![](/img/trans.png)
[英]How do I use BigQuery DML to transform some fields of a struct nested within an array, within a struct, within an array?
[英]How to select a list of fields within a struct by prefix using Bigquery
我想做类似的事情:
SELECT
STRUCT(
c.customer.name*,
r.revenue.type*
) AS customer_revenue
FROM customer_table c
JOIN revenue_table r
USING (transaction_id)
其中customer
是一个包含许多字段的结构体,包括一些以“name”开头的字段(有些字段不是),而revenue
同样是一个结构体,其中一些字段以“type”开头,一些字段不以“name”开头'不。
我似乎无法弄清楚如何在 BigQuery 中做到这一点。 我只是不断收到语法错误,让我相信这是不可能的。 有任何想法吗?
考虑以下替代方式
#standardSQL
SELECT (
SELECT STRING_AGG(x)
FROM UNNEST(REGEXP_EXTRACT_ALL(TRIM(TO_JSON_STRING(c.customer), '{}'), r'("name\w+":\w+)(?:,|$)')) x
) || ',' || (
SELECT STRING_AGG(y)
FROM UNNEST(REGEXP_EXTRACT_ALL(TRIM(TO_JSON_STRING(r.revenue), '{}'), r'("type\w+":\w+)(?:,|$)')) y
) AS customer_revenue
FROM customer_table c
JOIN revenue_table r
USING (transaction_id)
在以下[超级简化]数据的情况下
#standardSQL
WITH customer_table AS (
SELECT 1 transaction_id, STRUCT(1 AS name1, 2 AS name2, 3 AS aaa3, 4 AS bbb4) customer
), revenue_table AS (
SELECT 1 transaction_id, STRUCT(11 AS type11, 22 AS type22, 33 AS type33, 44 AS ccc44) revenue
)
结果将是
Row customer_revenue
1 "name1":1,"name2":2,"type11":11,"type22":22,"type33":33
显然 - 这不完全是你在问题中问的 - 但正如我已经评论过的 - “真的不可行:o(你可能需要重新考虑你的要求” - 所以上述解决方案可能会帮助你做到这一点
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.