MySQl篇（SQL - 基本介绍）（持续更新迭代）-CFANZ编程社区

在电商平台的数据分析中，处理品牌促销活动的日期交叉问题是一个挑战。本文将介绍几种高级SQL技巧，用于准确计算每个品牌的总优惠天数，即使在存在日期交叉的情况下。

问题背景

我们有一个促销活动表 shop_discount，记录了不同品牌的促销开始日期和结束日期。目标是计算每个品牌的总优惠天数，同时确保同一天内的多个优惠活动只计算一次。

示例数据

1.建表

create table shop_discount(
    brand string,
    stt string,
    edt string
);

2.导入数据

INSERT INTO shop_discount (brand, stt, edt) VALUES
('oppo', '2021-06-05', '2021-06-09'),
('oppo', '2021-06-11', '2021-06-21'),
('vivo', '2021-06-05', '2021-06-15'),
('vivo', '2021-06-09', '2021-06-21'),
('redmi', '2021-06-05', '2021-06-21'),
('redmi', '2021-06-09', '2021-06-15'),
('redmi', '2021-06-17', '2021-06-26'),
('huawei', '2021-06-05', '2021-06-26'),
('huawei', '2021-06-09', '2021-06-15'),
('huawei', '2021-06-17', '2021-06-21');

3.查询数据是否导入成功

select * from shop_discount;

高级SQL技巧

方法1：使用开窗进行连续区间划分及合并

这种方法通过识别每个品牌的连续促销区间来计算总天数。

SELECT 
    brand,
    SUM(days) AS promotion_day_count
FROM (
    SELECT 
        brand, DATEDIFF(MAX(edt), MIN(stt)) + 1 AS days 
    FROM (
        SELECT 
            brand, stt, edt,
            SUM(is_new_start) OVER (PARTITION BY brand ORDER BY stt) AS interval_id
        FROM (
            SELECT 
                brand, stt, edt,
                IF(stt > COALESCE(LAG_END_DATE, '1970-01-01'), 1, 0) AS is_new_start
            FROM (
                SELECT 
                    brand, stt, edt,
                    MAX(edt) OVER (PARTITION BY brand ORDER BY stt ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING) AS LAG_END_DATE
                FROM 
                    shop_discount
            ) t1
        ) t2
    ) t3 
    GROUP BY brand, interval_id
) t4 
GROUP BY brand;

方法2：使用开窗求出没有活动的天数

这种方法通过计算每个促销区间之间的空白天数来调整总天数。

SELECT 
    brand,
    DATEDIFF(MAX(edt), MIN(stt)) - SUM(no_promo_days) + 1 AS promotion_day_count
FROM (
    SELECT 
        brand, stt, edt,
        IF(stt > LAG_END_DATE, DATEDIFF(stt, LAG_END_DATE) - 1, 0) AS no_promo_days
    FROM (
        SELECT 
            brand, stt, edt,
            MAX(edt) OVER (PARTITION BY brand ORDER BY stt ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING) AS LAG_END_DATE
        FROM 
            shop_discount
    ) t1
) t2 
GROUP BY brand;

方法3：使用开窗去除区间之间重复的部分

这种方法通过确保每个促销日期只被计算一次来计算总天数。

SELECT 
    brand, 
    SUM(DATEDIFF(edt, start_date) + 1) AS promotion_day_count
FROM (
    SELECT 
        brand, MAX_END_DATE,
        IF(MAX_END_DATE IS NULL OR stt > MAX_END_DATE, stt, DATE_ADD(MAX_END_DATE, 1)) AS start_date, edt
    FROM (
        SELECT 
            brand, stt, edt,
            MAX(edt) OVER (PARTITION BY brand ORDER BY stt ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING) AS MAX_END_DATE
        FROM 
            shop_discount
    ) t1
) t2 
WHERE edt > start_date 
GROUP BY brand;

方法4：使用UDTF生成所有活动日期然后去重

这种方法通过生成所有可能的促销日期并去除重复来计算总天数。

SELECT 
    brand,
    COUNT(DISTINCT promo_date) AS promotion_day_count
FROM (
    SELECT 
        brand, DATE_ADD(stt, pos) AS promo_date 
    FROM 
        shop_discount
    LATERAL VIEW 
        POSEXPLODE(SPLIT(REPEAT(',', DATEDIFF(edt, stt)), ',')) tmp AS pos, element
) t1 
GROUP BY 
    brand;