G64DBS Database Systems. Lecture 6 More SQL Data. Creating Relations. Column Constraints. Last Lecture s Question

Size: px

Start display at page:

Download "G64DBS Database Systems. Lecture 6 More SQL Data. Creating Relations. Column Constraints. Last Lecture s Question"

Melanie Gardner
5 years ago
Views:

1 G6DBS Database Systems Lecture 6 More SQL Data Tim Brailsford Creating Relations From last lecture.. CREATE TABLE CREATE TABLE <name> ( <column definitions>, <constraints> ) Column definitions are made up of a:!, type (domain), and any column constraints you want to add. Last Lecture s Question CREATE TABLE Write ( the SQL that would Create a new table called stu INT, which NOT should NULL, have the following columns: stu VARCHAR(50), NOT NULL, stu, an integer that will be the primary key stuaddress VARCHAR(00), NOT NULL, stu, stu a string INT DEFAULT(), of up to 50 characters NOT NULL, stuaddress, CONSTRAINT a string pk of up to 00 characters stu, an PRIMARY integer that KEY defaults (stu), onto one ) Column Constraints NOT NULL PRIMARY KEY [AUTO_INCREMENT] FOREIGN KEY UNIQUE DEFAULT value CHECK (expession) COLLATE collation-name What can a domain be? A relational dataype can be ANYTHING numbers, words, objects, elephants, cucumbers anything. This is because you can put anything into a set, and relations are just sets. There are standard datatypes all SQL databases use:! e.g. INT, VARCHAR, TIMESTAMP Specific databases have other types built in:! e.g. NUMBER, POLYGONS, BLOBs. Consider the muppets... There is no reason that a Muppet relation couldn t be as follows: What can a domain be? (varchar) Kermit Miss Piggy Fozzy Picture (jpeg) Colour (colour) Animal (species) frog pig bear

2 It s still just a set of tuples: Muppets = {! (Kermit,,, frog)! (Miss Piggy,,, pig)! (Fozzy,,, bear) } To delete a table use DROP TABLE [IF EXISTS] <name> Example: DROP TABLE Module Deleting Tables Be EXTREMELY careful with any SQL statement with DROP in it! You will destroy any information in the table as well as the table itself You won t normally be asked to confirm There is no easy way to undo the changes - this is PERMANENT Changing Tables Sometimes you want to change the structure of an existing table One way is to DROP it then rebuild it This is dangerous, so there is the ALTER TABLE command instead ALTER TABLE can: Add a new column Remove an existing column Add a new constraint Remove an existing constraint ALTERing Columns To add or remove columns use: ALTER TABLE <table> ADD COLUMN <col> ALTER TABLE <table> DROP COLUMN <name> Examples: ALTER TABLE ADD COLUMN Degree VARCHAR(50) ALTER TABLE DROP COLUMN Degree ALTERing Constraints Examples: ALTER TABLE Module ADD CONSTRAINT ck UNIQUE (title) ALTER TABLE Module DROP CONSTRAINT ck INSERT, UPDATE, DELETE INSERT - add a row to a table UPDATE - change row(s) in a table DELETE - remove row(s) from a table UPDATE and DELETE use WHERE clauses to specify which rows to change or remove Be very CAREFUL with these - an incorrect WHERE clause can destroy a lot of data

3 INSERT INSERT example: The number of columns and values must be the same If you are adding a value to every column, you don t have to list them SQL doesn t require that all rows are different (unless a constraint says so) INSERT INTO <table> (col, col, ) VALUES (val, val, ) INSERT into Muppets VALUES ( Gordon Brown, White, Human ); Note that you have to remember the correct order for the attributes if you don t specify them, to pair up with the values you are putting in. This can get very confusing. In a database (unordered sets) order should not matter, but in SQL it does a good example of SQL confusion! INSERT illustration INSERT INTO (,, ) VALUES (,, ) INSERT INTO (, ) VALUES (, ) INSERT INTO VALUES (,, ) NULL Duplicates? Remember that a relation (table) is just a set. Every table must have at least one key, in order for each row can be identified. However SQL, not being a perfect implementation, allows duplicates and tables with no primary keys. Duplicates: Just say no! Remember that a relation (table) is just a set. A set such as: {,,, } is illegal an element appears twice. This makes no sense in set theory. Just the same for databases. No row may appear twice you would simply be recording the same fact again, wasting space. Never, ever create duplicate rows. UPDATE All rows where the condition is true have the columns set to the given values If no condition is given all rows are changed so BE CAREFUL Values are constants or can be computed from columns UPDATE <table> SET col = val [,col = val ] [WHERE <condition>]

4 UPDATE illustration UPDATE SET =, = Jane WHERE = UPDATE SET = + Jane UPDATE dangers An update can change one row, or many rows depending on how you specify the statement. The following will change every tuple (row): UPDATE Muppets SET colour = Brown ; So make sure you specify your WHERE statement: UPDATE Muppets SET colour = Brown WHERE name= Fozzy ; DELETE DELETE illustration Removes all rows which satisfy the condition DELETE FROM <table> [WHERE <condition>] If no condition is given then ALL rows are deleted - BE CAREFUL Some versions of SQL also have TRUNCATE TABLE <T> which is like DELETE FROM <T> but it is quicker as it doesn t record its actions DELETE FROM WHERE = DELETE FROM or TRUNCATE TABLE Being Careful Again When using DELETE and UPDATE You need to be careful to have the right WHERE clause You can check it by running a SELECT statement with the same WHERE clause first Before running DELETE FROM WHERE = run: SELECT * FROM WHERE = A table or a table variable? When you create a table you are actually creating a variable, with a name. This is what Codd called a time-dependent relation. Updates, Inserts and Deletes replace the value in that variable.

Back to Set Theory What are INSERT, UPDATE and DELETE actually doing? They replace the whole of the previous table with a whole new table. Or really they replace one relation value with another.

5 Back to Set Theory What are INSERT, UPDATE and DELETE actually doing? They replace the whole of the previous table with a whole new table. Or really they replace one relation value with another. A table is a relation variable (a relvar) and any change to it replaces the whole thing. Replacing Relation Values Muppets = { (Kermit, Green, Frog) (Miss Piggy, Pink, Pig) (Fozzy, Orange, Bear) } INSERT into Muppets VALUES (Robin, Green, Frog); Muppets = { (Kermit, Green, Frog) (Miss Piggy, Pink, Pig) (Fozzy, Orange, Bear) (Robin, Green, Frog) } Update Confusion It is easy to see with inserts and deletes that they are replacing the old relation (table) with a larger or smaller one. However it is vital to realise with updates the system is in theory: Memorizing some row in the old table Altering that memorized row Deleting the original row from the old table Creating a new table with the new altered row added. Rows are not variables. They cannot change. Sequences Often we want each tuple to have a unique number These are useful as primary keys Using integers to reference rows is more efficient We would like the DBMS to do this In most versions of SQL we can use autoincrementing fields to do this: Details differ between versions Usually the first entry is assigned, the next, and so on. AUTO_INCREMENT Auto_increment keeps track of an id number, adding one for each new statement you add to your database. CREATE TABLE s ( Then to add a statement: student_ INT NOT NULL AUTO_INCREMENT, name VARCHAR() NOT NULL, age INT NOT NULL ) INSERT INTO s VALUES (NULL, Gonzo, 5 ) These are NOT Row s row s would be identifiers for specific tuples This would violate Codd s information principle ( All information must be represented by data values ). This would be bad! Fields generated by auto_increment are simply data that is generated by the system Surrogate data Often used as keys - Surrogate keys This is a subtle difference, and one that many database practitioners do not realise. It is not an for a statement of fact (a row), it is a value of an entity in the real world, just as colour or name is.

6 Sequences In Oracle we can also use a Sequence Autoincrements just go up in integers. A sequence is a user-defined source of numbers. We can declare several sequences, giving each a name, a start point, and a step size. We can then generate unique numbers by asking for the next element from a sequence. Sequences in Oracle To declare a sequence: CREATE SEQUENCE <name> [START WITH <value>] [INCREMENT BY <value>] If no START WITH or INCREMENT BY values are given they default to To get the next value from a sequence <sequence name>.nextval Sequence Example Creating a sequence CREATE SEQUENCE myseq START WITH Using a sequence SELECT myseq.nextval FROM DUAL; INSERT INTO (stu, stu, stuaddress) VALUES (myseq.nextval, 'Tim Brailsford', ' Elm Street')

In this Lecture. More SQL Data Definition. Deleting Tables. Creating Tables. ALTERing Columns. Changing Tables. More SQL

In this Lecture. More SQL Data Definition. Deleting Tables. Creating Tables. ALTERing Columns. Changing Tables. More SQL In this Lecture Database Systems Lecture 6 Natasha Alechina More SQL DROP TABLE ALTER TABLE INSERT, UPDATE, and DELETE Data dictionary Sequences For more information Connolly and Begg chapters 5 and 6