SELECT DName
FROM drugs
WHERE DID IN
(
SELECT DID,SID
FROM transactions
GROUP BY TotalCost
HAVING SID = 1 AND TotalCost > 100
)
Doing such a query inside brackets will give me a result with 2 columns that I need to select results from one of them. In order to use SID in HAVING clause, I need to include it in SELECT operator inside brackets and that's why I am getting 2 columns as a result.
You can't have 2 columns in an IN statement.
You can just remove the SID from the select and only use it in the HAVING. You don't actually need to retrieve the data and your IN clause will work.
You can use EXISTS in the following way:
SELECT DName
FROM drugs
WHERE EXISTS
(
SELECT *
FROM transactions
WHERE SID = 1
AND TotalCost > 100
AND (
drugs.DID = transactions.SID
OR drugs.DID = transactions.DID
)
)
If I understood you correctly then you want something like this:
SELECT DName
FROM drugs
WHERE (DID, SID) IN
(
//Subquery
)
Answer updated. Sorry i have been deleted my previous answer. And thanks to ZoharPeled for reminding that my previous answer's wrong.
If I understood your question, This should works for you :
SELECT DName
FROM drugs
WHERE DID IN (
SELECT DID FROM(
SELECT DID,SID,TotalCost
FROM transactions
GROUP BY TotalCost
HAVING SID = 1 AND TotalCost > 100
) AS T
)
You have a few answers there - but I'll throw an alternative option into the ring:
select DName
from drugs d
inner join (
SELECT DID
FROM transactions
WHERE SID = 1
AND TotalCost > 100
) tx on d.Did =tx.Did
I think (and I'm open to correction here) that if you use something like...
select *
from table
where EXISTS in ( // some subquery )
... the subquery in the EXISTS clause must be run for each row found in table. With an inner join approach, the RDBMS will execute the sql for the inline table once and then hold the results in memory to join back to the other table. For small tables this is fine - but for larger tables there could be a significant performance hit.
Related
I am running this query on MySQL
SELECT ID FROM (
SELECT ID, msisdn
FROM (
SELECT * FROM TT2
)
);
and it is giving this error:
Every derived table must have its own alias.
What's causing this error?
Every derived table (AKA sub-query) must indeed have an alias. I.e. each query in brackets must be given an alias (AS whatever), which can the be used to refer to it in the rest of the outer query.
SELECT ID FROM (
SELECT ID, msisdn FROM (
SELECT * FROM TT2
) AS T
) AS T
In your case, of course, the entire query could be replaced with:
SELECT ID FROM TT2
I think it's asking you to do this:
SELECT ID
FROM (SELECT ID,
msisdn
FROM (SELECT * FROM TT2) as myalias
) as anotheralias;
But why would you write this query in the first place?
Here's a different example that can't be rewritten without aliases ( can't GROUP BY DISTINCT).
Imagine a table called purchases that records purchases made by customers at stores, i.e. it's a many to many table and the software needs to know which customers have made purchases at more than one store:
SELECT DISTINCT customer_id, SUM(1)
FROM ( SELECT DISTINCT customer_id, store_id FROM purchases)
GROUP BY customer_id HAVING 1 < SUM(1);
..will break with the error Every derived table must have its own alias. To fix:
SELECT DISTINCT customer_id, SUM(1)
FROM ( SELECT DISTINCT customer_id, store_id FROM purchases) AS custom
GROUP BY customer_id HAVING 1 < SUM(1);
( Note the AS custom alias).
I arrived here because I thought I should check in SO if there are adequate answers, after a syntax error that gave me this error, or if I could possibly post an answer myself.
OK, the answers here explain what this error is, so not much more to say, but nevertheless I will give my 2 cents, using my own words:
This error is caused by the fact that you basically generate a new table with your subquery for the FROM command.
That's what a derived table is, and as such, it needs to have an alias (actually a name reference to it).
Given the following hypothetical query:
SELECT id, key1
FROM (
SELECT t1.ID id, t2.key1 key1, t2.key2 key2, t2.key3 key3
FROM table1 t1
LEFT JOIN table2 t2 ON t1.id = t2.id
WHERE t2.key3 = 'some-value'
) AS tt
At the end, the whole subquery inside the FROM command will produce the table that is aliased as tt and it will have the following columns id, key1, key2, key3.
Then, with the initial SELECT, we finally select the id and key1 from that generated table (tt).
I am running this query on MySQL
SELECT ID FROM (
SELECT ID, msisdn
FROM (
SELECT * FROM TT2
)
);
and it is giving this error:
Every derived table must have its own alias.
What's causing this error?
Every derived table (AKA sub-query) must indeed have an alias. I.e. each query in brackets must be given an alias (AS whatever), which can the be used to refer to it in the rest of the outer query.
SELECT ID FROM (
SELECT ID, msisdn FROM (
SELECT * FROM TT2
) AS T
) AS T
In your case, of course, the entire query could be replaced with:
SELECT ID FROM TT2
I think it's asking you to do this:
SELECT ID
FROM (SELECT ID,
msisdn
FROM (SELECT * FROM TT2) as myalias
) as anotheralias;
But why would you write this query in the first place?
Here's a different example that can't be rewritten without aliases ( can't GROUP BY DISTINCT).
Imagine a table called purchases that records purchases made by customers at stores, i.e. it's a many to many table and the software needs to know which customers have made purchases at more than one store:
SELECT DISTINCT customer_id, SUM(1)
FROM ( SELECT DISTINCT customer_id, store_id FROM purchases)
GROUP BY customer_id HAVING 1 < SUM(1);
..will break with the error Every derived table must have its own alias. To fix:
SELECT DISTINCT customer_id, SUM(1)
FROM ( SELECT DISTINCT customer_id, store_id FROM purchases) AS custom
GROUP BY customer_id HAVING 1 < SUM(1);
( Note the AS custom alias).
I arrived here because I thought I should check in SO if there are adequate answers, after a syntax error that gave me this error, or if I could possibly post an answer myself.
OK, the answers here explain what this error is, so not much more to say, but nevertheless I will give my 2 cents, using my own words:
This error is caused by the fact that you basically generate a new table with your subquery for the FROM command.
That's what a derived table is, and as such, it needs to have an alias (actually a name reference to it).
Given the following hypothetical query:
SELECT id, key1
FROM (
SELECT t1.ID id, t2.key1 key1, t2.key2 key2, t2.key3 key3
FROM table1 t1
LEFT JOIN table2 t2 ON t1.id = t2.id
WHERE t2.key3 = 'some-value'
) AS tt
At the end, the whole subquery inside the FROM command will produce the table that is aliased as tt and it will have the following columns id, key1, key2, key3.
Then, with the initial SELECT, we finally select the id and key1 from that generated table (tt).
I have a complicated aggregate-functions query that produces a result-set, and which has to be amended with a single row that contains the totals and averages of that result-set.
My idea is to assign an alias to the result-set, and then use that alias in a second query, after a UNION ALL statement.
But, I can't successfully use the alias, in the subsequent SELECT statement, after the UNION ALL statement.
For the sake of simplicity, I won't post the original query here, just a simplified list of the variants I've tried:
SELECT * FROM fees AS Test1 WHERE Percentage = 15
UNION ALL
(SELECT * FROM fees AS Test2 WHERE Percentage > 15)
UNION ALL
(SELECT * FROM (SELECT * FROM fees AS Test3 WHERE Percentage < 10) AS Test4)
UNION ALL
SELECT * FROM Test3
The result is:
MySQL said: Documentation
#1146 - Table 'xxxxxx.Test3' doesn't exist
The result is the same if the last query references to the table Test1, Test2, or Test4.
So, how should I assign an alias to a result-set/derived table in earlier queries and use that same alias in latter queries, all within a UNION query?
Amendment:
My primary query is:
SELECT
COALESCE(referrers.name,order_items.ReferrerID),
SUM(order_items.quantity) as QtySold,
ROUND(SUM((order_items.quantity*order_items.price+order_items.shippingcosts)/((100+order_items.vat)/100)), 2) as TotalRevenueNetto,
ROUND(100*SUM(order_items.quantity*order_items.purchasepricenet)/SUM((order_items.quantity*order_items.price+order_items.shippingcosts)/((100+order_items.vat)/100)), 1) as PurchasePrice,
ROUND(100*SUM(order_items.quantity*COALESCE(order_items.calculatedfee,0)+order_items.quantity*COALESCE(order_items.calculatedcost,0))/SUM((order_items.quantity*order_items.price+order_items.shippingcosts)/((100+order_items.vat)/100)), 1) as Costs,
ROUND(100*SUM(order_items.calculatedprofit) / SUM( (order_items.quantity*order_items.price + order_items.shippingcosts)/((100+order_items.vat)/100) ) , 1) as Profit,
COALESCE(round(100*Returns.TotalReturns_Qty/SUM(order_items.quantity),2),0) as TotalReturns
FROM order_items LEFT JOIN (SELECT order_items.ReferrerID as ReferrerID, sum(order_items.quantity) as TotalReturns_Qty FROM order_items WHERE OrderType='returns' and OrderTimeStamp>='2017-12-1 00:00:00' GROUP BY order_items.ReferrerID) as Returns ON Returns.ReferrerID = order_items.ReferrerID LEFT JOIN `referrers` on `referrers`.`referrerId` = `order_items`.`ReferrerID`
WHERE ( ( order_items.BundleItemID in ('-1', '0') and order_items.OrderType in ('order', '') ) or ( order_items.BundleItemID is NULL and order_items.OrderType = 'returns' ) ) and order_items.OrderTimestamp >= '2017-12-1 00:00:00'
GROUP BY order_items.ReferrerID
ORDER BY referrers.name ASC
I want to make a grand-total of all the rows resulting from query above with:
SELECT 'All marketplaces', SUM(QtySold), SUM(TotalRevenueNetto), AVG(PurchasePrice), AVG(Costs), AVG(Profit), AVG(TotalReturns) FROM PrimaryQuery
I want to do this with a single query.
Your query is well-written. You may be able to get a total line by using a surrounding query with a dummy GROUP BY clause and WITH ROLLUP:
SELECT
COALESCE(Referrer, 'All marketplaces'),
SUM(QtySold) AS QtySold,
SUM(TotalRevenueNetto) AS TotalRevenueNetto,
AVG(PurchasePrice) AS PurchasePrice,
AVG(Costs) AS Costs,
AVG(Profit) AS Profit,
AVG(TotalReturns) AS TotalReturns
FROM
(
SELECT
COALESCE(referrers.name,order_items.ReferrerID) AS Referrer,
SUM(order_items.quantity) AS QtySold,
...
) PrimaryQuery
GROUP BY Referrer ASC WITH ROLLUP;
I'm not entirely sure what you are attempting to solve, but I guess something like the following:
Hypothetical 'main' query:
SELECT T1.ID
, Sum(total_grade)/COUNT(subjects) as AverageGrade
FROM A_Table T1
JOIN AnotherTable T2
ON T2.id = T1.id
GROUP BY T1.ID
You want sub resultsets, without having to keep querying the same data.
Edit: I mistakenly thought the linked documentation and method mentioned below was for the current version of mySQL. It is however a draft for a future version, and CTE's are not currently supported.
In the absence of CTE support, I would probably just insert the resultset into a temporary table. Something like:
CREATE TABLE TEMP_TABLE(ID INT, AverageGrade DECIMAL(15, 3))
INSERT INTO TEMP_TABLE
SELECT T1.ID
, Sum(total_grade)/COUNT(subjects) as AverageGrade
FROM A_Table T1
JOIN AnotherTable T2
ON T2.id = T1.id
GROUP BY T1.ID
SELECT ID, AverageGrade FROM TEMP_TABLE WHERE AverageGrade > 5
UNION ALL
SELECT COUNT(ID) AS TotalCount, SUM(AverageGrade) AS Total_AVGGrade FROM TEMP_TABLE
DROP TABLE TEMP_TABLE
(Disclaimer: I'm not too familiar with mySQL, there may be some syntax errors here. The general idea should be clear, though.)
That is, of course, if i had to do it like this, there are probably better ways to achieve the same. See Thorsten Kettner's comments on the matter.
(Previous answer assuming CTE is a posibility:)
A CTE approach looks like:
WITH CTE AS
(
SELECT T1.ID
, Sum(total_grade)/COUNT(subjects) as AverageGrade
FROM A_Table T1
JOIN AnotherTable T2
ON T2.id = T1.id
GROUP BY T1.ID
)
SELECT ID, AverageGrade FROM CTE WHERE AverageGrade > 5
UNION ALL
SELECT COUNT(ID) AS TotalCount, SUM(AverageGrade) AS Total_AVGGrade FROM CTE
You have the error because every query involved in UNION doens't know the alias of other.
DB Engine execute, in your case, 4 queries and then paste them with UNION operation.
Your real table is fees. Test3 is an alias used in the third query.
If you want to process the results of UNION operation, you must encapsulate your queries in a MAIN query.
It looks like you need something like below. Please try
SELECT * FROM fees AS Test2 WHERE Percentage >= 15
UNION ALL
SELECT * FROM fees AS Test3 WHERE Percentage < 10
You can't use a table alias based on a subquery (is not in the scope of the outer united select) you must repeat the code eg:
SELECT * FROM fees AS Test1 WHERE Percentage = 15
UNION ALL
SELECT * FROM fees AS Test2 WHERE Percentage > 15
UNION ALL
SELECT * FROM (
SELECT * FROM fees AS Test3 WHERE Percentage < 10
) AS Test4
UNION ALL
SELECT * FROM fees AS Test3 WHERE Percentage < 10
Is it possible to combine a result of two cte to another cte. I wrote a query combining two cte. The result gave a three column data in which I want to group the third column and averaging the second column. The second column resulted from a case sum statement.
If you are asking whether you can re-use CTEs after they have been used in a query, the answer is no. You can't do this:
with A
as (
-- query
)
select A.*
from A;
-- this is a separate query
select id
, count(*)
from A
group by
id
You can, however, combine CTEs in all kinds of ways, as long as you do it in a single statement. You can do this, which uses the hypothetical CTE A in two CTEs and the final query:
with A
as (
-- some query
)
, ACustomers
as (
select *
from Customers
join A
on ....
)
, AVendors
as (
select *
from Vendors
join A
on ....
)
select A.StateId
, ACount = COUNT(*)
, CustomerCount = (select count(*) from ACustomers ac where ac.StateId = A.StateId )
, VendorCount = (select count(*) from AVendors av where av.StateId = A.StateId )
from A
group by
A.StateId
Firstly, if it is relevant, I'm using MySQL, though I assume a solution would work across DB products. My problem is thus:
I have a simple table with a single column. There are no constraints on the column. Within this column there is some simple data, e.g.
a
a
b
c
d
d
I need to get the number/count of values that only appear once. From the example above that would be 2 (since only b and c occur once in the column).
Hopefully it's clear I don't want DISTINCT values, but UNIQUE values. I have actually done this before, by creating an additional table with a UNIQUE constraint on the column and simply INSERTing to the new table from the old one, handling the duplicates accordingly.
I was hoping to find a solution that did not require the temporary table, and could somehow just be accomplished with a nifty SELECT.
Assuming your table is called T and your field is called F:
SELECT COUNT(F)
FROM (
SELECT F
FROM T
GROUP BY F
HAVING COUNT(*) = 1
) AS ONLY_ONCE
select count(*) from
(
select
col1, count(*)
from
Table
group by
Col1
Having
Count(Col1) = 1
)
just nest it a little...
select count( cnt ) from
( select count(mycol) cnt from mytab group by mycol )
where cnt = 1
select field1, count(field1) from my_table group by field1 having count(field1) = 1
select count(*) from (select field1, count(field1) from my_table group by field1 having count(field1) = 1)
first one will return the ones that are unique and second one will return the number of unique elements.
Could it be as simple as this:
Select count(*) From MyTable Group By MyColumn Where Count(MyColumn) = 1
This is what I did and it worked:
SELECT name
FROM people JOIN stars ON stars.person_id = people.id
JOIN movies ON movies.id = stars.movie_id
WHERE year = 2004
GROUP BY name, person_id ORDER BY birth;
note: I was working with several tables here.
CS50 Problem Set 7 (pset7) 9.sql fix!!