"""Per instant: targets whose corr_x / corr_y jump against their local level (median of the 3 measurements on each side, without the point), rejected sensors outside the always-rejected ones, and the labels covering the instant. Usage: python3 scan.py START END [K]""" import json, statistics, sys sys.path.insert(0, "tools/scripts") import data A = "labeling/auto/analyses/NIEDERHORN_2026-10-11T0500_2026-10-21T1200/work/" start, end = data.parse_time(sys.argv[1]), data.parse_time(sys.argv[2]) K = float(sys.argv[3]) if len(sys.argv) > 3 else 5 ALWAYS = {"e03", "e04", "e05", "e07", "e08", "e11", "e12"} by = data.load("data/NIEDERHORN") labels = [json.loads(l) for f in ("existing.jsonl", "spanning.jsonl") for l in open(A + f)] for l in labels: l["s"], l["e"] = data.parse_time(l["start"]), data.parse_time(l["end"]) dev = {} # t -> list of strings rej = {} for s, ms in by.items(): if s in ALWAYS: continue ms = [m for m in ms if start - __import__("datetime").timedelta(hours=6) <= m["t"] <= end] for c in ("corr_x", "corr_y"): v = [(m["t"], m[c]) for m in ms if m[c] is not None] if len(v) < 8: continue diffs = [abs(v[i + 1][1] - v[i][1]) for i in range(len(v) - 1)] noise = statistics.median(diffs) or 1e-9 for i, (t, x) in enumerate(v): nb = [y for _, y in v[max(0, i - 3):i] + v[i + 1:i + 4]] d = (x - statistics.median(nb)) / noise if abs(d) >= K and start <= t <= end: dev.setdefault(t, []).append(f"{s.replace('PWD_','').replace('_C09','')}:{c[-1]}{d:+.0f}") for m in ms: if data.rejected(m) and start <= m["t"] <= end: rej.setdefault(m["t"], []).append(s.replace("PWD_", "").replace("_C09", "")) for t in sorted(set(dev) | set(rej)): cov = [l["id"].split("-")[1] + l["family"][:3] for l in labels if l["s"] <= t <= l["e"]] print(data.iso(t), "| rej:", ",".join(rej.get(t, [])), "| dev:", " ".join(dev.get(t, [])), "| lab:", ",".join(cov))