Cross apply MSSQL operator equivalent in bigquery

Question

is there an equivalent to cross apply in Sql server syntaxe or in bigquery syntaxe.

I need to write this query without the cross apply:

   select *
    from t1 
    cross apply (select top 1 col1,col2,col3
    from t2
    where col1 = t1.col1
    and (col2 = '0' or col2 = t1.col2)
and (col3 = '0' or (col3 = left(t1.col3)  and t1.col4 = 'fr'))
    order by col2 desc, col3 desc)

t1 called 'Delivery' contains deliveries informations:

Delivery_ID,Delivery_ShipmentDate,Country_Code,address_ZipCode and the providerservice_id

t2 called 'ProviderService' contains informations about the providerservice:

Providerservice_id,DCountry_ID , DZone_Code  as well as a numeric ProviderService_DeliveryTime.

So each Delivery_ID has a ProviderService_ID And For the same Providerservice we could have several distinct ProviderService_DeliveryTime according to the other columns in this table(DCountry_ID , DZone_Code)

So the final query will be:

    Select t1.Delivery_ShipmentDate,t2.ProviderService_DeliveryTime
    from Delivery t1
    cross apply (select top 1 ProviderService_DeliveryTime,DCountry_ID , DZone_Code 
from ProviderService 
    where ProviderService_id = t1.ProviderService_id
And (DCountry_ID = '0' or DCountry_ID = t1.Country_Code)
And (DZone_Code = '0' or (DZone_Code = left(t1.address_ZipCode,2) and t1.Country_Code='FR'))
        order by t2.DCountry_ID desc, t2.DZone_Code desc)sq

In fact I need first to write the same query in sql server syntaxe and then to write it in Bigquery because BigQuery don't recognize "cross apply" operator.

I tried to do it with row_number window function but it did not work:

    with cte as (Select t1.Delivery_ShipmentDate,t2.ProviderService_DeliveryTime,row_number() over (partition by t2.providerservice_id order by t2.DCountry_ID desc, t2.DZone_Code desc) as num
    from Delivery t1
    inner join providerservice t2 on t1.providerservice_id = t2.providerservice_id
And (DCountry_ID = '0' or DCountry_ID = t1.Country_Code)
And (DZone_Code = '0' or (DZone_Code = left(t1.address_ZipCode,2) and t1.Country_Code='FR')))


select * from cte where num = 1

it works only when I filter by delivery_id:

 with cte as (Select t1.Delivery_ShipmentDate,t2.ProviderService_DeliveryTime,row_number() over (partition by t2.providerservice_id order by t2.DCountry_ID desc, t2.DZone_Code desc) as num
    from Delivery t1
    inner join providerservice t2 on t1.providerservice_id = t2.providerservice_id
And (DCountry_ID = '0' or DCountry_ID = t1.Country_Code)
And (DZone_Code = '0' or (DZone_Code = left(t1.address_ZipCode,2) and t1.Country_Code='FR'))
where delivery_id = xxxx)


select * from cte where num = 1

Can any one help me please? Thanks!!

Answer 1

Below is for BigQuery Standard SQL

#standardSQL
SELECT * EXCEPT(pos)
FROM (
  SELECT *
    ROW_NUMBER() OVER(PARTITION BY TO_JSON_STRING(t1) ORDER BY t2.col2 DESC, t2.col3 DESC) pos
  FROM t1 INNER JOIN t2
  ON t2.col1 = t1.col1
  AND (t2.col2 = '0' OR t2.col2 = t1.col2)
  AND (t2.col3 = '0' OR (t2.col3 = left(t1.col3)  AND t1.col4 = 'fr'))
)
WHERE pos = 1

I wish you gave us some data to play with - but having absence of it - above is just quick shot for you to try

I checked this approach (of implementing CROSS APPLY in BigQuery) against classic example with customers and orders tables and it worked

Answer 2

Would correlated subquery work the same as cross apply?

   select *
    from t1 , (select col1,col2,col3
    from t2
    where col1 = t1.col1
    and (col2 = '0' or col2 = t1.col2)
and (col3 = '0' or (col3 = left(t1.col3)  and t1.col4 = 'fr'))
    order by col2 desc, col3 desc
    limit 1)