"""Summarize one run's probe.csv (and routerping.csv, if present) between two HH:MM:SS times.

Same phase rule, idle baseline and percentile as video #1's analyze.py (so medians and p95 match it exactly),
plus what video #2's plan needs: the worst packet, packets over +100 ms, and the router ping per phase.
Usage: python analyze_v2.py START END [probe.csv] [routerping.csv]
"""
import csv, statistics, sys

start, end = sys.argv[1], sys.argv[2]
probe = sys.argv[3] if len(sys.argv) > 3 else 'probe.csv'
router = sys.argv[4] if len(sys.argv) > 4 else None


def secs(ts):
    return int(ts[:2]) * 3600 + int(ts[3:5]) * 60 + float(ts[6:])


rows = [r for r in csv.DictReader(open(probe)) if start <= r['ts'][:8] <= end]
for r in rows:
    for k in ('udp_ms', 'icmp_ms', 'tx_mbps', 'rx_mbps'):
        r[k] = float(r[k])


def phase(r):  # analyze.py's rule
    if r['tx_mbps'] > 100 and r['tx_mbps'] > r['rx_mbps']: return 'upload'
    if r['rx_mbps'] > 100 and r['rx_mbps'] > r['tx_mbps']: return 'download'
    if r['tx_mbps'] < 5 and r['rx_mbps'] < 5: return 'idle'
    return 'other'


def pct(v, p):  # analyze.py's percentile
    v = sorted(v); return v[min(len(v) - 1, int(round(p / 100 * (len(v) - 1))))]


times = [secs(r['ts']) for r in rows]
gap = max((b - a for a, b in zip(times, times[1:])), default=0)
print(f"samples {len(rows)}  max gap {gap:.2f}s")
idle_udp = [r['udp_ms'] for r in rows if phase(r) == 'idle' and r['udp_ms'] > 0]
base_udp = statistics.median(idle_udp) if idle_udp else float('nan')
idle_icmp = [r['icmp_ms'] for r in rows if phase(r) == 'idle' and r['icmp_ms'] > 0]
base_icmp = statistics.median(idle_icmp) if idle_icmp else float('nan')
print(f"idle baseline: udp {base_udp:.1f}ms (n={len(idle_udp)})  icmp {base_icmp:.1f}ms")

gw = []
if router:
    gw = [(secs(r['ts']), float(r['gw_ms'])) for r in csv.DictReader(open(router)) if start <= r['ts'][:8] <= end]
    targets = {r['target'] for r in csv.DictReader(open(router))}
    gidle = [ms for t, ms in gw if ms >= 0 and any(abs(t - pt) < 0.25 and phase(pr) == 'idle' for pt, pr in zip(times, rows))]
    print(f"router ping target {', '.join(sorted(targets))}: {len(gw)} samples, idle median "
          f"{statistics.median(gidle) if gidle else float('nan'):.1f}ms")

for ph in ('download', 'upload'):
    sel = [r for r in rows if phase(r) == ph]
    u = [r['udp_ms'] for r in sel if r['udp_ms'] > 0]; lost_u = sum(1 for r in sel if r['udp_ms'] < 0)
    if not u:
        print(f"{ph:8s} n={len(sel)} (no valid samples)")
        continue
    mb = statistics.median([r['tx_mbps' if ph == 'upload' else 'rx_mbps'] for r in sel])
    i = [r['icmp_ms'] for r in sel if r['icmp_ms'] >= 0]
    over = sum(1 for x in u if x - base_udp > 100)
    line = (f"{ph:8s} n={len(sel)} ~{mb:.0f} Mbps | UDP added: median +{statistics.median(u) - base_udp:.0f}ms "
            f"p95 +{pct(u, 95) - base_udp:.0f}ms worst +{max(u) - base_udp:.0f}ms over+100 {over}/{len(u)} timeouts {lost_u}")
    if i:
        line += f" | ICMP added: median +{statistics.median(i) - base_icmp:.0f}ms p95 +{pct(i, 95) - base_icmp:.0f}ms"
    if gw:
        st = {round(secs(r['ts']), 1) for r in sel}
        g = [ms for t, ms in gw if ms >= 0 and any(abs(t - s) < 0.25 for s in st)]
        lost_g = sum(1 for t, ms in gw if ms < 0 and any(abs(t - s) < 0.25 for s in st))
        if g:
            line += f" | router ping: median {statistics.median(g):.0f}ms p95 {pct(g, 95):.0f}ms worst {max(g):.0f}ms lost {lost_g}"
    print(line)