SQL. SQL Queries. CISC437/637, Lecture #8 Ben Cartere9e. Basic form of a SQL query: 3/17/10

Transcription

1 SQL CISC437/637, Lecture #8 Ben Cartere9e Copyright Ben Cartere9e 1 SQL Queries Basic form of a SQL query: SELECT [DISTINCT] target FROM relations WHERE qualification target is a list of a9ributes/fields rela'ons is a list of relajons to get data from qualifica'on is a a set of comparisons joined with AND/OR/NOT DISTINCT is an opjonal keyword for dropping duplicates Copyright Ben Cartere9e 2 1

2 Conceptual EvaluaJon A simple evaluajon strategy for SQL queries: Compute cross product of rela'ons Discard records that do not match qualifica'on Delete a9ributes not in the target If DISTINCT is specified, drop duplicates This works for any basic query But it is usually the least efficient way to evaluate a query Query opjmizajon finds be9er ways to compute the same answer Copyright Ben Cartere9e 3 ProjecJon in SQL: SELECT SELECT A1, A2, FROM RelaJon Use SELECT * to skip projecjon (get all fields) Perform generalized projecjon on numeric fields by including arithmejc expressions or funcjons SELECT c+a1 from RelaJon SELECT A1*A2 from RelaJon String funcjons can be applied to character fields SELECT upper(a1) from RelaJon SELECT substring(a1, a, b) from RelaJon Copyright Ben Cartere9e 4 2

3 SelecJon in SQL: WHERE SELECT A1, A2, FROM RelaJon WHERE P P is a sentence of comparison operators on fields A1, A2, and constants, linked with AND, OR, NOT Other useful keywords: A1 BETWEEN c AND d A1 >= c AND A1 <= d A1 LIKE str% match strings starjng with str Copyright Ben Cartere9e 5 Cross- Product in SQL: FROM SELECT R1.A1, R1.A2,, R2.B1, R2.B2, FROM RelaJon1 R1, RelaJon2 R2, WHERE P Usually you d want a natural join Specify R1.foreignKey = R2.primaryKey in P Copyright Ben Cartere9e 6 3

4 Renaming in SQL: AS SELECT R1.A1 AS a, R1.A2 as b,, R2.B1 as c, FROM RelaJon1 R1, RelaJon2 R2, WHERE a = c AND b > 100 Useful for defining shorter names to refer to fields and for defining column names for arithmejc expressions in generalized projecjon and for defining column names for funcjons Copyright Ben Cartere9e 7 SorJng SELECT * FROM R1, R2, WHERE R1.fKey = R2.pKey AND ORDER BY R1.A1 Return results sorted in increasing order of R1.A1 Specify muljple fields to break Jes ORDER BY R1.A1, R2.B1, Specify DESC or ASC for decreasing or increasing ORDER BY R1.A1 DESC, R2.B1 ASC, SorJng can be expensive! Copyright Ben Cartere9e 8 4

5 Set Union in SQL: UNION (SELECT A1, A2 FROM R1 WHERE P) UNION (SELECT A1, A2 FROM R2 WHERE Q) The two resuljng relajons must be union- compajble Same number of fields, same domains Use UNION ALL to keep duplicates Copyright Ben Cartere9e 9 Set IntersecJon in SQL: INTERSECT (SELECT A1, A2 FROM R1 WHERE P) INTERSECT (SELECT A1, A2 FROM R2 WHERE Q) The two resuljng relajons must be union- compajble Same number of fields, same domains Use INTERSECT ALL to keep duplicates MySQL does not support this Copyright Ben Cartere9e 10 5

6 Set Difference in SQL: EXCEPT (SELECT A1, A2 FROM R1 WHERE P) EXCEPT (SELECT A1, A2 FROM R2 WHERE Q) The two resuljng relajons must be union- compajble Same number of fields, same domains Use EXCEPT ALL to keep duplicates MySQL does not support this Copyright Ben Cartere9e 11 SELECT f1(a1), f2(a2), FROM RelaJon GROUP BY A3 AggregaJon in SQL f1, f2, are funcjons that compute a quanjty over all values of A1, A2, AVG, MIN, MAX, SUM, COUNT are in the SQL standard GROUP BY is used to calculate aggregate funcjons for values of A1, A2 within groups uniquely defined by values of A3 Copyright Ben Cartere9e 12 6

7 More on AggregaJon SELECT COUNT(DISTINCT A1) FROM R Return the number of unique values of A1 SELECT A1, AVG(A2) FROM R GROUP BY A1 HAVING AVG(A2) > 3 Only select groups where the average A2 is more than three Precedence order: First apply WHERE to select records Then GROUP BY to form the groups Then HAVING to select the groups Copyright Ben Cartere9e 13 Nested Subqueries SELECT * FROM R1 WHERE A1 IN (SELECT A1 FROM R2 WHERE P) EvaluaJon: For each record in R1, compute subquery and check whether value of A1 is in the result P can include condijons on R1 even if R1 not involved in subquery IN can become NOT IN, EXISTS, UNIQUE EXISTS = check whether subquery returns any records UNIQUE = check whether subquery has any duplicates Copyright Ben Cartere9e 14 7

8 Nested Subquery Comparisons SELECT * FROM R1 WHERE A1 > ANY (SELECT A2 FROM R2 WHERE P) Test whether the value of A1 is greater than some value of A2 in the subquery Any comparison operator ANY can be replaced by ALL or SOME ALL equivalent to A1 greater than every value of A2 SOME equivalent to ANY Copyright Ben Cartere9e 15 Set IntersecJon Revisited Implement INTERSECT using nested subqueries SELECT * FROM RelaJon1 R1, RelaJon2 R2 WHERE R1.fKey = R2.pKey AND R2.B1 = b AND R1.A1 IN (SELECT R3.A1 FROM RelaJon1 R3, RelaJon2 R4 WHERE R3.fKey = R4.fKey AND R2.B1 = c) Copyright Ben Cartere9e 16 8

9 Set Difference Revisited Implement INTERSECT using nested subqueries SELECT * FROM RelaJon1 R1, RelaJon2 R2 WHERE R1.fKey = R2.pKey AND R2.B1 = b AND R1.A1 NOT EXISTS (SELECT R3.A1 FROM RelaJon1 R3, RelaJon2 R4 WHERE R3.fKey = R4.fKey AND R2.B1 = c) Copyright Ben Cartere9e 17 Set Division in SQL SELECT * FROM Students S WHERE NOT EXISTS ( (SELECT C.Course FROM Course C) EXCEPT (SELECT E.Course FROM Enrolled E WHERE E.sid = S.sid) ) Copyright Ben Cartere9e 18 9

10 Outer Joins Recall the idea of an outer join: Joining R1 and R2 on condijon P Preserve records in R1 or R2 or both even if there are no matching records in the other relajon Fill in missing values with NULLs Ley outer join: preserve records in R1 Right outer join: preserve records in R2 Full outer join: preserve records in both Copyright Ben Cartere9e 19 Outer Joins in SQL SELECT * FROM R1 LEFT OUTER JOIN R2 ON R1.fKey = R2.pKey LEFT can be replaced with RIGHT MySQL doesn t support full outer joins Specify NATURAL LEFT OUTER JOIN for a natural join (no ON clause required) Use WHERE to select rows NULL values processed according to three- value logic rules Copyright Ben Cartere9e 20 10

11 Integrity Constraints An IC describes condijons that every legal instance of a relajon must sajsfy Inserts/deletes/updates that violate ICs are forbidden ICs can be used to ensure applicajon semanjcs or prevent inconsistencies Domain constraints, primary key constraints, foreign key constraints, general constraints Copyright Ben Cartere9e 21 General Constraints General constraints can be used to ensure field values have a parjcular format or fall in a given range CREATE TABLE Students (sid INTEGER, name CHAR(30), gpa REAL, age INTEGER, PRIMARY KEY (sid), CHECK (gpa >= 0 AND gpa <= 4)) CHECK not supported by MySQL, but GCs can sjll be enforced Copyright Ben Cartere9e 22 11

12 General Constraints Over Two Tables Suppose we want to ensure that there are at most 20 students for every course CREATE TABLE Students (sid INTEGER, CHECK ( (SELECT COUNT(*) FROM Students)/ (SELECT COUNT(*) FROM Courses) <= 20)) This doesn t work! Copyright Ben Cartere9e 23 AsserJons AsserJons allow complex general constraints CREATE ASSERTION smallclass CHECK( (SELECT COUNT(*) FROM Students)/ (SELECT COUNT(*) FROM Courses) <= 20) Every asserjon will be checked with every modificajon of data This can get very expensive if there are many asserjons Use sparingly Copyright Ben Cartere9e 24 12

13 Triggers A trigger is a procedure that automajcally starts if some specified change occurs Triggers consist of three parts: Ac(va(on event, a change to the database that begins the trigger Test condi(on to determine whether trigger should run Ac(on to execute when the trigger is acjvated and the test condijon is true Copyright Ben Cartere9e 26 Trigger Complexity Recursive triggers The acjon in trigger A can acjvate other triggers One of those other triggers can then acjvate trigger A Endless loop of triggering DBMS does not process triggers in any predictable order MulJple triggers can have unpredictable effects Copyright Ben Cartere9e 27 13