SQL Workshop. Subqueries. Doug Shook

Transcription

1 SQL Workshop Subqueries Doug Shook

2 Subqueries A SELECT statement within a SELECT statement Four ways to code one: WHERE (*) HAVING (*) FROM SELECT Syntax is the same Rare to see GROUP BY or HAVING 2

3 Subqueries A subquery in the WHERE clause SELECT InvoiceNumber, InvoiceDate, InvoiceTotal FROM Invoices WHERE InvoiceTotal > (SELECT AVG(InvoiceTotal) FROM Invoices) ORDER BY InvoiceTotal; The value returned by the subquery The result set 3

4 Subqueries vs. Joins Advantages of subqueries Aggregates can be passed to outer query More intuitive than joins for ad-hoc relationships Can be easier to code Advantages of joins Can include columns from both tables More intuitive than subqueries for existing relationships Performance is generally better 4

5 Subqueries vs. Joins What would this look like with a subquery instead? A query that uses an inner join SELECT InvoiceNumber, InvoiceDate, InvoiceTotal FROM Invoices JOIN Vendors ON Invoices.VendorID = Vendors.VendorID WHERE VendorState = 'CA' ORDER BY InvoiceDate; The result set ( 4 0 r o w s ) 5

6 Subqueries vs. Joins The same query restated with a subquery SELECT InvoiceNumber, InvoiceDate, InvoiceTotal FROM Invoices WHERE VendorID IN (SELECT VendorID FROM Vendors WHERE VendorState = 'CA') ORDER BY InvoiceDate; Most queries can be written either way 6

7 Subqueries and WHERE The syntax of a WHERE clause that uses an IN phrase with a subquery WHERE test_expression [NOT] IN (subquery) A query that returns vendors without invoices SELECT VendorID, VendorName, VendorState FROM Vendors WHERE VendorID NOT IN (SELECT DISTINCT VendorID FROM Invoices); Subquery must return a single column of values here Why? Could this be rewritten without the subquery? 7

8 Subqueries and WHERE The query restated without a subquery SELECT Vendors.VendorID, VendorName, VendorState FROM Vendors LEFT JOIN Invoices ON Vendors.VendorID = Invoices.VendorID WHERE Invoices.VendorID IS NULL; The same result set (88 rows) NOT IN can typically be replaced with an outer join. 8

9 Subqueries and WHERE A query with a subquery in the WHERE condition SELECT InvoiceNumber, InvoiceDate, InvoiceTotal, InvoiceTotal - PaymentTotal - CreditTotal AS BalanceDue FROM Invoices WHERE InvoiceTotal - PaymentTotal - CreditTotal > 0 AND InvoiceTotal - PaymentTotal - CreditTotal < (SELECT AVG(InvoiceTotal - PaymentTotal - CreditTotal) FROM Invoices WHERE InvoiceTotal - PaymentTotal - CreditTotal > 0) ORDER BY InvoiceTotal DESC; What is the subquery returning here? 9

10 ALL Can be thought of as a sequence of ANDs: How the ALL keyword works Condition Equivalent expression x > ALL (1, 2) x > 2 x < ALL (1, 2) x < 1 x = ALL (1, 2) (x = 1) AND (x = 2) x <> ALL (1, 2) (x <> 1) AND (x <> 2) Special cases: If no rows are returned, ALL will be true What if all rows are null? 10

11 ALL A query that uses the ALL keyword SELECT VendorName, InvoiceNumber, InvoiceTotal FROM Invoices JOIN Vendors ON Invoices.VendorID = Vendors.VendorID WHERE InvoiceTotal > ALL (SELECT InvoiceTotal FROM Invoices WHERE VendorID = 34) ORDER BY VendorName; The result of the subquery The result set (25 rows) 11

12 ANY and SOME Can be thought of as a sequence of Ors How the ANY and SOME keywords work Condition Equivalent expression x > ANY (1, 2) x > 1 x < ANY (1, 2) x < 2 x = ANY (1, 2) (x = 1) OR (x = 2) x <> ANY (1, 2) (x <> 1) OR (x <> 2) Special cases No rows returned? All null values? 12

13 ANY A query that uses the ANY keyword SELECT VendorName, InvoiceNumber, InvoiceTotal FROM Vendors JOIN Invoices ON Vendors.VendorID = Invoices.VendorID WHERE InvoiceTotal < ANY (SELECT InvoiceTotal FROM Invoices WHERE VendorID = 115); The result of the subquery The result set (17 rows) 13

14 Correlated Subqueries Each subquery we ve seen so far has been executed exactly once Correlated subqueries execute once for each row in the outer query Typically as part of a conditional statement Potential conflicts Same column in outer query and subquery? Same table in outer query and subquery? 14

15 Correlated Subqueries A query that uses a correlated subquery SELECT VendorID, InvoiceNumber, InvoiceTotal FROM Invoices AS Inv_Main WHERE InvoiceTotal > (SELECT AVG(InvoiceTotal) FROM Invoices AS Inv_Sub WHERE Inv_Sub.VendorID = Inv_Main.VendorID) ORDER BY VendorID, InvoiceTotal; The value returned by the subquery for vendor The result set (36 rows) 15

16 EXISTS Tests whether a subquery returns any rows Does not return the rows, even if they exist Most often used with correlated subquery A query that returns vendors without invoices SELECT VendorID, VendorName, VendorState FROM Vendors WHERE NOT EXISTS (SELECT * FROM Invoices WHERE Invoices.VendorID = Vendors.VendorID); The result set (88 rows) 16

17 Subqueries with FROM This type of subquery returns a derived table Must assign it to an alias Can be used in the outer query like any other table Often used to provide additional summaries to a summary query 17

18 Subqueries with FROM A subquery coded in the FROM clause SELECT Invoices.VendorID, MAX(InvoiceDate) AS LatestInv, AVG(InvoiceTotal) AS AvgInvoice FROM Invoices JOIN (SELECT TOP 5 VendorID, AVG(InvoiceTotal) AS AvgInvoice FROM Invoices GROUP BY VendorID ORDER BY AvgInvoice DESC) AS TopVendor ON Invoices.VendorID = TopVendor.VendorID GROUP BY Invoices.VendorID ORDER BY LatestInv DESC; 18

19 Subqueries with FROM The derived table generated by the subquery The result set 19

20 Subqueries with SELECT Used in place of a column specifier Must return a single value Usually correlated Can often be stated with a join instead And should be! 20

21 Subqueries with SELECT A correlated subquery in the SELECT clause SELECT DISTINCT VendorName, (SELECT MAX(InvoiceDate) FROM Invoices WHERE Invoices.VendorID = Vendors.VendorID) AS LatestInv FROM Vendors ORDER BY LatestInv DESC; The result set (122 rows) How could we restate this with a join? 21

22 Subqueries with SELECT The same query restated using a join SELECT VendorName, MAX(InvoiceDate) AS LatestInv FROM Vendors LEFT JOIN Invoices ON Vendors.VendorID = Invoices.VendorID GROUP BY VendorName ORDER BY LatestInv DESC; The same result set ( r o w s ) 22

23 Complex Queries A complex query that uses three subqueries SELECT Summary1.VendorState, Summary1.VendorName, TopInState.SumOfInvoices FROM (SELECT V_Sub.VendorState, V_Sub.VendorName, SUM(I_Sub.InvoiceTotal) AS SumOfInvoices FROM Invoices AS I_Sub JOIN Vendors AS V_Sub ON I_Sub.VendorID = V_Sub.VendorID GROUP BY V_Sub.VendorState, V_Sub.VendorName) AS Summary1 JOIN (SELECT Summary2.VendorState, MAX(Summary2.SumOfInvoices) AS SumOfInvoices FROM (SELECT V_Sub.VendorState, V_Sub.VendorName, SUM(I_Sub.InvoiceTotal) AS SumOfInvoices FROM Invoices AS I_Sub JOIN Vendors AS V_Sub ON I_Sub.VendorID = V_Sub.VendorID GROUP BY V_Sub.VendorState, V_Sub.VendorName) AS Summary2 GROUP BY Summary2.VendorState) AS TopInState ON Summary1.VendorState = TopInState.VendorState AND Summary1.SumOfInvoices = TopInState.SumOfInvoices ORDER BY Summary1.VendorState; 23

24 Complex Queries The result set ( 1 0 r o w s ) Such queries are useful, but can we simplify this process? 24

25 Complex Queries Procedure State the problem to be solved Use pseudocode to outline the query Identify subqueries Identify results from subqueries Include aliases Use pseudocode for each subquery Code the subqueries Test! Combine and test final query 25

26 Complex Queries Pseudocode for the query SELECT Summary1.VendorState, Summary1.VendorName, TopInState.SumOfInvoices FROM (Derived table returning VendorState, VendorName, SumOfInvoices) AS Summary1 JOIN (Derived table returning VendorState, MAX(SumOfInvoices)) AS TopInState ON Summary1.VendorState = TopInState.VendorState AND Summary1.SumOfInvoices = TopInState.SumOfInvoices ORDER BY Summary1.VendorState; 26

27 Complex Queries The code for the Summary1 and Summary2 subqueries SELECT V_Sub.VendorState, V_Sub.VendorName, SUM(I_Sub.InvoiceTotal) AS SumOfInvoices FROM Invoices AS I_Sub JOIN Vendors AS V_Sub ON I_Sub.VendorID = V_Sub.VendorID GROUP BY V_Sub.VendorState, V_Sub.VendorName; The result of the Summary1 and Summary2 subqueries ( 3 4 r o w s ) 27

28 Complex Queries Pseudocode for the TopInState subquery SELECT Summary2.VendorState, MAX(Summary2.SumOfInvoices) FROM (Derived table returning VendorState, VendorName, SumOfInvoices) AS Summary2 GROUP BY Summary2.VendorState; The result of the TopInState subquery ( 1 0 r o w s ) 28

29 Common Table Expressions Allows you to code an expression that defines a derived table This table can be used with the query immediately following the CTE Most often used with SELECT 29

30 Common Table Expressions Two CTEs and a query that uses them WITH Summary AS ( SELECT VendorState, VendorName, SUM(InvoiceTotal) AS SumOfInvoices FROM Invoices JOIN Vendors ON Invoices.VendorID = Vendors.VendorID GROUP BY VendorState, VendorName ), TopInState AS ( SELECT VendorState, MAX(SumOfInvoices) AS SumOfInvoices FROM Summary GROUP BY VendorState ) SELECT Summary.VendorState, Summary.VendorName, TopInState.SumOfInvoices FROM Summary JOIN TopInState ON Summary.VendorState = TopInState.VendorState AND Summary.SumOfInvoices = TopInState.SumOfInvoices ORDER BY Summary.VendorState; 30

31 31 Exercises Write a SELECT statement that returns the same result set as this SELECT statement. Substitute a subquery in a WHERE clause for the inner join. SELECT DISTINCT VendorName FROM Vendors JOIN Invoices ON Vendors.VendorID = Invoices.VendorID ORDER BY VendorName;

32 32 Exercises Write a SELECT statement that answers this question: Which invoices have a PaymentTotal that's greater than the average PaymentTotal for all paid invoices? Return the InvoiceNumber and InvoiceTotal for each invoice.

33 33 Exercises Write a SELECT statement that answers this question: which invoices have a PaymentTotal that's greater than the median PaymentTotal for all paid invoices? Return the InvoiceNumber and InvoiceTotal for each invoice. Hint: Start with the solution to the previous problem then use the ALL keyword in the WHERE clause and code "TOP 50 PERCENT PaymentTotal" in the subquery.

34 34 Exercises Write a SELECT statement that returns two columns from the GLAccounts table: AccountNo and AccountDescription. The result set should have one row for each account number that has never been used. Use a correlated subquery introduced with the NOT EXISTS operator. Sort the final result set by AccountNo.

35 35 Exercises Write a SELECT statement that returns four columns: VendorName, InvoiceID, InvoiceSequence, and InvoiceLineItemAmount for each invoice that has more than one line item in the InvoiceLineItems table. Hint: Use a subquery that tests for InvoiceSequence > 1.

36 36 Exercises Write a SELECT statement that returns a single value that represents the sum of the largest unpaid invoices submitted by each vendor. Use a derived table that reutrns MAX(InvoiceTotal) grouped by VendorID, filtering for invoices with a balance due.

37 37 Exercises Write a SELECT statement that returns the name, city, and state of each vendor that's located in a unique city and state. In other words, don't include vendors that have a city and state in common with another vendor. Write a SELECT statement that returns four columns: VendorName, InvoiceNumber, InvoiceDate, and InvoiceTotal. Return one row per vendor, representing the vendor's invoice with the earliest date.