BigQuery中获取Step1与Step2公共值:Step1全为null时取Step2值
BigQuery 实现特定逻辑的customer_id匹配需求
需求
在BigQuery中实现:获取Step1与Step2的公共customer_id值;若Step1的customer_id全为null,则取Step2的所有customer_id。
现有问题示例
场景1:Step_a全为null时(不符合预期)
现有查询:
with step_a as (select null as customer_id union all select null as customer_id union all select null as customer_id union all select null as customer_id), step_b as (select 1 as customer_id union all select 2 as customer_id union all select 3 as customer_id union all select 4 as customer_id ) (Select customer_id from step_a) intersect distinct (select customer_id from step_b)
- 预期结果:
customer_id 1 2 3 4 - 实际结果:无返回值
场景2:Step_a非null时(符合预期)
现有查询:
with step_a as (select 1 as customer_id union all select 2 as customer_id union all select 3 as customer_id union all select 4 as customer_id), step_b as (select 1 as customer_id union all select 2 as customer_id union all select 3 as customer_id union all select 4 as customer_id ) (Select customer_id from step_a) intersect distinct (select customer_id from step_b)
- 结果:符合预期,返回:
customer_id 1 2 3 4
需覆盖的全量场景
- 场景1:Step_a全为null,Step_b为1、2 → 输出:1、2
- 场景2:Step_a为1、2,Step_b为5 → 输出:无
- 场景3:Step_a为1、2,Step_b为1、2、5 → 输出:1、2
- 场景4:Step_a为5,Step_b全为null → 输出:无
解决方案
以下SQL可满足所有场景需求:
WITH step_a AS ( -- 替换为你的Step1实际查询逻辑 SELECT NULL AS customer_id UNION ALL SELECT NULL AS customer_id UNION ALL SELECT NULL AS customer_id UNION ALL SELECT NULL AS customer_id ), step_b AS ( -- 替换为你的Step2实际查询逻辑 SELECT 1 AS customer_id UNION ALL SELECT 2 AS customer_id UNION ALL SELECT 3 AS customer_id UNION ALL SELECT 4 AS customer_id ), step_a_stats AS ( SELECT COUNT(DISTINCT customer_id) AS non_null_count FROM step_a ) SELECT DISTINCT b.customer_id FROM step_b b CROSS JOIN step_a_stats s WHERE (s.non_null_count = 0 AND b.customer_id IS NOT NULL) OR b.customer_id IN (SELECT customer_id FROM step_a WHERE customer_id IS NOT NULL)
逻辑说明
- 通过
step_a_stats统计Step_a中非null的customer_id数量 - 条件分支:
- 若
non_null_count = 0(Step_a全为null),则选取Step_b所有非null的customer_id - 否则,选取Step_b中存在于Step_a非null集合中的
customer_id
- 若
- 自动处理Step_b全为null的场景,此时返回空结果,符合需求
内容的提问来源于stack exchange,提问作者hepzi
相关产品推荐
相关产品推荐

