beam: verify_toposort_gen probe for one-shot iterator materialisation

wizard-work/beam/verify_toposort_gen.py

@@ -0,0 +1,63 @@
+"""Verification: toposort materialises one-shot iterators exactly once.
+
+Covers the defect Q61 measured: a dependency delivered as a generator must
+produce the same ordering/cycle result as the equivalent list, including
+when the graph contains a cycle. Also asserts the iterator is consumed
+exactly once (call counter).
+"""
+import os, sys
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+from toposort import topo_sort
+
+
+def _check(cond, msg="check failed"):
+    if not cond:
+        raise SystemExit(f"FAIL: {msg}")
+
+
+class Counter:
+    def __init__(self):
+        self.n = 0
+
+    def gen(self):
+        self.n += 1
+        yield "b"
+
+
+# 1. generator dep equals list dep (acyclic)
+c = Counter()
+r_gen = topo_sort({"a": c.gen()})
+r_list = topo_sort({"a": ["b"]})
+_check(r_gen.sorted == r_list.sorted == ["b", "a"],
+       f"acyclic mismatch: gen={r_gen} list={r_list}")
+# materialised exactly once
+_check(c.n == 1, f"iterator consumed {c.n} times, expected 1")
+
+# 2. None deps still work
+_check(topo_sort({"a": None}).sorted == ["a"])
+_check(topo_sort({"a": ["z"], "z": None}).sorted == ["z", "a"])
+
+# 3. cycle delivered as generators must be reported as a cycle (not ordered)
+def g_ab():
+    yield "b"
+def g_bc():
+    yield "c"
+def g_ca():
+    yield "a"
+r_cyc_gen = topo_sort({"a": g_ab(), "b": g_bc(), "c": g_ca()})
+r_cyc_list = topo_sort({"a": ["b"], "b": ["c"], "c": ["a"]})
+_check(r_cyc_gen.sorted is None and r_cyc_list.sorted is None,
+       f"cycle not detected: gen={r_cyc_gen} list={r_cyc_list}")
+_check(set(r_cyc_gen.cycle) == set(r_cyc_list.cycle) == {"a", "b", "c"},
+       f"cycle mismatch: gen={r_cyc_gen} list={r_cyc_list}")
+
+# 4. self-loop
+_check(topo_sort({"a": ["a"]}).cycle == ["a"])
+
+# 5. multi-form equivalence: filter/map objects
+r_filter = topo_sort({"a": filter(lambda x: True, ["b"])})
+r_map = topo_sort({"a": map(str, ["b"])})
+_check(r_filter.sorted == r_map.sorted == ["b", "a"],
+       f"filter/map mismatch: {r_filter} {r_map}")
+
+print("ALL VERIFIED")