#!/usr/bin/env python3 -I
# usage: analyse_timing.py round.csv [round2.csv ...]
import csv, sys, random
import numpy as np
rng = np.random.default_rng(7)
rows = []
for p in sys.argv[1:]:
    with open(p) as f:
        for r in csv.DictReader(f):
            rows.append({**r, "src": p})
def col(kind, host, key, f=lambda r: True):
    return [float(r[key]) for r in rows if r["kind"] == kind and r["host"] == host and f(r)]
def pairs(kind, fn):
    d = {}
    for r in rows:
        d.setdefault((r["src"], r["iter"], r["kind"]), {})[r["host"]] = fn(r)
    return [(v["vercel"], v["cloudflare"]) for (s, i, k), v in d.items() if k == kind and len(v) == 2]
def ci(diffs, stat=np.median, n=10000):
    a = np.array(diffs)
    boots = [stat(rng.choice(a, len(a))) for _ in range(n)]
    return stat(a), np.percentile(boots, 2.5), np.percentile(boots, 97.5)
def q(a, p): return np.percentile(a, p)
ms = lambda s: s * 1000
for kind in ("home", "post", "avif"):
    print(f"\n=== {kind} ===")
    for label, fn in (("TTFB excl DNS (P1)", lambda r: ms(float(r["ttfb"]) - float(r["dns"]))),
                      ("TCP connect", lambda r: ms(float(r["connect"]) - float(r["dns"]))),
                      ("TLS handshake", lambda r: ms(float(r["tls"]) - float(r["connect"]))),
                      ("server wait (ttfb-tls)", lambda r: ms(float(r["ttfb"]) - float(r["tls"]))),
                      ("total", lambda r: ms(float(r["total"]) - float(r["dns"])))):
        ps = pairs(kind, fn)
        if not ps: continue
        v = np.array([p[0] for p in ps]); c = np.array([p[1] for p in ps])
        m, lo, hi = ci(v - c)
        print(f"{label:24s} n={len(ps):3d}  vercel med {np.median(v):7.1f} p90 {q(v,90):7.1f} p99 {q(v,99):7.1f} | "
              f"cloudflare med {np.median(c):7.1f} p90 {q(c,90):7.1f} p99 {q(c,99):7.1f} | "
              f"median(V-C) {m:7.1f} [{lo:7.1f}, {hi:7.1f}]  CF faster in {np.mean(v>c)*100:3.0f}% of pairs")
    for h in ("vercel", "cloudflare"):
        caches = {}
        for r in rows:
            if r["kind"] == kind and r["host"] == h: caches[r["cache"]] = caches.get(r["cache"], 0) + 1
        codes = {}
        for r in rows:
            if r["kind"] == kind and r["host"] == h: codes[r["code"]] = codes.get(r["code"], 0) + 1
        ips = sorted({r["ip"] for r in rows if r["kind"] == kind and r["host"] == h})
        print(f"  {h}: cache {caches} codes {codes} ips {len(ips)} http {sorted({r['http'] for r in rows if r['host']==h})}")