BigQuery: select * replace from multiple nested column

Question

I'm trying to remove a field which is nested inside multiple records. Following this question, I tried the following (remove column d which has path a.b.d) but get an error: Unrecognized name: b at [6:68]

WITH T AS (
  SELECT * from unnest([struct(struct(struct(10 as c,'test' as d) as b,'field2' as field2) as a)])

)

select * replace ((select as struct a.* replace( (select as struct b.* except (d)) as b)) as a)  from T;

Mikhail Berlyant Mikhail Berlyant · Accepted Answer · 2018-03-19T18:15:29

from what I understood in your question - your original row is as below

STRUCT(STRUCT(STRUCT(10 AS c,'test' AS d) AS b,'field2' AS field2) AS a) original_struct

and you need to get it to

STRUCT(STRUCT(STRUCT(10 AS c) AS b,'field2' AS field2) AS a) original_struct

Below is for BigQuery Standard SQL for this to accomplish

#standardSQL
WITH t AS (
  SELECT STRUCT(STRUCT(STRUCT(10 AS c,'test' AS d) AS b,'field2' AS field2) AS a) original_struct 
) 
SELECT 
  original_struct,
  (
    SELECT AS STRUCT (
      SELECT AS STRUCT (
        SELECT AS STRUCT * EXCEPT(d) 
        FROM UNNEST([original_struct.a.b]) b
      ) b, 
      original_struct.a.field2
    ) a
  ) new_struct
FROM t

original struct look as

Row original_struct.a.b.c   original_struct.a.b.d   original_struct.a.field2    
1   10                      test                    field2

and new struct is

Row new_struct.a.b.c    new_struct.a.field2  
1   10                  field2

So, to further apply this to your particular case - see example below

#standardSQL
WITH t AS (
  SELECT 1 id, 2 x, 3 y, STRUCT(STRUCT(STRUCT(10 AS c,'test' AS d) AS b,'field2' AS field2) AS a) xyz, 4 v, 5 w 
) 
SELECT * REPLACE (
  (
    SELECT AS STRUCT (
      SELECT AS STRUCT (
        SELECT AS STRUCT * EXCEPT(d) 
        FROM UNNEST([xyz.a.b]) b
      ) b, 
      xyz.a.field2
    ) a
  ) AS xyz)
FROM t

BigQuery: select * replace from multiple nested column

1 Answers