Flink sql 支持在流式数据上使用LAG和LEAD函数吗
如何在流式数据源上使用分析函数LAG和EAD函数：
kafka输入数据如：
{"t":"2020-04-01T05:00:00Z", "id":"1", "speed":1.0}
{"t":"2020-04-01T05:05:00Z", "id":"1", "speed":2.0}
{"t":"2020-04-01T05:10:00Z", "id":"1", "speed":3.0}
{"t":"2020-04-01T05:15:00Z", "id":"1", "speed":4.0}
{"t":"2020-04-01T05:20:00Z", "id":"1", "speed":5.0}
{"t":"2020-04-01T05:25:00Z", "id":"1", "speed":6.0}
sql如下：
INSERT INTO topic_sink
SELECT
t,
id,
speed,
LAG(speed, 1) OVER w AS speed_1,
LAG(speed, 2) OVER w AS speed_2
FROM topic_source
WINDOW w AS (
PARTITION BY id
ORDER BY t
)
我期望得到的结果数据是
{"t":"2020-04-01T05:00:00Z", "id":"1", "speed":1.0, "speed_1":null, "speed_2":null}
{"t":"2020-04-01T05:05:00Z", "id":"1", "speed":2.0,"speed_1":1.0, "speed_2":null}
{"t":"2020-04-01T05:10:00Z", "id":"1", "speed":3.0,"speed_1":2.0, "speed_2":1.0}
{"t":"2020-04-01T05:15:00Z", "id":"1", "speed":4.0,"speed_1":3.0, "speed_2":2.0}
{"t":"2020-04-01T05:20:00Z", "id":"1", "speed":5.0,"speed_1":4.0, "speed_2":3.0}
{"t":"2020-04-01T05:25:00Z", "id":"1", "speed":6.0",speed_1":5.0, "speed_2":4.0}
实际得到的结果数据是：
{"t":"2020-04-01T05:00:00Z", "id":"1", "speed":1.0, "speed_1":1.0, "speed_2":1.0}
{"t":"2020-04-01T05:05:00Z", "id":"1", "speed":2.0,"speed_1":2.0, "speed_2":2.0}
{"t":"2020-04-01T05:10:00Z", "id":"1", "speed":3.0,"speed_1":3.0, "speed_2":3.0}
{"t":"2020-04-01T05:15:00Z", "id":"1", "speed":4.0,"speed_1":4.0, "speed_2":4.0}
{"t":"2020-04-01T05:20:00Z", "id":"1", "speed":5.0,"speed_1":5.0, "speed_2":5.0}
{"t":"2020-04-01T05:25:00Z", "id":"1", "speed":6.0",speed_1":6.0, "speed_2":6.0}
想问一下flink sql里的LAG函数能完成我期望的计算吗？如果可以sql该如何写？
写回答
-
INSERT INTO topic_sink SELECT t, id, speed, LAG(speed, 1, null) OVER w AS speed_1, LAG(speed, 2, null) OVER w AS speed_2 FROM topic_source WINDOW w AS ( PARTITION BY id ORDER BY t )这样应该可以
2020-11-18 11:39:13赞同 1 展开评论 打赏 -
SQL Server从2012开始，引入了LEAD和LAG函数，这两个函数可以把之前要关联查询的方法，改为可直接获取当前数据上下相邻多少行数据，描述的有点不清楚，上数据吧，测试数据：
--测试数据 if not object_id(N'Tempdb..#T') is null drop table #T Go Create table #T([Id] int,[Value] int) Insert #T select 1,10 union all select 2,20 union all select 3,30 union all select 4,40 union all select 5,50 Go --测试数据结束 以LEAD方法为例（LAG一样），语法是这样的：LEAD ( scalar_expression [ ,offset ] , [ default ] ) OVER ( [ partition_by_clause ] order_by_clause )，读取的数据，偏移多少行，如果没有值默认值是多少，写法如下：
SELECT *, LEAD(Value, 1, 666) OVER (ORDER BY Value) AS LEADVALUE, --提前1行，默认值666 LAG(Value, 2, 888) OVER (ORDER BY Value) AS LAGVALUE --滞后2行，默认值888 FROM #T; 结果如下：
我们也可以不写偏移数和默认值，结果就是这样的：
SELECT *, LEAD(Value) OVER (ORDER BY Value) AS LEADVALUE, --提前1行 LAG(Value) OVER (ORDER BY Value) AS LAGVALUE --滞后1行
2020-06-20 13:49:33赞同 展开评论 打赏
版权声明：本文内容由阿里云实名注册用户自发贡献，版权归原作者所有，阿里云开发者社区不拥有其著作权，亦不承担相应法律责任。具体规则请查看《阿里云开发者社区用户服务协议》和《阿里云开发者社区知识产权保护指引》。如果您发现本社区中有涉嫌抄袭的内容，填写侵权投诉表单进行举报，一经查实，本社区将立刻删除涉嫌侵权内容。
相关课程
更多
相关文章
相关电子书
更多
