"""Rebuild the Canada 2026-11-01 container census classes and headline counts from census.json.

    python3 rebuild.py census.json [--check]

Re-derives every scanned row's class from its recorded tzscan copies and recounts the headline
numbers for D_maint (tag pushed on or after 2026-07-16) and D_all (all pinned images).
--check exits 1 if any class or count differs from the file's own values.

To re-measure one row instead of trusting the recorded copies (needs crane and tzscan):
    crane export --platform linux/amd64 <image> - | python3 tzscan.py --tar - --json
"""
import collections, json, math, sys

CANADA = ("America/Vancouver", "America/Edmonton", "America/Inuvik", "America/Winnipeg")


def classify(r):
    copies = r["copies"]
    kinds = r["tzscanProblemKinds"]
    cw = [c for c in copies if set(c.get("wrong") or []) & set(CANADA)]
    if any(c["tier"] == "A" for c in cw):
        return "WRONG_DEFAULT"
    if cw:
        return "WRONG_LIBRARY_ONLY"
    if not copies:
        return "NO_TZ_DATA" if all(k == "nothing-judged" for k in kinds) else "UNKNOWN"
    if r["tzscanComplete"] and not kinds and all(c["verdict"] == "OK" for c in copies):
        return "CURRENT"
    return "UNKNOWN"


def wilson(k, n, z=1.96):
    if n == 0:
        return None
    p = k / n
    d = 1 + z * z / n
    c = (p + z * z / (2 * n)) / d
    h = z * math.sqrt(p * (1 - p) / n + z * z / (4 * n * n)) / d
    return [round(c - h, 4), round(c + h, 4)]


def counts(sub):
    cnt = collections.Counter(r["class"] for r in sub)
    scanned = sum(v for k, v in cnt.items() if k != "NOT_SCANNED")
    decidable = scanned - cnt["UNKNOWN"]
    h1, h2 = cnt["WRONG_DEFAULT"], cnt["WRONG_DEFAULT"] + cnt["WRONG_LIBRARY_ONLY"]
    out = {"n": len(sub), "scanned": scanned, "notScanned": cnt["NOT_SCANNED"], "classes": dict(sorted(cnt.items()))}
    for name, den in (("scanned", scanned), ("decidable", decidable)):
        out[f"H1_{name}"] = {"k": h1, "n": den, "rate": round(h1 / den, 4) if den else None, "wilson95": wilson(h1, den)}
        out[f"H2_{name}"] = {"k": h2, "n": den, "rate": round(h2 / den, 4) if den else None, "wilson95": wilson(h2, den)}
    return out


def main():
    data = json.load(open(sys.argv[1], encoding="utf-8"))
    rows = data["rows"]
    bad = []
    for r in rows:
        if r["class"] in ("EXCLUDED", "NOT_SCANNED"):
            continue
        c = classify(r)
        if c != r["class"]:
            bad.append((r["image"], r["class"], c))
    pinned = [r for r in rows if r["class"] != "EXCLUDED"]
    got = {"D_maint": counts([r for r in pinned if r["maintained"]]), "D_all": counts(pinned)}
    for lbl, v in got.items():
        print(f"{lbl}: n={v['n']} scanned={v['scanned']} not_scanned={v['notScanned']} classes={v['classes']}")
        for h in ("H1_scanned", "H2_scanned"):
            print(f"  {h}: {v[h]['k']}/{v[h]['n']} = {v[h]['rate']} (Wilson 95% {v[h]['wilson95']})")
    print(f"rows: {len(rows)} (pinned {len(pinned)}, excluded {len(rows) - len(pinned)}); class mismatches: {len(bad)}")
    for b in bad[:20]:
        print("  mismatch", *b)
    if "--check" in sys.argv:
        want = data["meta"]["headline"]
        diff = [lbl for lbl in got if json.dumps(got[lbl], sort_keys=True) != json.dumps(want[lbl], sort_keys=True)]
        if bad or diff:
            print(f"CHECK FAILED: {len(bad)} class mismatches, headline differs for {diff}")
            return 1
        print("CHECK OK: classes and headline counts match the file")
    return 0


if __name__ == "__main__":
    sys.exit(main())