cangku/backend/scripts/backfillPickupHistory.js

#!/usr/bin/env node

require('dotenv').config();

const dayjs = require('dayjs');

const database = require('../src/config/database');
const cloudDb = require('../src/config/cloudDatabase');
const redis = require('../src/config/redis');
const { refreshAllCache } = require('../src/jobs/dailyRefresh');

const localDb = database.local;

const PICKUP_EXCLUDED_LABEL_KEYWORDS = [
  '\u914d\u4ef6',
  '\u7ecf\u7ba1\u5927\u5355',
  '\u4e0d\u53d1',
  '\u7b49\u901a\u77e5',
  '\u5ef6\u671f',
  '\u5ef6\u8fdf',
  '\u6b66\u8fdb\u5b89\u80fd',
  '\u4e2d\u94c1',
  '\u7269\u6d41',
  '\u81ea\u63d0',
  '\u5916\u8d38',
  '\u91c7\u8d2d\u5355',
  '\u5907\u8d27',
  '\u5b9a\u5236',
  '\u4e2d\u8f6c\u96c6\u8fd0',
  '\u4e2d\u8f6c\u8ba2\u5355',
  '\u65b0\u7586\u897f\u85cf\u5355',
  'ECLP\u8ba2\u5355',
];

const PICKUP_EXCLUDED_COURIER_KEYWORDS = [
  '\u6b66\u8fdb\u5b89\u80fd',
  '\u4e2d\u94c1',
  '\u7269\u6d41',
  '\u81ea\u63d0',
  '\u7ebf\u4e0b\u5feb\u9012',
  '\u4e13\u7ebf',
];

const PICKUP_EXCLUDED_REMARK_KEYWORDS = [
  'bbff',
  'ccdd',
  '\u8f6cX00006',
];

const PICKUP_TIMELY_NO_FETCH_COURIER_KEYWORDS = [
  '\u4eac\u4e1c\u5feb\u8fd0',
  '\u4eac\u4e1c\u81ea\u8425',
  '\u987a\u4e30',
];

const PICKUP_TIMELY_LABEL_KEYWORDS = [
  'CLPS\u8ba2\u5355',
  'ECLP\u8ba2\u5355',
];

const UNKNOWN_WAREHOUSE = '\u672a\u77e5\u4ed3\u5e93';
const UNKNOWN_SUPPLIER = '\u672a\u77e5\u4f9b\u5e94\u5546';
const UNKNOWN_PROVINCE = '\u672a\u77e5\u7701\u4efd';
const UNKNOWN_COURIER = '\u672a\u77e5\u5feb\u9012';
const NO_AFTERSALE = '\u6682\u65e0\u552e\u540e';
const NO_REFUND_REQUEST = '\u672a\u7533\u8bf7\u9000\u6b3e';

const q = (value) => `N'${String(value).replace(/'/g, "''")}'`;

const buildSqlLikeAny = (column, keywords) => keywords
  .map((keyword) => `${column} LIKE ${q(`%${keyword}%`)}`)
  .join(' OR ');

const excludedLabelSql = buildSqlLikeAny(`ISNULL(src.order_labels, N'')`, PICKUP_EXCLUDED_LABEL_KEYWORDS);
const excludedCourierSql = buildSqlLikeAny(`ISNULL(src.courier_name, N'')`, PICKUP_EXCLUDED_COURIER_KEYWORDS);
const excludedRemarkSql = buildSqlLikeAny(
  `ISNULL(src.buyer_message, N'') + N' ' + ISNULL(src.remark, N'') + N' ' + ISNULL(src.remark_content, N'')`,
  PICKUP_EXCLUDED_REMARK_KEYWORDS
);
const timelyNoFetchCourierSql = buildSqlLikeAny(`ISNULL(src.courier_name, N'')`, PICKUP_TIMELY_NO_FETCH_COURIER_KEYWORDS);
const timelyLabelSql = buildSqlLikeAny(`ISNULL(src.order_labels, N'')`, PICKUP_TIMELY_LABEL_KEYWORDS);

const BACKFILL_PICKUP_SQL = `
SET NOCOUNT ON;

SELECT
  NULLIF(LTRIM(RTRIM(undefined1)), '') AS inner_order_id,
  NULLIF(LTRIM(RTRIM(undefined2)), '') AS order_id,
  NULLIF(LTRIM(RTRIM(undefined10)), '') AS sku_id,
  TRY_CONVERT(datetime, REPLACE(undefined5, '/', '-')) AS pay_time,
  TRY_CONVERT(datetime, REPLACE(undefined6, '/', '-')) AS afs_send_time,
  NULLIF(LTRIM(RTRIM(undefined8)), '') AS shop_name,
  NULLIF(LTRIM(RTRIM(undefined62)), '') AS order_labels,
  NULLIF(LTRIM(RTRIM(undefined72)), '') AS express_no,
  NULLIF(LTRIM(RTRIM(undefined73)), '') AS afs_courier_name,
  NULLIF(LTRIM(RTRIM(undefined74)), '') AS province,
  NULLIF(LTRIM(RTRIM(undefined75)), '') AS city,
  NULLIF(LTRIM(RTRIM(undefined66)), '') AS afs_warehouse_name,
  NULLIF(LTRIM(RTRIM(undefined32)), '') AS order_type,
  TRY_CONVERT(datetime, REPLACE(undefined33, '/', '-')) AS aftersale_apply_time
INTO #afs_raw
FROM dbo.jst_goods_profit_kpi_items_afs
WHERE undefined5 IS NOT NULL
  AND undefined5 LIKE @payDatePrefix;

CREATE NONCLUSTERED INDEX IX_afs_raw_dedup ON #afs_raw(inner_order_id, sku_id);

WITH ranked AS (
  SELECT
    *,
    ROW_NUMBER() OVER (
      PARTITION BY inner_order_id, sku_id
      ORDER BY
        CASE WHEN express_no IS NULL OR express_no = '' THEN 1 ELSE 0 END,
        CASE WHEN afs_send_time IS NULL THEN 1 ELSE 0 END,
        afs_send_time DESC,
        pay_time DESC
    ) AS rn
  FROM #afs_raw
  WHERE inner_order_id IS NOT NULL
    AND inner_order_id <> ''
    AND sku_id IS NOT NULL
    AND sku_id <> ''
)
SELECT
  inner_order_id,
  order_id,
  sku_id,
  pay_time,
  afs_send_time,
  shop_name,
  order_labels,
  express_no,
  afs_courier_name,
  province,
  city,
  afs_warehouse_name,
  order_type,
  aftersale_apply_time
INTO #afs_dedup
FROM ranked
WHERE rn = 1;

CREATE NONCLUSTERED INDEX IX_afs_dedup_order_id ON #afs_dedup(order_id);
CREATE NONCLUSTERED INDEX IX_afs_dedup_sku_id ON #afs_dedup(sku_id);

SELECT DISTINCT express_no
INTO #express_keys
FROM #afs_dedup
WHERE express_no IS NOT NULL
  AND express_no <> '';

SELECT DISTINCT order_id
INTO #order_keys
FROM #afs_dedup
WHERE order_id IS NOT NULL
  AND order_id <> '';

CREATE UNIQUE CLUSTERED INDEX IX_order_keys ON #order_keys(order_id);

SELECT DISTINCT sku_id
INTO #sku_keys
FROM #afs_dedup
WHERE sku_id IS NOT NULL
  AND sku_id <> '';

CREATE UNIQUE CLUSTERED INDEX IX_sku_keys ON #sku_keys(sku_id);

SELECT
  NULLIF(LTRIM(RTRIM(e.logistic_id)), '') AS logistic_id,
  MAX(NULLIF(LTRIM(RTRIM(e.so_id)), '')) AS so_id,
  MAX(NULLIF(LTRIM(RTRIM(e.logistic_company_name)), '')) AS logistic_company_name,
  MAX(NULLIF(LTRIM(RTRIM(e.order_status_name)), '')) AS order_status_name,
  MAX(NULLIF(LTRIM(RTRIM(e.aftersale_status_name)), '')) AS aftersale_status_name,
  MAX(NULLIF(LTRIM(RTRIM(e.remark)), '')) AS remark,
  MAX(NULLIF(LTRIM(RTRIM(e.buyer_message)), '')) AS buyer_message,
  MAX(NULLIF(LTRIM(RTRIM(e.remark_content)), '')) AS remark_content,
  MAX(NULLIF(LTRIM(RTRIM(e.shop_name)), '')) AS shop_name,
  MAX(NULLIF(LTRIM(RTRIM(e.warehouse_name)), '')) AS warehouse_name,
  MIN(e.send_time) AS send_time,
  MIN(e.fetch_time) AS fetch_time,
  MAX(NULLIF(LTRIM(RTRIM(e.receiver_address)), '')) AS receiver_address
INTO #express_by_logistic
FROM dbo.jst_order_express_details e
INNER JOIN #express_keys k
  ON k.express_no = NULLIF(LTRIM(RTRIM(e.logistic_id)), '')
GROUP BY NULLIF(LTRIM(RTRIM(e.logistic_id)), '');

CREATE UNIQUE CLUSTERED INDEX IX_express_by_logistic ON #express_by_logistic(logistic_id);

SELECT
  NULLIF(LTRIM(RTRIM(e.so_id)), '') AS so_id,
  MAX(NULLIF(LTRIM(RTRIM(e.logistic_id)), '')) AS logistic_id,
  MAX(NULLIF(LTRIM(RTRIM(e.logistic_company_name)), '')) AS logistic_company_name,
  MAX(NULLIF(LTRIM(RTRIM(e.order_status_name)), '')) AS order_status_name,
  MAX(NULLIF(LTRIM(RTRIM(e.aftersale_status_name)), '')) AS aftersale_status_name,
  MAX(NULLIF(LTRIM(RTRIM(e.remark)), '')) AS remark,
  MAX(NULLIF(LTRIM(RTRIM(e.buyer_message)), '')) AS buyer_message,
  MAX(NULLIF(LTRIM(RTRIM(e.remark_content)), '')) AS remark_content,
  MAX(NULLIF(LTRIM(RTRIM(e.shop_name)), '')) AS shop_name,
  MAX(NULLIF(LTRIM(RTRIM(e.warehouse_name)), '')) AS warehouse_name,
  MIN(e.send_time) AS send_time,
  MIN(e.fetch_time) AS fetch_time,
  MAX(NULLIF(LTRIM(RTRIM(e.receiver_address)), '')) AS receiver_address
INTO #express_by_order
FROM dbo.jst_order_express_details e
INNER JOIN #order_keys k
  ON k.order_id = NULLIF(LTRIM(RTRIM(e.so_id)), '')
GROUP BY NULLIF(LTRIM(RTRIM(e.so_id)), '');

CREATE UNIQUE CLUSTERED INDEX IX_express_by_order ON #express_by_order(so_id);

SELECT
  NULLIF(LTRIM(RTRIM(s.sku_id)), '') AS sku_id,
  MAX(NULLIF(LTRIM(RTRIM(s.supplier_name)), '')) AS supplier_name
INTO #sku_supplier
FROM dbo.jst_sku s
INNER JOIN #sku_keys k
  ON k.sku_id = NULLIF(LTRIM(RTRIM(s.sku_id)), '')
GROUP BY NULLIF(LTRIM(RTRIM(s.sku_id)), '');

CREATE UNIQUE CLUSTERED INDEX IX_sku_supplier ON #sku_supplier(sku_id);

SELECT
  d.inner_order_id,
  d.order_id,
  d.sku_id,
  d.pay_time,
  COALESCE(el.send_time, eo.send_time, d.afs_send_time) AS send_time,
  COALESCE(el.fetch_time, eo.fetch_time) AS fetch_time,
  COALESCE(el.shop_name, eo.shop_name, d.shop_name) AS shop_name,
  COALESCE(el.warehouse_name, eo.warehouse_name, d.afs_warehouse_name) AS warehouse_name,
  COALESCE(el.logistic_company_name, eo.logistic_company_name, d.afs_courier_name) AS courier_name,
  COALESCE(d.express_no, el.logistic_id, eo.logistic_id, N'') AS express_no,
  COALESCE(el.order_status_name, eo.order_status_name, N'') AS order_status_name,
  COALESCE(el.aftersale_status_name, eo.aftersale_status_name, N'') AS aftersale_status_name,
  COALESCE(el.remark, eo.remark, N'') AS remark,
  COALESCE(el.buyer_message, eo.buyer_message, N'') AS buyer_message,
  COALESCE(el.remark_content, eo.remark_content, N'') AS remark_content,
  COALESCE(d.order_labels, N'') AS order_labels,
  COALESCE(d.province, N'') AS province,
  COALESCE(d.city, N'') AS city,
  COALESCE(el.receiver_address, eo.receiver_address, N'') AS receiver_address,
  COALESCE(d.order_type, N'') AS order_type,
  d.aftersale_apply_time,
  COALESCE(s.supplier_name, N'') AS supplier_name,
  COALESCE(shop_meta.responsible_department, N'') AS shop_responsible_department
INTO #pickup_enriched
FROM #afs_dedup d
LEFT JOIN #express_by_logistic el ON d.express_no = el.logistic_id
LEFT JOIN #express_by_order eo ON d.order_id = eo.so_id
LEFT JOIN #sku_supplier s ON d.sku_id = s.sku_id
OUTER APPLY (
  SELECT TOP 1 NULLIF(LTRIM(RTRIM(responsibleDepartment)), N'') AS responsible_department
  FROM dbo.fys_shopInfo
  WHERE NULLIF(LTRIM(RTRIM(erpShopName)), N'') = COALESCE(el.shop_name, eo.shop_name, d.shop_name)
) shop_meta;

SELECT
  src.*,
  CASE
    WHEN (${excludedLabelSql}) THEN 1
    WHEN ISNULL(src.shop_responsible_department, N'') LIKE N'%灞变笢椤圭洰缁?%' THEN 1
    WHEN src.supplier_name IN (${q('\u865a\u62df')}, ${q('\u666f\u74f4\u5341\u65b9')}) THEN 1
    WHEN src.order_status_name IN (${q('\u88ab\u62c6\u5206')}, ${q('\u88ab\u5408\u5e76')}) THEN 1
    WHEN (${excludedCourierSql}) THEN 1
    WHEN ISNULL(src.express_no, N'') LIKE ${q('%F800%')} THEN 1
    WHEN (${excludedRemarkSql}) THEN 1
    WHEN ISNULL(src.shop_name, N'') LIKE ${q('%\u4eac\u559c%')}
         AND ISNULL(src.aftersale_status_name, ${q(NO_AFTERSALE)}) <> ${q(NO_AFTERSALE)} THEN 1
    WHEN ISNULL(src.shop_name, N'') LIKE ${q('%\u6837\u54c1\u4e13\u7528\u5e97\u94fa%')} THEN 1
    WHEN ISNULL(src.order_status_name, N'') LIKE ${q('%\u53d6\u6d88%')}
         AND src.send_time IS NULL
         AND ISNULL(src.aftersale_status_name, ${q(NO_AFTERSALE)}) IN (N'', ${q(NO_AFTERSALE)}, ${q(NO_REFUND_REQUEST)}) THEN 1
    WHEN src.aftersale_apply_time IS NOT NULL
         AND src.pay_time IS NOT NULL
         AND src.aftersale_apply_time >= src.pay_time
         AND DATEDIFF(SECOND, src.pay_time, src.aftersale_apply_time) <= 24 * 3600 THEN 1
    WHEN ISNULL(src.order_type, N'') LIKE ${q('%\u9884\u552e%')}
         OR ISNULL(src.order_status_name, N'') LIKE ${q('%\u9884\u552e%')} THEN 1
    ELSE 0
  END AS is_excluded,
  CASE
    WHEN src.pay_time IS NULL THEN 1
    WHEN ISNULL(src.shop_name, N'') LIKE ${q('%\u4eac\u559c%')}
         AND src.send_time IS NOT NULL
         AND src.send_time >= src.pay_time
         AND DATEDIFF(SECOND, src.pay_time, src.send_time) <= 24 * 3600 THEN 0
    WHEN (${timelyNoFetchCourierSql})
         AND src.fetch_time IS NULL
         AND src.send_time IS NOT NULL
         AND src.send_time >= src.pay_time
         AND DATEDIFF(SECOND, src.pay_time, src.send_time) <= 24 * 3600 THEN 0
    WHEN (${timelyLabelSql})
         AND src.send_time IS NOT NULL
         AND src.send_time >= src.pay_time
         AND DATEDIFF(SECOND, src.pay_time, src.send_time) <= 24 * 3600 THEN 0
    WHEN src.fetch_time IS NOT NULL
         AND src.fetch_time >= src.pay_time
         AND DATEDIFF(SECOND, src.pay_time, src.fetch_time) <= 24 * 3600 THEN 0
    ELSE 1
  END AS is_timeout,
  CASE
    WHEN src.fetch_time IS NOT NULL
         AND src.pay_time IS NOT NULL
         AND src.fetch_time >= src.pay_time
      THEN CAST(DATEDIFF(SECOND, src.pay_time, src.fetch_time) AS DECIMAL(20, 2)) / 3600.0
    ELSE NULL
  END AS pickup_hours,
  CASE
    WHEN src.receiver_address LIKE ${q('%-%-%')}
      THEN REVERSE(LEFT(REVERSE(src.receiver_address), CHARINDEX('-', REVERSE(src.receiver_address)) - 1))
    ELSE N''
  END AS district
INTO #pickup_scored
FROM #pickup_enriched src;

SELECT *
INTO #pickup_valid
FROM #pickup_scored
WHERE is_excluded = 0;

SELECT
  COUNT_BIG(*) AS total_count,
  SUM(CASE WHEN is_timeout = 1 THEN 1 ELSE 0 END) AS timeout_count,
  CAST(ROUND(AVG(pickup_hours), 2) AS DECIMAL(10, 2)) AS avg_hours
FROM #pickup_valid;

SELECT
  ISNULL(NULLIF(warehouse_name, N''), ${q(UNKNOWN_WAREHOUSE)}) AS warehouse_name,
  COUNT_BIG(*) AS total_count,
  SUM(CASE WHEN is_timeout = 1 THEN 1 ELSE 0 END) AS timeout_count,
  CAST(ROUND(AVG(pickup_hours), 2) AS DECIMAL(10, 2)) AS avg_hours
FROM #pickup_valid
GROUP BY ISNULL(NULLIF(warehouse_name, N''), ${q(UNKNOWN_WAREHOUSE)});

SELECT
  ISNULL(NULLIF(supplier_name, N''), ${q(UNKNOWN_SUPPLIER)}) AS supplier_name,
  COUNT_BIG(*) AS total_count,
  SUM(CASE WHEN is_timeout = 1 THEN 1 ELSE 0 END) AS timeout_count,
  CAST(ROUND(AVG(pickup_hours), 2) AS DECIMAL(10, 2)) AS avg_hours
FROM #pickup_valid
GROUP BY ISNULL(NULLIF(supplier_name, N''), ${q(UNKNOWN_SUPPLIER)});

SELECT
  ISNULL(NULLIF(province, N''), ${q(UNKNOWN_PROVINCE)}) AS province,
  COUNT_BIG(*) AS total_count,
  SUM(CASE WHEN is_timeout = 1 THEN 1 ELSE 0 END) AS timeout_count,
  CAST(ROUND(AVG(pickup_hours), 2) AS DECIMAL(10, 2)) AS avg_hours
FROM #pickup_valid
GROUP BY ISNULL(NULLIF(province, N''), ${q(UNKNOWN_PROVINCE)});

SELECT
  ISNULL(NULLIF(courier_name, N''), ${q(UNKNOWN_COURIER)}) AS courier_name,
  COUNT_BIG(*) AS total_count,
  SUM(CASE WHEN is_timeout = 1 THEN 1 ELSE 0 END) AS timeout_count,
  CAST(ROUND(AVG(pickup_hours), 2) AS DECIMAL(10, 2)) AS avg_hours
FROM #pickup_valid
GROUP BY ISNULL(NULLIF(courier_name, N''), ${q(UNKNOWN_COURIER)});

SELECT
  ISNULL(NULLIF(province, N''), ${q(UNKNOWN_PROVINCE)}) AS province,
  ISNULL(NULLIF(city, N''), N'') AS city,
  ISNULL(NULLIF(district, N''), N'') AS district,
  ISNULL(NULLIF(courier_name, N''), ${q(UNKNOWN_COURIER)}) AS courier_name,
  COUNT_BIG(*) AS total_count,
  SUM(CASE WHEN is_timeout = 1 THEN 1 ELSE 0 END) AS timeout_count,
  CAST(ROUND(AVG(pickup_hours), 2) AS DECIMAL(10, 2)) AS avg_hours
FROM #pickup_valid
GROUP BY
  ISNULL(NULLIF(province, N''), ${q(UNKNOWN_PROVINCE)}),
  ISNULL(NULLIF(city, N''), N''),
  ISNULL(NULLIF(district, N''), N''),
  ISNULL(NULLIF(courier_name, N''), ${q(UNKNOWN_COURIER)});
`;

const TABLE_DELETES = [
  'summary_daily_pickup_region_courier',
  'summary_daily_pickup_courier',
  'summary_daily_pickup_region',
  'summary_daily_pickup_supplier',
  'summary_daily_pickup_warehouse',
  'summary_daily_pickup',
];

const toInt = (value) => Number.parseInt(value, 10) || 0;
const toFloat = (value) => Number.parseFloat(value) || 0;

const calcPickupRate = (totalCount, timeoutCount) => (
  totalCount > 0 ? Number((((totalCount - timeoutCount) / totalCount) * 100).toFixed(2)) : 0
);

const TRANSIENT_LOCAL_WRITE_ERROR_CODES = new Set([
  'ETIMEDOUT',
  'ECONNRESET',
  'ECONNREFUSED',
  'EPIPE',
  'PROTOCOL_CONNECTION_LOST',
  'PROTOCOL_ENQUEUE_AFTER_FATAL_ERROR',
  'PROTOCOL_ENQUEUE_AFTER_QUIT',
  'PROTOCOL_PACKETS_OUT_OF_ORDER',
]);

const isDateString = (value) => /^\d{4}-\d{2}-\d{2}$/.test(value) && dayjs(value).format('YYYY-MM-DD') === value;

const buildPayDatePrefix = (statDate) => `${dayjs(statDate).subtract(1, 'day').format('YYYY/M/D')}%`;

const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

const isTransientLocalWriteError = (error) => {
  if (!error) {
    return false;
  }

  const message = String(error.message || '').toLowerCase();
  const code = String(error.code || '');

  return (
    TRANSIENT_LOCAL_WRITE_ERROR_CODES.has(code) ||
    message.includes('timed out') ||
    message.includes('closed state') ||
    message.includes('connection lost') ||
    message.includes('socket hang up')
  );
};

const chunkRows = (rows, size) => {
  const chunks = [];
  for (let index = 0; index < rows.length; index += size) {
    chunks.push(rows.slice(index, index + size));
  }
  return chunks;
};

const buildInsertSql = (tableName, columns, rowCount) => {
  const valuePlaceholder = `(${columns.map(() => '?').join(', ')})`;
  const placeholders = new Array(rowCount).fill(valuePlaceholder).join(', ');
  const quotedColumns = columns.map((column) => `\`${column}\``).join(', ');
  return `INSERT INTO ${tableName} (${quotedColumns}) VALUES ${placeholders}`;
};

const executeInsertChunks = async (connection, tableName, columns, rows, chunkSize = 500) => {
  if (rows.length === 0) {
    return;
  }

  for (const chunk of chunkRows(rows, chunkSize)) {
    const sql = buildInsertSql(tableName, columns, chunk.length);
    const params = chunk.flat();
    await connection.query(sql, params);
  }
};

const normalizeOverallRow = (statDate, row = {}) => {
  const totalCount = toInt(row.total_count);
  const timeoutCount = toInt(row.timeout_count);

  if (totalCount === 0 && timeoutCount === 0) {
    return null;
  }

  return [
    statDate,
    totalCount,
    timeoutCount,
    calcPickupRate(totalCount, timeoutCount),
    toFloat(row.avg_hours),
  ];
};

const normalizeNamedRows = (statDate, rows, nameField) => rows
  .map((row) => {
    const totalCount = toInt(row.total_count);
    const timeoutCount = toInt(row.timeout_count);
    return [
      statDate,
      row[nameField] || '',
      totalCount,
      timeoutCount,
      calcPickupRate(totalCount, timeoutCount),
      toFloat(row.avg_hours),
    ];
  });

const normalizeRegionCourierRows = (statDate, rows) => rows
  .map((row) => {
    const totalCount = toInt(row.total_count);
    const timeoutCount = toInt(row.timeout_count);
    return [
      statDate,
      row.province || UNKNOWN_PROVINCE,
      row.city || '',
      row.district || '',
      row.courier_name || UNKNOWN_COURIER,
      totalCount,
      timeoutCount,
      calcPickupRate(totalCount, timeoutCount),
      toFloat(row.avg_hours),
    ];
  });

const fetchPickupRecordsets = async (statDate) => {
  const pool = await cloudDb.getPool();
  const request = pool.request();
  request.input('payDatePrefix', buildPayDatePrefix(statDate));
  const result = await request.query(BACKFILL_PICKUP_SQL);
  return result.recordsets;
};

const replacePickupSummaryForDateOnce = async (statDate, recordsets) => {
  const [
    overallRows = [],
    warehouseRows = [],
    supplierRows = [],
    regionRows = [],
    courierRows = [],
    regionCourierRows = [],
  ] = recordsets;

  const overallRow = normalizeOverallRow(statDate, overallRows[0]);
  const warehouseInsertRows = normalizeNamedRows(statDate, warehouseRows, 'warehouse_name');
  const supplierInsertRows = normalizeNamedRows(statDate, supplierRows, 'supplier_name');
  const regionInsertRows = normalizeNamedRows(statDate, regionRows, 'province');
  const courierInsertRows = normalizeNamedRows(statDate, courierRows, 'courier_name');
  const regionCourierInsertRows = normalizeRegionCourierRows(statDate, regionCourierRows);

  const connection = await localDb.pool.getConnection();

  try {
    await connection.beginTransaction();

    for (const tableName of TABLE_DELETES) {
      await connection.query(`DELETE FROM ${tableName} WHERE stat_date = ?`, [statDate]);
    }

    if (overallRow) {
      await executeInsertChunks(
        connection,
        'summary_daily_pickup',
        ['stat_date', 'total_order_count', 'timeout_count', 'pickup_rate', 'avg_pickup_hours'],
        [overallRow],
        1
      );
    }

    await executeInsertChunks(
      connection,
      'summary_daily_pickup_warehouse',
      ['stat_date', 'warehouse_name', 'total_order_count', 'timeout_count', 'pickup_rate', 'avg_pickup_hours'],
      warehouseInsertRows
    );

    await executeInsertChunks(
      connection,
      'summary_daily_pickup_supplier',
      ['stat_date', 'supplier_name', 'total_order_count', 'timeout_count', 'pickup_rate', 'avg_pickup_hours'],
      supplierInsertRows
    );

    await executeInsertChunks(
      connection,
      'summary_daily_pickup_region',
      ['stat_date', 'province', 'total_order_count', 'timeout_count', 'pickup_rate', 'avg_pickup_hours'],
      regionInsertRows
    );

    await executeInsertChunks(
      connection,
      'summary_daily_pickup_courier',
      ['stat_date', 'courier_name', 'total_order_count', 'timeout_count', 'pickup_rate', 'avg_pickup_hours'],
      courierInsertRows
    );

    await executeInsertChunks(
      connection,
      'summary_daily_pickup_region_courier',
      ['stat_date', 'province', 'city', 'district', 'courier_name', 'total_order_count', 'timeout_count', 'pickup_rate', 'avg_pickup_hours'],
      regionCourierInsertRows
    );

    await connection.commit();

    console.log(
      `[pickup-backfill] ${statDate} total=${overallRow ? overallRow[1] : 0} timeout=${overallRow ? overallRow[2] : 0} `
        + `warehouse=${warehouseInsertRows.length} supplier=${supplierInsertRows.length} `
        + `region=${regionInsertRows.length} courier=${courierInsertRows.length} regionCourier=${regionCourierInsertRows.length}`
    );
  } catch (error) {
    try {
      await connection.rollback();
    } catch (rollbackError) {
      console.warn(`[pickup-backfill] rollback failed for ${statDate}: ${rollbackError.message}`);
    }
    throw error;
  } finally {
    connection.release();
  }
};

const replacePickupSummaryForDate = async (statDate, recordsets) => {
  const maxAttempts = 3;

  for (let attempt = 1; attempt <= maxAttempts; attempt += 1) {
    try {
      await replacePickupSummaryForDateOnce(statDate, recordsets);
      return;
    } catch (error) {
      const shouldRetry = attempt < maxAttempts && isTransientLocalWriteError(error);

      if (!shouldRetry) {
        throw error;
      }

      console.warn(
        `[pickup-backfill] local write retry for ${statDate} (${attempt}/${maxAttempts}) due to: ${error.message}`
      );
      await sleep(attempt * 1000);
    }
  }
};

const backfillRange = async (startDate, endDate) => {
  const start = dayjs(startDate);
  const end = dayjs(endDate);
  const totalDays = end.diff(start, 'day') + 1;

  if (totalDays <= 0) {
    throw new Error(`Invalid date range: ${startDate} -> ${endDate}`);
  }

  for (let offset = 0; offset < totalDays; offset += 1) {
    const statDate = start.add(offset, 'day').format('YYYY-MM-DD');
    const dayStart = Date.now();
    console.log(`[pickup-backfill] start ${statDate} (${offset + 1}/${totalDays}) payPrefix=${buildPayDatePrefix(statDate)}`);
    const recordsets = await fetchPickupRecordsets(statDate);
    await replacePickupSummaryForDate(statDate, recordsets);
    console.log(`[pickup-backfill] done ${statDate} elapsed=${Date.now() - dayStart}ms`);
  }
};

const main = async () => {
  const [startDate, endDate = startDate] = process.argv.slice(2);

  if (!isDateString(startDate) || !isDateString(endDate)) {
    throw new Error('Usage: node backend/scripts/backfillPickupHistory.js <YYYY-MM-DD> [YYYY-MM-DD]');
  }

  console.log(`[pickup-backfill] range ${startDate} -> ${endDate}`);

  try {
    await backfillRange(startDate, endDate);
    try {
      await refreshAllCache();
      console.log('[pickup-backfill] cache refresh done');
    } catch (error) {
      console.warn(`[pickup-backfill] cache refresh skipped: ${error.message}`);
    }
  } finally {
    await Promise.allSettled([
      cloudDb.close(),
      database.close(),
      redis.close(),
    ]);
  }
};

main().catch((error) => {
  console.error('[pickup-backfill] failed');
  console.error(error);
  process.exit(1);
});