import boto3, numpy as np
                from datetime import datetime
                timestream = boto3.client("timestream-query")
                sitewise = boto3.client("iotsitewise")
                sns = boto3.client("sns")
                REPLACEMENT_THRESHOLD_RATIO = 1.20   # 差壓達初始值 120% 視為應更換
                WARNING_DAYS_AHEAD = 7            # 提前幾天預警
                def get_baseline_and_recent(asset_id: str) -> tuple:
                    # baseline：濾網更換後首週平均值；recent：近14天數據
                    query = f"""
                        SELECT time, measure_value::double
                        FROM "cleanroom"."ffu_differential_pressure"
                        WHERE assetId = '{asset_id}'
                        AND time > ago(14d)
                        ORDER BY time ASC
                    """
                    result = timestream.query(QueryString=query)
                    values = [float(row["Value"]) for row in result["Rows"]]
                    baseline = sitewise.get_asset_property_value(
                        assetId=asset_id, propertyId="baseline_dp"
                    )["propertyValue"]["value"]["doubleValue"]
                    return baseline, values
                def predict_replacement_date(asset_id: str, ffu_name: str):
                    baseline, readings = get_baseline_and_recent(asset_id)
                    if len(readings) < 10:
                        return  # 資料量不足，暫不預測
                    threshold = baseline * REPLACEMENT_THRESHOLD_RATIO
                    x = np.arange(len(readings))
                    slope, intercept = np.polyfit(x, readings, 1)
                    if slope <= 0:
                        return  # 差壓沒有上升趨勢，濾網狀態穩定
                    current = readings[-1]
                    steps_to_threshold = (threshold - current) / slope
                    days_to_threshold = steps_to_threshold * (14 / len(readings))
                    if days_to_threshold <= WARNING_DAYS_AHEAD:
                        sns.publish(
                            TopicArn="arn:aws:sns:ap-northeast-1:xxx:ffu-maintenance",
                            Message=f"{ffu_name} 預估 {days_to_threshold:.1f} 天後達到更換閾值（目前 {current:.2f} Pa，基準 {baseline:.2f} Pa）"
                        )
                def handler(event, context):
                    ffu_assets = sitewise.list_assets(assetModelId="ffu-unit-model")["assetSummaries"]
                    for asset in ffu_assets:
                        predict_replacement_date(asset["id"], asset["name"])

import boto3
                sitewise = boto3.client("iotsitewise")
                sns = boto3.client("sns")
                DROP_THRESHOLD_RATIO = 0.85  # 低於基準值 85% 視為異常下降
                WIDESPREAD_RATIO = 0.6       # 60% 以上監測點同時異常 → 判定為全區性
                def classify_fault_source(ahu_zone_id: str) -> dict:
                    monitoring_points = sitewise.list_associated_assets(
                        assetId=ahu_zone_id, hierarchyId="monitoring-points"
                    )["assetSummaries"]
                    abnormal_points = []
                    for point in monitoring_points:
                        current = sitewise.get_asset_property_value(
                            assetId=point["id"], propertyId="differential_pressure"
                        )["propertyValue"]["value"]["doubleValue"]
                        baseline = sitewise.get_asset_property_value(
                            assetId=point["id"], propertyId="baseline_dp"
                        )["propertyValue"]["value"]["doubleValue"]
                        if current < baseline * DROP_THRESHOLD_RATIO:
                            abnormal_points.append(point["name"])
                    ratio = len(abnormal_points) / len(monitoring_points)
                    if ratio == 0:
                        return {"status": "normal"}
                    elif ratio >= WIDESPREAD_RATIO:
                        return {
                            "status": "abnormal",
                            "likely_source": "HVAC（全區性下降）",
                            "affected_points": abnormal_points,
                            "recommendation": "檢查 AHU 供氣機狀態與備用系統切換"
                        }
                    else:
                        return {
                            "status": "abnormal",
                            "likely_source": "FFU（局部性下降）",
                            "affected_points": abnormal_points,
                            "recommendation": f"檢查以下 FFU 單元的電源與濾網: {abnormal_points}"
                        }
                def handler(event, context):
                    result = classify_fault_source(event["ahu_zone_id"])
                    if result["status"] == "abnormal":
                        sns.publish(
                            TopicArn="arn:aws:sns:ap-northeast-1:xxx:cleanroom-fault",
                            Message=f"疑似故障來源: {result['likely_source']}\n建議動作: {result['recommendation']}"
                        )
                    return result