#!/usr/bin/env python3
"""Compare the Go tracker's /live (and /trails) against the live Python service.

Run on the server (or anywhere with loopback access to both):
    python3 compare_live.py            # default loopback ports
    python3 compare_live.py --py http://127.0.0.1:8765 --go http://127.0.0.1:8770

Parity strategy for a live firehose
-----------------------------------
The two services rebuild their /live cache independently every 5s, so an
actively-updating character can legitimately show a newer telemetry row in one
than the other. We separate "is this a real divergence?" from "is this just
cache timing?" using the server-stamped received_at:

  * SAME ROW  (py.received_at == go.received_at): both rendered the *same*
    telemetry_events row -> every field MUST match (numbers within epsilon,
    timestamps compared as instants). This is the rigorous render-parity proof.
  * DIFFERENT ROW: a newer row arrived between the two refreshes -> we only
    require identity + key-set + type/null-pattern parity, and report the
    volatile-field skew (which should be small and recent).

Exit code 0 if no real parity violations, 1 otherwise.
"""
import argparse
import json
import sys
import urllib.request
from datetime import datetime, timezone

EPS = 1e-6

# Fields that identify the entity / join keys — must always match for a player
# present in both outputs.
IDENTITY = ("character_name", "char_tag", "session_id")
# Slowly-changing aggregates — informational when they differ on a same-row pair
# (a kill/rare recorded between refreshes can bump these even for the same
# telemetry row).
AGGREGATES = ("total_kills", "total_rares", "session_rares")
TIMESTAMP_FIELDS = ("timestamp", "received_at")


def fetch(base, path):
    with urllib.request.urlopen(base.rstrip("/") + path, timeout=8) as r:
        return json.load(r)


def jtype(v):
    if v is None:
        return "null"
    if isinstance(v, bool):
        return "bool"
    if isinstance(v, (int, float)):
        return "num"
    if isinstance(v, str):
        return "str"
    return type(v).__name__


def parse_ts(s):
    if s is None:
        return None
    return datetime.fromisoformat(s.replace("Z", "+00:00"))


def values_equal(key, a, b):
    """Semantic equality for a single field value."""
    if a is None or b is None:
        return a is b or a == b
    if key in TIMESTAMP_FIELDS and isinstance(a, str) and isinstance(b, str):
        return parse_ts(a) == parse_ts(b)
    an, bn = isinstance(a, (int, float)) and not isinstance(a, bool), isinstance(b, (int, float)) and not isinstance(b, bool)
    if an and bn:
        return abs(float(a) - float(b)) <= EPS
    return a == b


def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--py", default="http://127.0.0.1:8765")
    ap.add_argument("--go", default="http://127.0.0.1:8770")
    args = ap.parse_args()

    py = fetch(args.py, "/live")["players"]
    go = fetch(args.go, "/live")["players"]
    now = datetime.now(timezone.utc)

    pyi = {p["character_name"]: p for p in py}
    goi = {p["character_name"]: p for p in go}
    common = sorted(set(pyi) & set(goi))
    only_py = sorted(set(pyi) - set(goi))
    only_go = sorted(set(goi) - set(pyi))

    print("=" * 72)
    print("/live PARITY  python(%s) vs go(%s)" % (args.py, args.go))
    print("=" * 72)
    print(f"python players : {len(py)}")
    print(f"go     players : {len(go)}")
    print(f"common         : {len(common)}")

    violations = 0

    # --- key-set parity (all players) ---
    py_keys = set().union(*[set(p) for p in py]) if py else set()
    go_keys = set().union(*[set(p) for p in go]) if go else set()
    if py_keys == go_keys:
        print(f"key set        : IDENTICAL ({len(py_keys)} keys)")
    else:
        violations += 1
        print("key set        : MISMATCH")
        print("   only in python:", sorted(py_keys - go_keys))
        print("   only in go    :", sorted(go_keys - py_keys))

    # --- online-set parity (boundary-aware) ---
    def age(p):
        ts = parse_ts(p.get("received_at") or p.get("timestamp"))
        return (now - ts).total_seconds() if ts else None

    print("\n-- online set --")
    if not only_py and not only_go:
        print("online set     : IDENTICAL")
    else:
        # Players near the 30s boundary can flap between the two refreshes.
        def explain(names, idx):
            for n in names:
                a = age(idx[n])
                tag = "boundary-flap (age %.1fs)" % a if a is not None and 22 <= a <= 38 else "age %s" % (None if a is None else round(a, 1))
                print(f"   only_{('py' if idx is pyi else 'go')}: {n:<20} {tag}")
        if only_py:
            print(f"only in python : {len(only_py)}")
            explain(only_py, pyi)
        if only_go:
            print(f"only in go     : {len(only_go)}")
            explain(only_go, goi)
        unexplained = [n for n in (only_py + only_go)
                       if not (lambda a: a is not None and 22 <= a <= 38)(age((pyi.get(n) or goi.get(n))))]
        if unexplained:
            violations += 1
            print("   UNEXPLAINED set difference (not near 30s boundary):", unexplained)
        else:
            print("   (all set differences explained by the 30s online boundary)")

    # --- per-player field parity ---
    same_row = []          # py.received_at == go.received_at -> must fully match
    diff_row = []          # newer row arrived between refreshes
    for n in common:
        a, b = pyi[n], goi[n]
        if a.get("received_at") is not None and a.get("received_at") == b.get("received_at"):
            same_row.append(n)
        else:
            diff_row.append(n)

    print("\n-- per-player parity --")
    print(f"same-row pairs (identical received_at, must fully match): {len(same_row)}")
    print(f"diff-row pairs (newer telemetry between refreshes)      : {len(diff_row)}")

    # Identity + type/null-pattern parity across ALL common players.
    id_bad = type_bad = 0
    for n in common:
        a, b = pyi[n], goi[n]
        for k in IDENTITY:
            if a.get(k) != b.get(k):
                id_bad += 1
                print(f"   IDENTITY mismatch {n}.{k}: py={a.get(k)!r} go={b.get(k)!r}")
        for k in py_keys:
            ta, tb = jtype(a.get(k)), jtype(b.get(k))
            if ta != tb:
                # null vs num/str is a real null-pattern divergence; num-vs-num
                # whole-float (0.0) vs int (0) is already unified under "num".
                type_bad += 1
                print(f"   TYPE mismatch {n}.{k}: py={ta}({a.get(k)!r}) go={tb}({b.get(k)!r})")
    if id_bad:
        violations += id_bad
    if type_bad:
        violations += type_bad
    if not id_bad and not type_bad:
        print("identity+type  : IDENTICAL for all common players")

    # Rigorous: same-row pairs must match on every field.
    sr_full_match = 0
    for n in same_row:
        a, b = pyi[n], goi[n]
        diffs = []
        for k in py_keys:
            if not values_equal(k, a.get(k), b.get(k)):
                diffs.append((k, a.get(k), b.get(k)))
        if not diffs:
            sr_full_match += 1
        else:
            # Aggregate-only diffs are timing-explainable even on a same row.
            non_agg = [d for d in diffs if d[0] not in AGGREGATES]
            if non_agg:
                violations += 1
                print(f"   SAME-ROW FIELD divergence {n}: " +
                      ", ".join(f"{k}: py={pa!r} go={ga!r}" for k, pa, ga in non_agg))
            else:
                print(f"   (same-row {n}: only aggregate fields differ — kill/rare between refreshes: "
                      + ", ".join(f"{k} py={pa} go={ga}" for k, pa, ga in diffs) + ")")
    print(f"same-row full-field matches: {sr_full_match}/{len(same_row)}")

    # Volatile-field skew on diff-row pairs (informational).
    if diff_row:
        ts_deltas = []
        for n in diff_row:
            da, db = parse_ts(pyi[n].get("timestamp")), parse_ts(goi[n].get("timestamp"))
            if da and db:
                ts_deltas.append(abs((da - db).total_seconds()))
        if ts_deltas:
            ts_deltas.sort()
            print(f"diff-row timestamp skew: min={ts_deltas[0]:.1f}s "
                  f"median={ts_deltas[len(ts_deltas)//2]:.1f}s max={ts_deltas[-1]:.1f}s "
                  "(bounded by the two 5s refresh cycles)")

    print("\n" + "=" * 72)
    if violations == 0:
        print("RESULT: PARITY OK — no structural or same-row divergences.")
    else:
        print(f"RESULT: {violations} PARITY VIOLATION(S) — see above.")
    print("=" * 72)
    return 1 if violations else 0


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