dashboard-statistics-lambda/src/main/java/com/dashboard/aws/lambda/service/MeasureAverageService.java

package com.dashboard.aws.lambda.service;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import java.sql.*;
import java.time.*;
import java.util.*;

public class MeasureAverageService {

    private static final Logger logger = LoggerFactory.getLogger(MeasureAverageService.class);

    private static final String MYSQL_URL = System.getenv("DB_URL");
    private static final String DB_USER = System.getenv("DB_USER");
    private static final String DB_PASSWORD = System.getenv("DB_PASSWORD");

    private Connection getConnection() throws SQLException {
        return DriverManager.getConnection(MYSQL_URL, DB_USER, DB_PASSWORD);
    }

    public void aggregateLastHour(Long companyId, ZoneId zone, LocalDateTime now, LocalDateTime start) {
        long startTs = start.atZone(zone).toInstant().toEpochMilli();
        long endTs = now.atZone(zone).toInstant().toEpochMilli();

        logger.info("Aggregating data for company {} from {} to {}", companyId, start, now);

        String schema = "data_center_dongjian_" + companyId;
        String realtimeTable = "dashboard_realtime_measure";
        String sourceTable = "dashboard_record_measure";
        String targetTable = "dashboard_aggregate_measure_hour";

        try (Connection conn = getConnection()) {

            // 1: 从实时表获取所有上报过的设备ID集合
            Set<String> deviceSet = new HashSet<>();
            String realtimeSql = String.format("SELECT device_id FROM %s.%s", schema, realtimeTable);
            try (Statement st = conn.createStatement();
                 ResultSet rs = st.executeQuery(realtimeSql)) {
                while (rs.next()) {
                    deviceSet.add(rs.getString("device_id"));
                }
            }
            logger.info("Found {} devices from realtime table", deviceSet.size());

            // 2: 查询过去一小时的数据
            Map<String, List<Record>> currentHourData = queryRecordData(conn, schema, sourceTable, startTs, endTs);

            // 3: 查询上一小时的聚合结果（用于回退）
            LocalDateTime prevHour = start.minusHours(1);
            Map<String, AggregateResult> prevHourData = queryPrevHourAggregates(conn, schema, targetTable, prevHour);

            // 4: 计算当前小时平均、最大、最小值
            Map<String, AggregateResult> resultMap = new HashMap<>();
            for (Map.Entry<String, List<Record>> entry : currentHourData.entrySet()) {
                String deviceId = entry.getKey();
                List<Record> records = entry.getValue();

                double avg = computeTimeWeightedAverage(records, startTs, endTs);
                double max = computeMaxValue(records);
                double min = computeMinValue(records);

                resultMap.put(deviceId, new AggregateResult(avg, max, min));
            }

            // 5: 对比集合 A，确保所有设备都有值（否则用上一小时或 0）
            for (String deviceId : deviceSet) {
                if (!resultMap.containsKey(deviceId)) {
                    AggregateResult prev = prevHourData.get(deviceId);
                    if (prev != null) {
                        resultMap.put(deviceId, prev);
                    } else {
                        resultMap.put(deviceId, new AggregateResult(0.0, 0.0, 0.0));
                    }
                }
            }

            // 6: 写入本小时聚合表
            String insertSql = String.format(
                    "INSERT INTO %s.%s (device_id, average_value, max_value, min_value, " +
                            "date_year, date_month, date_day, date_hour, time_start, time_end, aggregated_at) " +
                            "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
                    schema, targetTable
            );

            try (PreparedStatement ps = conn.prepareStatement(insertSql)) {
                for (Map.Entry<String, AggregateResult> entry : resultMap.entrySet()) {
                    String deviceId = entry.getKey();
                    AggregateResult r = entry.getValue();
                    LocalDateTime dt = start;

                    ps.setString(1, deviceId);
                    ps.setString(2, String.format("%.6f", r.avg));
                    ps.setString(3, String.format("%.6f", r.max));
                    ps.setString(4, String.format("%.6f", r.min));
                    ps.setInt(5, dt.getYear());
                    ps.setInt(6, dt.getMonthValue());
                    ps.setInt(7, dt.getDayOfMonth());
                    ps.setInt(8, dt.getHour());
                    ps.setLong(9, startTs);
                    ps.setLong(10, endTs);
                    ps.setTimestamp(11, Timestamp.valueOf(now));

                    ps.addBatch();
                }
                ps.executeBatch();
            }

            logger.info("Hourly aggregation finished for {} devices", resultMap.size());

        } catch (SQLException e) {
            logger.error("Hourly aggregation failed", e);
        }
    }

    // 查询当前小时数据
    private Map<String, List<Record>> queryRecordData(Connection conn, String schema, String table, long startTs, long endTs) throws SQLException {
        String sql = String.format(
                "SELECT device_id, upload_value, upload_at FROM %s.%s " +
                        "WHERE upload_at >= ? AND upload_at < ? ORDER BY device_id, upload_at",
                schema, table
        );
        Map<String, List<Record>> map = new HashMap<>();
        try (PreparedStatement ps = conn.prepareStatement(sql)) {
            ps.setLong(1, startTs);
            ps.setLong(2, endTs);
            ResultSet rs = ps.executeQuery();
            while (rs.next()) {
                String deviceId = rs.getString("device_id");
                String valStr = rs.getString("upload_value");
                long ts = rs.getLong("upload_at");
                double val = (valStr == null || valStr.isEmpty()) ? 0.0 : Double.parseDouble(valStr);
                map.computeIfAbsent(deviceId, k -> new ArrayList<>()).add(new Record(ts, val));
            }
        }
        return map;
    }

    // 查询上一小时聚合结果
    private Map<String, AggregateResult> queryPrevHourAggregates(Connection conn, String schema, String table, LocalDateTime prevHour) throws SQLException {
        Map<String, AggregateResult> map = new HashMap<>();
        String sql = String.format(
                "SELECT device_id, average_value, max_value, min_value FROM %s.%s " +
                        "WHERE date_year = ? AND date_month = ? AND date_day = ? AND date_hour = ?",
                schema, table
        );
        try (PreparedStatement ps = conn.prepareStatement(sql)) {
            ps.setInt(1, prevHour.getYear());
            ps.setInt(2, prevHour.getMonthValue());
            ps.setInt(3, prevHour.getDayOfMonth());
            ps.setInt(4, prevHour.getHour());
            ResultSet rs = ps.executeQuery();
            while (rs.next()) {
                map.put(rs.getString("device_id"),
                        new AggregateResult(
                                rs.getDouble("average_value"),
                                rs.getDouble("max_value"),
                                rs.getDouble("min_value")));
            }
        }
        return map;
    }

    // 时间加权平均
    private double computeTimeWeightedAverage(List<Record> records, long startTs, long endTs) {
        if (records == null || records.isEmpty()) return 0.0;

        double sum = 0.0;
        double totalMinutes = (endTs - startTs) / 60000.0; // 毫秒 → 分钟

        for (int i = 0; i < records.size(); i++) {
            Record rec = records.get(i);
            long nextTs = (i < records.size() - 1) ? records.get(i + 1).timestamp : endTs;
            double intervalMinutes = (nextTs - rec.timestamp) / 60000.0; // 转分钟
            sum += rec.value * intervalMinutes;
        }

        return totalMinutes > 0 ? sum / totalMinutes : 0.0;
    }

    // 最大值
    private double computeMaxValue(List<Record> records) {
        return records.stream().mapToDouble(r -> r.value).max().orElse(0.0);
    }

    // 最小值
    private double computeMinValue(List<Record> records) {
        return records.stream().mapToDouble(r -> r.value).min().orElse(0.0);
    }

    private static class Record {
        long timestamp;
        double value;

        Record(long timestamp, double value) {
            this.timestamp = timestamp;
            this.value = value;
        }
    }

    private static class AggregateResult {
        double avg;
        double max;
        double min;

        AggregateResult(double avg, double max, double min) {
            this.avg = avg;
            this.max = max;
            this.min = min;
        }
    }
}