Advanced SQL
Features
Immanuel Trummer
[email protected]
www.itrummer.org
Database Management
Systems (DBMS)
Application 1 Connections, Security, Utilities, ...
Application 2
DBMS Interface Query Processor
Query Parser Query Rewriter
... Query Optimizer Query Executor
Storage Manager
Data Access Buffer Manager
Transaction Manager Recovery Manager
Slides by Immanuel Trummer, Cornell University
Data
Reminder: Queries So Far
SELECT ...
FROM ...
WHERE ...
GROUP BY ...
HAVING ...
Slides by Immanuel Trummer, Cornell University
Two More Features
SELECT ...
FROM ...
WHERE ...
GROUP BY ...
HAVING ...
ORDER BY ...
LIMIT ...
Slides by Immanuel Trummer, Cornell University
Syntax for Ordering
• ORDER BY <order-item-list>
• <order-item> : <column> <direction>
• <direction> : either ASC or DESC
• Orders result rows by values in order items
• Prioritize order items that appear earlier in list
• Applied after grouping (for group-by queries)
• Items must have unique value per group
Slides by Immanuel Trummer, Cornell University
Limiting Output Size
• Limit <Number> : only shows first <Number> result rows
Slides by Immanuel Trummer, Cornell University
Unknown Values
• Unknown values are called NULL values in SQL
• SQL uses Ternary (i.e., Three-Valued Logic)
• Outcome may be true, false, or unknown
• Check for corresponding outcome
• <expression> = TRUE
• <expression> = FALSE
• <expression> IS NULL (not: "= NULL")
• WHERE condition evaluates to NULL - no result row!
Slides by Immanuel Trummer, Cornell University
Exercise (5 Minutes)
• Guess (or try in PostgreSQL) the results:
• SELECT 3 = NULL
• SELECT NULL = NULL
• SELECT NULL IS NULL
• SELECT NULL IS NOT NULL
• SELECT TRUE OR NULL
• SELECT TRUE AND NULL
Slides by Immanuel Trummer, Cornell University
Joins with Unknowns I
• Standard join keeps only matching row pairs
• Eliminates rows without matching rows in other table
• Sometimes we want to keep rows regardless
• Can do that with OUTER JOINs
• Fills up fields in missing row with NULL values
Slides by Immanuel Trummer, Cornell University
Joins with Unknowns II
• Keep each row in left table (plus standard join result):
• <table-1> LEFT OUTER JOIN <table-2> ON ...
• Keep each row in right table (plus standard result):
• <table-1> RIGHT OUTER JOIN <table-2> ON ...
• Keep rows in both tables (plus standard result):
• <table-1> FULL OUTER JOIN <table-2> ON
Slides by Immanuel Trummer, Cornell University
Outer Join Example
Database Relations:
Students(Sid, Sname)
Enrollment(Sid, Cid)
Courses(Cid, Cname)
SELECT Sname, Count(*)
FROM Students JOIN Enrollment
ON (Students.sid = Enrollment.sid)
GROUP BY Sname
Want to count number of enrollments per student
Slides by Immanuel Trummer, Cornell University
Outer Join Example
Database Relations:
Students(Sid, Sname)
Enrollment(Sid, Cid)
Courses(Cid, Cname)
Will not consider students
without enrollments!
SELECT Sname, Count(*)
FROM Students JOIN Enrollment
ON (Students.sid = Enrollment.sid)
GROUP BY Sname
Want to count number of enrollments per student
Slides by Immanuel Trummer, Cornell University
Outer Join Example
Database Relations:
Students(Sid, Sname)
Enrollment(Sid, Cid)
Courses(Cid, Cname)
Will count one row for
students without enrollments!
SELECT Sname, Count(*)
FROM Students LEFT OUTER JOIN Enrollment
ON (Students.sid = Enrollment.sid)
GROUP BY Sname
Want to count number of enrollments per student
Slides by Immanuel Trummer, Cornell University
Outer Join Example
Database Relations:
Students(Sid, Sname)
Enrollment(Sid, Cid)
Courses(Cid, Cname)
Count only students
matched against courses
SELECT Sname, Count(cid)
FROM Students LEFT OUTER JOIN Enrollment
ON (Students.sid = Enrollment.sid)
GROUP BY Sname
Want to count number of enrollments per student
Slides by Immanuel Trummer, Cornell University
Set Operations
• Union result tuples from two queries
• <query-1> UNION <query-2> : eliminates duplicates
• <query-1> UNION ALL <query-2> : keep duplicates
• Intersect results from two queries
• <query-1> INTERSECT <query-2>
• Set difference between queries
• <query-1> EXCEPT <query-2>
• Results from <query-1> and <query-2> must be union-compatible
Slides by Immanuel Trummer, Cornell University
Query Nesting
• Can use queries as part of another query, e.g.
• Query instead of table in FROM clause,
• Query instead of conjunct in WHERE clause,
• ...
• Query (containing query) vs. sub-query (contained query)
• Correlated vs. uncorrelated sub-queries
• Correlated sub-queries reference containing query
Slides by Immanuel Trummer, Cornell University
Nesting Examples
Database Relations:
Students(Sid, Sname, gpa)
(Not yet exciting)
SELECT SQ.Sname FROM
(SELECT Sname FROM Students) AS SQ
Assign
name for sub-queries in
FROM clause
Slides by Immanuel Trummer, Cornell University
Nesting Examples
Database Relations:
Students(Sid, Sname, gpa)
SELECT Sname FROM Students
WHERE gpa >= (SELECT MAX(gpa) FROM Students)
Slides by Immanuel Trummer, Cornell University
Sub-Queries in Conditions
• Check if sub-query result is empty
• EXISTS(<sub-query>) : TRUE if non-empty
• Check if sub-query result contains value
• <value> IN (<sub-query>) : TRUE if contained
• Check if condition holds for all/some sub-query rows
• E.g., <value> >= ALL(<sub-query>) : TRUE if satisfied for all
• E.g., <value> >= ANY(<sub-query>) : TRUE if satisfied for some
Slides by Immanuel Trummer, Cornell University
Nesting Examples
Database Relations:
Students(Sid, Sname, gpa)
SELECT Sname FROM Students
WHERE gpa >= ALL(SELECT gpa FROM Students)
What does this do?
Slides by Immanuel Trummer, Cornell University
Correlated Sub-Queries
• So far: have seen uncorrelated sub-queries
• Uncorrelated sub-queries are a bit "easier"
• Correlated sub-queries: sub-query refers to the "outside"
Slides by Immanuel Trummer, Cornell University
Nesting Examples
Database Relations:
Students(Sid, Sname, gpa)
SELECT S1.Sname FROM Students S1 WHERE S1.gpa >=
ALL(SELECT S2.gpa FROM Students S2
WHERE S1.Sname = S2.Sname)
What does this do?
Slides by Immanuel Trummer, Cornell University
Evaluating
Correlated Sub-Queries
• Iterate over rows from outer (containing) query
• Evaluate sub-query for fixed row in outer query
• (Decide whether outer row belongs into result)
Slides by Immanuel Trummer, Cornell University
Nesting Examples
Database Relations:
Students(Sid, Sname, gpa)
SELECT S1.Sname FROM Students S1 WHERE
EXISTS (SELECT S2.gpa FROM Students S2
WHERE S1.gpa < S2.gpa)
Names of all
students except for students
with highest gpa
Slides by Immanuel Trummer, Cornell University
Multiple Nesting Levels
Database Relations:
Students(Sid, Sname)
Enrollment(Sid, Cid)
Courses(Cid, Cname)
SELECT C.Cname FROM Courses C WHERE NOT EXISTS (
SELECT * FROM Students S WHERE NOT EXISTS(
SELECT * FROM Enrollment E
WHERE E.cid = C.cid AND E.sid = S.sid
)
)
What does this do ... ?
Slides by Immanuel Trummer, Cornell University
