I have three SQL selects, the results of which I need to add together. Two of the three use fairly complex joins.
select sum(field_one) from t_a join t_b on (t_a.bid = t_b.id) where t_b.user_id=:id
select sum(field_two) from t_c join t_d on (t_c.did = t_d.id) where t_d.user_id=:id
select sum(field_three) from t_e where t_e.user_id=:id
What I need is the sum of all three values. sum(field_one)+sum(field_two)+sum(field_three). Is there anyway to do this in a single statement?
You could UNION ALL them.
Do not use UNION, since it omits duplicate values (5+5+5 would result in 5).
Select Sum(s)
From
(
Select Sum(field_one) As s ...
Union All
Select Sum(field_two) ...
Union All
Select Sum(field_three) ...
) x
You can do this without using Union like this
Sample Query
select( (select 15) + (select 10) + (select 20))
Your Query
select
(
(select sum(field_one) from t_a join t_b on (t_a.bid = t_b.id) where t_b.user_id=:id) +
(select sum(field_two) from t_c join t_d on (t_c.did = t_d.id) where t_d.user_id=:id) +
(select sum(field_three) from t_e where t_e.user_id=:id)
)
You can use a UNION and a subselect to do that:
select sum(`sum`) FROM
(
select sum(field_one) as `sum` from t_a join t_b on (t_a.bid = t_b.id) where t_b.user_id=:id
UNION ALL
select sum(field_two) from t_c join t_d on (t_c.did = t_d.id) where t_d.user_id=:id
UNION ALL
select sum(field_three) from t_e where t_e.user_id=:id
) as x;
Edit: Updated my answer to use UNION ALL, as suggested by Peter Lang.
Related
have a problem with my query:
select Sources.dataset.name setName, x.element_name Commodity_Is, y.element_name Provider_Is
from meta.object_meta_v x, meta.object_meta_v y
join dw.load_set2_curve on x.object_id = dw.load_set2_curve.curve_id
join Sources.dataset on Sources.dataset.id = dw.load_set2_curve.load_set_id
where dw.load_set2_curve.curve_id in (
select max(curve_id) sample_curve_id from dw.load_Set2_curve
group by load_set_id
)
and (meta.object_meta_v.attribute = 'Provider' or meta.object_meta_v.attribute = 'Commodity');
the error is on the line:
join dw.load_set2_curve on x.object_id = dw.load_set2_curve.curve_id
I know why, because, according to this article 'https://stackoverflow.com/questions/10500048/invalid-identifier-on-oracle-inner-join' - "Looks like you cannot refer to an outer table alias in the join condition of the inner query." Unfortunately, I don't know how to find a workaround as I am looking for two different records (Commodity_is and Provider_is) from the same table in my query (with aliases 'x' and 'y').
Do you have any hints?
Your problem is that you are not using the table aliases in the SELECT, ON and WHERE clauses and are trying to refer to identifiers as schema.table.column and, in some cases that is ambiguous and you need to use table_alias.column.
Additionally, you are trying to mix legacy comma joins with ANSI joins (which does work but the comma joins need to be last, not first, so its easier just to use ANSI joins all the way through):
select ds.name setName,
x.element_name Commodity_Is,
y.element_name Provider_Is
from meta.object_meta_v x
CROSS JOIN meta.object_meta_v y
INNER JOIN dw.load_set2_curve lsc
ON x.object_id = lsc.curve_id
INNER JOIN Sources.dataset ds
ON ds.id = lsc.load_set_id
where lsc.curve_id in (
select max(curve_id) sample_curve_id
from dw.load_Set2_curve
group by load_set_id
)
and ( x.attribute = 'Provider'
or y.attribute = 'Commodity');
Which, for the sample data:
CREATE TABLE meta.object_meta_v (object_id, element_name, attribute) AS
SELECT 1, 'A', 'Provider' FROM DUAL UNION ALL
SELECT 2, 'B', 'Commodity' FROM DUAL;
CREATE TABLE dw.load_set2_curve (curve_id, load_set_id) AS
SELECT 1, 100 FROM DUAL UNION ALL
SELECT 2, 200 FROM DUAL;
CREATE TABLE sources.dataset (id, name) AS
SELECT 100, 'DS1' FROM DUAL UNION ALL
SELECT 200, 'DS2' FROM DUAL;
Outputs:
SETNAME
COMMODITY_IS
PROVIDER_IS
DS1
A
B
DS1
A
A
DS2
B
B
db<>fiddle here
I'm currently trying to find the percentage of certain amount of preregistered users in my postgres db, the operation would be (1185 * 100) / 3104 = 38.17. To do that I'm using two select statements to retrieve each count, but I've been unable to operate between them:
+ count +
- 1185 -
- 3104 -
This is what I have:
select
count(*)
from crm_user_preregistred
left join crm_player on crm_user_preregistred."document" = crm_player."document"
left join crm_user on crm_player.user_id = crm_user.id
where crm_user.email is not null
union all
select
count(*)
from crm_user_preregistred
Thanks in advance for any hint or help.
you can use some with clause to simplifie your selects, substitute the values with your count(*) selects, maybe some formating to the result, and a check for 0 on value2
with temp_value1 as (
select 1185 as value1 ),
temp_value2 as (
select 3104 as value2 )
select (select temp_value1.value1::float * 100 from temp_value1) /
(select temp_value2.value2::float from temp_value2)
result :
38.17654639175258
with your selects:
with temp_value1 as (
select
count(*) as value1
from crm_user_preregistred
left join crm_player on crm_user_preregistred."document" = crm_player."document"
left join crm_user on crm_player.user_id = crm_user.id
where crm_user.email is not null
),
temp_value2 as (
select
count(*) as value2
from crm_user_preregistred
)
select (select temp_value1.value1::float * 100 from temp_value1) / (select temp_value2.value2::float from temp_value2)
You can do this in one query:
select count(*) filter (where cu.email is not null) * 100.0 / max(cup.cnt)
from (select cup.*, count(*) over () as cnt
from crm_user_preregistred cup
) cup left join
crm_player cp
on cup."document" = cp."document" left join
crm_user cu
on cp.user_id = cu.id
where cu.email is not null;
I suspect that the query could be simplified further, but without knowing your data model, it is hard to make specific suggestions.
I am not able to validate this legacy sql into standard bigquery sql as I don't know what else is required to change here(This query fails during validation if I choose standard SQL as big query dialect):
SELECT
lineitem.*,
proposal_lineitem.*,
porder.*,
company.*,
product.*,
proposal.*,
trafficker.name,
salesperson.name,
rate_card.*
FROM (
SELECT
*
FROM
dfp_data.dfp_order_lineitem
WHERE
DATE(end_datetime) >= DATE(DATE_ADD(CURRENT_TIMESTAMP(), -1, 'YEAR'))
OR end_datetime IS NULL ) lineitem
JOIN (
SELECT
*
FROM
dfp_data.dfp_order) porder
ON
lineitem.order_id = porder.id
LEFT JOIN (
SELECT
*
FROM
adpoint_data.dfp_proposal_lineitem) proposal_lineitem
ON
lineitem.id = proposal_lineitem.dfp_lineitem_id
JOIN (
SELECT
*
FROM
dfp_data.dfp_company) company
ON
porder.advertiser_id = company.id
LEFT JOIN (
SELECT
*
FROM
adpoint_data.dfp_product) product
ON
proposal_lineitem.product_id=product.id
LEFT JOIN (
SELECT
*
FROM
adpoint_data.dfp_proposal) proposal
ON
proposal_lineitem.proposal_id=proposal.id
LEFT JOIN (
SELECT
*
FROM
adpoint_data.dfp_rate_card) rate_card
ON
proposal_lineitem.ratecard_id=rate_card.id
LEFT JOIN (
SELECT
id,
name
FROM
dfp_data.dfp_user) trafficker
ON
porder.trafficker_id =trafficker.id
LEFT JOIN (
SELECT
id,
name
FROM
dfp_data.dfp_user) salesperson
ON
porder. salesperson_id =salesperson.id
Most likely the error you are getting is something like below
Duplicate column names in the result are not supported. Found duplicate(s): name
Legacy SQL adjust trafficker.name and salesperson.name in your SELECT statement into respectively trafficker_name and salesperson_name thus effectively eliminating column names duplication
Standard SQL behaves differently and treat both those columns as named name thus producing duplication case. To avoid it - you just need to provide aliases as in example below
SELECT
lineitem.*,
proposal_lineitem.*,
porder.*,
company.*,
product.*,
proposal.*,
trafficker.name AS trafficker_name,
salesperson.name AS salesperson_name,
rate_card.*
FROM ( ...
You can easily check above explained using below simplified/dummy queries
#legacySQL
SELECT
porder.*,
trafficker.name,
salesperson.name
FROM (
SELECT 1 order_id, 'abc' order_name, 1 trafficker_id, 2 salesperson_id
) porder
LEFT JOIN (SELECT 1 id, 'trafficker' name) trafficker
ON porder.trafficker_id =trafficker.id
LEFT JOIN (SELECT 2 id, 'salesperson' name ) salesperson
ON porder. salesperson_id =salesperson.id
and
#standardSQL
SELECT
porder.*,
trafficker.name AS trafficker_name,
salesperson.name AS salesperson_name
FROM (
SELECT 1 order_id, 'abc' order_name, 1 trafficker_id, 2 salesperson_id
) porder
LEFT JOIN (SELECT 1 id, 'trafficker' name) trafficker
ON porder.trafficker_id =trafficker.id
LEFT JOIN (SELECT 2 id, 'salesperson' name ) salesperson
ON porder. salesperson_id =salesperson.id
Note: if you have more duplicate names - you need to alias all of them too
I have a table variable that holds orderID, UnitID and OrderServiceId (it is already populated via a query with insert statement).
I then have a query under this that returns 15 columns which also include the OrderId, UnitId, OrderServiceId
I need to only return the rows from this query where the same combination of OrderId, UnitId, and OrderServiceId are not in the table variable.
You can use NOT EXISTS. e.g.
FROM YourQuery q
WHERE NOT EXISTS
(
SELECT * FROM #TableVar t
WHERE t.OrderId = q.OrderId
and t.UnitId = q.UnitId
and t.OrderServiceId=q.OrderServiceId
)
select q.*
from (
MyQuery
) q
left outer join MyTableVariable t on q.ORDERID = t.ORDERID
and q.UNITID= t.UNITID
and q.ORDERSERVICESID = t.ORDERSERVICESID
where t.ORDERID is null
You can use EXCEPT | INTERSECT operators for this (link).
Example:
(select 3,4,1
union all
select 2,4,1)
intersect
(select 1,2,9
union all
select 3,4,1)
Suppose I have this query
SELECT * FROM (
SELECT * FROM table_a
WHERE id > 10 )
AS a_results LEFT JOIN
(SELECT * from table_b
WHERE id IN
(SElECT id FROM a_results)
ON (a_results.id = b_results.id)
I would get the error "a_results is not a table". Anywhere I could use the re-use the results of the subquery?
Edit: It has been noted that this query doesn't make sense...it doesn't, yes. This is just to illustrate the question which I am asking; the 'real' query actually looks something like this:
SELECT SQL_CALC_FOUND_ROWS * FROM
( SELECT wp_pod_tbl_hotel . *
FROM wp_pod_tbl_hotel, wp_pod_rel, wp_pod
WHERE wp_pod_rel.field_id =12
AND wp_pod_rel.tbl_row_id =1
AND wp_pod.tbl_row_id = wp_pod_tbl_hotel.id
AND wp_pod_rel.pod_id = wp_pod.id
) as
found_hotel LEFT JOIN (
SELECT COUNT(*) as review_count, avg( (
location_rating + staff_performance_rating + condition_rating + room_comfort_rating + food_rating + value_rating
) /6 ) AS average_score, hotelid
FROM (
SELECT r. * , wp_pod_rel.tbl_row_id AS hotelid
FROM wp_pod_tbl_review r, wp_pod_rel, wp_pod
WHERE wp_pod_rel.field_id =11
AND wp_pod_rel.pod_id = wp_pod.id
AND r.id = wp_pod.tbl_row_id
AND wp_pod_rel.tbl_row_id
IN (
SELECT wp_pod_tbl_hotel .id
FROM wp_pod_tbl_hotel, wp_pod_rel, wp_pod
WHERE wp_pod_rel.field_id =12
AND wp_pod_rel.tbl_row_id =1
AND wp_pod.tbl_row_id = wp_pod_tbl_hotel.id
AND wp_pod_rel.pod_id = wp_pod.id
)
) AS hotel_reviews
GROUP BY hotel_reviews.hotelid
ORDER BY average_score DESC
AS sorted_hotel ON (id = sorted_hotel.hotelid)
As you can see, the sub-query which makes up the found_query table is repeated elsewhere downward as another sub-query, so I was hoping to re-use the results
You can not use a sub-query like this.
I'm not sure I understand your query, but wouldn't that be sufficient?
SELECT * FROM table_a a
LEFT JOIN table_b b ON ( b.id = a.id )
WHERE a.id > 10
It would return all rows from table_a where id > 10 and LEFT JOIN rows from table_b where id matches.