Arrays BigQuery-按数组分组

Arrays BigQuery-按数组分组,arrays,group-by,google-cloud-platform,google-bigquery,Arrays,Group By,Google Cloud Platform,Google Bigquery,我想按数组分组 示例查询: #standardSQL WITH `project.dataset.table` AS ( SELECT 'compute' description, '[{"key":"application","value":"scaled-server"},{"key":"department","value":"hrd"}]' labels, 0.323316 cost UNION ALL SELECT 'compute' description, '[{"key"

我想按数组分组

示例查询:

#standardSQL
WITH `project.dataset.table` AS (
  SELECT 'compute' description, '[{"key":"application","value":"scaled-server"},{"key":"department","value":"hrd"}]' labels, 0.323316 cost UNION ALL
  SELECT 'compute' description, '[{"key":"application","value":"scaled-server"},{"key":"department","value":"hrd"}]' labels, 0.342825 cost 
)
SELECT 
  description, 
  ARRAY(
    SELECT AS STRUCT 
      JSON_EXTRACT_SCALAR(kv, '$.key') key, 
      JSON_EXTRACT_SCALAR(kv, '$.value') value 
    FROM UNNEST(SPLIT(labels, '},{')) kv_temp, 
    UNNEST([CONCAT('{', REGEXP_REPLACE(kv_temp, r'^\[{|}]$', ''), '}')]) kv
  ) labels,
  cost
FROM `project.dataset.table`
上述查询的结果:

Row description labels.key      labels.value    cost     
1   compute         application     scaled-server   0.323316     
                    department      hrd      
2   compute         application    scaled-server  0.342825     
                    department     hrd
我希望得到如下结果:

Row description labels.key      labels.value    cost     
1   compute         application     scaled-server   0.666141     
                    department      hrd   

分组的逻辑是什么?是吗?所有键和值都应该相同才能分组?我刚刚更新了这个问题,目的是,“描述”列有VM、cloudSQL、BQ这样的服务,而“标签”列告诉我们有关标签的信息,如VM标签(env=prod),所以,如果我想得到具有env=staging的compute或vm的成本,那么就结束这一天。这就是为什么我需要先按描述分组,然后按键值分组
#standardSQL
WITH `project.dataset.table` AS (
  SELECT 'compute' description, '[{"key":"application","value":"scaled-server"},{"key":"department","value":"hrd"}]' labels, 0.323316 cost UNION ALL
  SELECT 'compute' description, '[{"key":"application","value":"scaled-server"},{"key":"department","value":"hrd"}]' labels, 0.342825 cost 
), temp AS (
  SELECT description, labels, SUM(cost) AS cost
  FROM `project.dataset.table`
  GROUP BY description, labels
)
SELECT 
  description, 
  ARRAY(
    SELECT AS STRUCT 
      JSON_EXTRACT_SCALAR(kv, '$.key') key, 
      JSON_EXTRACT_SCALAR(kv, '$.value') value 
    FROM UNNEST(SPLIT(labels, '},{')) kv_temp, 
    UNNEST([CONCAT('{', REGEXP_REPLACE(kv_temp, r'^\[{|}]$', ''), '}')]) kv
  ) labels,
  cost
FROM temp