如何取消将BigQuery字段存储为字符串?

下面是BigQuery标准SQL

#standardSQL
WITH t AS (
  SELECT "1234" AS row_id, '{ "id" : "123" , "items" : [ { "quantity" : 1 , "product" : { "id" : "p1" , "categories" : [ "cat1","cat2","cat3"]  }}] }' AS parts 
)
SELECT row_id, REPLACE(_categories, '"', '') _categories
FROM t, UNNEST(SPLIT(REGEXP_EXTRACT(
  JSON_EXTRACT(parts, '$.items'), 
  r'"categories":\[(.+?)]'))
) _categories 

并产生预期结果

Row row_id  _categories  
1   1234    cat1     
2   1234    cat2     
3   1234    cat3      

更新

上述解决方案主要集中于修复提取中使用的regexp,但没有解决具有多个产品的更一般的情况。下面的解决方案解决了这种更一般的情况

#standardSQL
WITH t AS (
  SELECT "1234" AS row_id, '''{ "id" : "123" , "items" : [ 
      { "quantity" : 1 , "product" : { "id" : "p1" , "categories" : [ "cat1","cat2","cat3"]  }},
      { "quantity" : 2 , "product" : { "id" : "p2" , "categories" : [ "cat4","cat5","cat6"]  }}
    ] }''' AS parts 
)
SELECT row_id, REPLACE(category, '"', '') category
FROM t, UNNEST(REGEXP_EXTRACT_ALL(parts, r'"categories" : \[(.+?)]')) categories,
UNNEST(SPLIT(categories)) category  

以结果

Row row_id  category     
1   1234    cat1     
2   1234    cat2     
3   1234    cat3     
4   1234    cat4     
5   1234    cat5     
6   1234    cat6