Databases and SQL - Library edition

Filtering

One of the most powerful features of a database is the ability to filter data, i.e., to select only those records that match certain criteria. For example, suppose we want to list all the books in the catalogue that have been published by Wiley. We can select these records from the Works table by using a WHERE clause in our query:

SELECT * FROM Works WHERE Publisher="Wiley";

The database manager executes this query in two stages. First, it checks at each row in the Works table to see which ones satisfy the WHERE. It then uses the column names following the SELECT keyword to determine what columns to display.

This processing order means that we can filter records using WHERE based on values in columns that aren’t then displayed:

SELECT Title FROM Works WHERE Publisher="Wiley";

We can use many other Boolean operators to filter our data. For example, we can display only titles that have been published by Wiley in 2011 or later:

SELECT * FROM Works WHERE (Publisher='Wiley') AND (Date>='2011');

(The parentheses around the individual tests aren’t strictly required, but they help make the query easier to read.)

If we want to list all books published either by Wiley or O’Reilly, we can combine the query using OR. Note in passing how the apostrophe in the name O’Reilly is escaped:

SELECT * FROM Works WHERE Publisher='Wiley' OR Publisher='O''Reilly';

Alternatively, we can use IN to see if a value is in a specific set:

SELECT * FROM Works WHERE Publisher IN('Wiley','O''Reilly');

We can combine AND with OR, but we need to be careful about which operator is executed first. If we don’t use parentheses, we get this:

SELECT * FROM Works WHERE Publisher='Wiley' OR Publisher='O''Reilly' AND Date>=2011;

which is a list of the books published either by Wiley at any given time and those by O’Reilly in 2011 or later. We probably want this instead:

SELECT * FROM Works WHERE (Publisher='Wiley' OR Publisher='O''Reilly') AND Date>=2011;

We can also filter by partial matches. For example, if we want to know something just about the site names beginning with “DR” we can use the LIKE keyword. The percent symbol acts as a wildcard, matching any characters in that place. It can be used at the beginning, middle, or end of the string:

SELECT * FROM Works WHERE Publisher LIKE "%y";

Finally, we can use DISTINCT with WHERE to give a second level of filtering:

SELECT DISTINCT Date, Publisher FROM Works WHERE (Publisher='Wiley' OR Publisher='O''Reilly') AND Date>=2011;

But remember: DISTINCT is applied to the values displayed in the chosen columns, not to the entire rows as they are being processed.

What we have just done is how most people “grow” their SQL queries. We started with something simple that did part of what we wanted, then added more clauses one by one, testing their effects as we went. This is a good strategy — in fact, for complex queries it’s often the only strategy — but it depends on quick turnaround, and on us recognizing the right answer when we get it.

The best way to achieve quick turnaround is often to put a subset of data in a temporary database and run our queries against that, or to fill a small database with synthesized records. For example, instead of trying our queries against an actual database of 20 million bibliographic records, we could run it against a sample of ten thousand, or write a small program to generate ten thousand random (but plausible) records and use that.