputs[0] for output in outputs[1:]):
+    raise SystemExit("compiler hosts disagreed on deterministic fuzz output")
+print("ok   compiler/runtime/stdlib fuzz: 128 generated expressions, 300 scoped allocations and GC, "
+      "16 eval recompiles, unchanged source strings, and 32 randomized Text/Bytes cases on "
+      + ", ".join(name for name, _ in hosts))