KingbaseES lag 和 lead 函数

1.简介

lag与lead函数是跟偏移量相关的两个分析函数，通过这两个函数可以在一次查询中取出同一字段的前N行的数据(lag)和后N行的数据(lead)作为独立的列,从而更方便地进行进行数据过滤。

2.语法

LEAD/lag (expression [, offset [, default_value]])
OVER (
    [PARTITION BY partition_expression, ... ]
    ORDER BY sort_expression [ASC | DESC], ...
)

over()表示 lag()与lead()操作的数据都在over()的范围内，他里面可以使用partition by 语句（用于分组） order by 语句（用于排序）。partition by a order by b表示以a字段进行分组，再以b字段进行排序，对数据进行查询。

3.示例

示例数据：
test=# select * from sales;
 year | group_id | amount
------+----------+---------
 2021 |        1 | 1474.00
 2021 |        2 | 1787.00
 2021 |        3 | 1760.00
 2019 |        1 | 1915.00
 2019 |        2 | 1911.00
 2019 |        3 | 1118.00
 2020 |        1 | 1646.00
 2020 |        2 | 1975.00
 2020 |        3 | 1516.00
(9 行记录)

查询当年和下一年的销售额：
test=# WITH cte AS (
test(#     SELECT
test(#         year,
test(#         SUM(amount) amount
test(#     FROM sales
test(#     GROUP BY year
test(#     ORDER BY year
test(# )
test-# SELECT
test-#     year,
test-#     amount,
test-#     LEAD(amount, 1) OVER (ORDER BY year) next_year_sales
test-# from cte;
 year | amount  | next_year_sales
------+---------+-----------------
 2019 | 4944.00 |         5137.00
 2020 | 5137.00 |         5021.00
 2021 | 5021.00 |
(3 行记录)

每个产品组的当年销售额与下一年的销售额进行比较，没有销售额的默认为0：
test=# select year, amount, group_id,
test-#    LEAD(amount, 1,0::numeric) OVER (PARTITION BY group_id ORDER BY year) next_year_sales
test-# FROM sales;
 year | amount  | group_id | next_year_sales
------+---------+----------+-----------------
 2019 | 1915.00 |        1 |         1646.00
 2020 | 1646.00 |        1 |         1474.00
 2021 | 1474.00 |        1 |               0
 2019 | 1911.00 |        2 |         1975.00
 2020 | 1975.00 |        2 |         1787.00
 2021 | 1787.00 |        2 |               0
 2019 | 1118.00 |        3 |         1516.00
 2020 | 1516.00 |        3 |         1760.00
 2021 | 1760.00 |        3 |               0

lag函数用法与lead一致，这里就不再演示了。