"""Per day and per target: what the daytime curves keep, outside the brief moments. For each day D (04:00-16:00 UTC), each target, each column: - night level: median of accepted measurements from D-1 17:00 to D 04:00 (evening + night before), - 'clean' instants: those where the compensation keeps all its references except R3 (5 or 6 retained), that no label of --exclude covers (brief labels: passages, reference losses), accepted; - wave: max |median-of-5 smoothed (clean) - night level| in noises, and its instant; - agit: max |v - median of 4 clean neighbours each side| in noises (the fast part), and the count of clean points > 5; - night agit, same on the night measurements, for comparison. Usage: python3 days_dev.py EXCLUDE.jsonl [--maxdur 90] """ import json, sys, statistics from datetime import timedelta sys.path.insert(0, "tools/scripts") import data S, E = data.parse_time("2026-09-23T17:00Z"), data.parse_time("2026-10-06T19:45Z") by = data.load("data/WILDSTRUBEL") refs = [s for s, ms in by.items() if data.is_reference(ms)] tg = [s for s, ms in by.items() if not data.is_reference(ms)] maxdur = int(sys.argv[sys.argv.index("--maxdur") + 1]) if "--maxdur" in sys.argv else 90 excl = [] for l in open(sys.argv[1]): L = json.loads(l) a, b = data.parse_time(L["start"]), data.parse_time(L["end"]) if (b - a) <= timedelta(minutes=maxdur) or L.get("_force"): excl.append((a, b)) # instants with refs other than R3 out bad = set() reft = {} for r in refs: for m in by[r]: if S <= m["t"] <= E and r != "R3" and (data.rejected(m) or m.get("inlier") in (0, None)): bad.add(m["t"]) def covered(t): return any(a < t < b for a, b in excl) noise = {} W0, W1 = data.parse_time("2026-09-24T02:30Z"), E for s in tg: for c in ("corr_x", "corr_y"): v = [m[c] for m in by[s] if W0 <= m["t"] <= W1 and m[c] is not None] noise[s, c] = statistics.median(abs(b - a) for a, b in zip(v, v[1:])) def agit(ser, nz): out = [] for i, (t, v) in enumerate(ser): nb = [x for _, x in ser[max(0, i - 4):i] + ser[i + 1:i + 5]] if len(nb) >= 4: out.append((abs(v - statistics.median(nb)) / nz, t)) return out day = data.parse_time("2026-09-24T00:00Z") while day < data.parse_time("2026-10-07T00:00Z"): print(f"=== {day:%m-%d}") for s in tg: for c in ("corr_x", "corr_y"): nz = noise[s, c] ms = [m for m in by[s] if m[c] is not None] night = [m[c] for m in ms if day - timedelta(hours=7) <= m["t"] < day + timedelta(hours=4)] dser = [(m["t"], m[c]) for m in ms if day + timedelta(hours=4) <= m["t"] < day + timedelta(hours=16) and m["t"] not in bad and not covered(m["t"])] nser = [(m["t"], m[c]) for m in ms if day - timedelta(hours=7) <= m["t"] < day + timedelta(hours=4) and m["t"] not in bad and not covered(m["t"])] if len(night) < 3 or len(dser) < 5: lvl = None else: lvl = statistics.median(night) wave = "" if lvl is not None: sm = [] for i in range(len(dser)): w = [v for _, v in dser[max(0, i - 2):i + 3]] sm.append((statistics.median(w) - lvl, dser[i][0])) mx = max(sm, key=lambda x: abs(x[0])) over = [t for d, t in sm if abs(d) / nz > 5] wave = f"wave {mx[0]/nz:+6.1f} @{mx[1]:%H:%M}" + (f" >5: {over[0]:%H:%M}-{over[-1]:%H:%M} ({len(over)})" if over else "") ag = agit(dser, nz) na = agit(nser, nz) agd = max(ag, default=(0, None)) n5 = sum(1 for a, _ in ag if a > 5) agn = max(na, default=(0, None)) print(f" {s:6s}{c[-1]} {wave:42s} agit day max {agd[0]:5.1f}" f"{'@'+format(agd[1],'%H:%M') if agd[1] else '':7s} n>5 {n5:2d}/{len(ag):2d} night max {agn[0]:4.1f} ({len(na)})") day += timedelta(days=1)