Last modified on 23 March 2013, at 10:04

Structured Query Language/Data Query Language

← Relational Databases | Data Manipulation Language → Data Query Language is used to extract data from the database. It doesn't modify any data in the database. It describes only one query: SELECT.

SQL data typesEdit

Each column has a type. Here are the standard SQL data types:

Data type Explanation Allowed values Example
VARCHAR(n) A string with a maximum length of n [0-9a-zA-Z]+{n} "foo"
CHAR(n) A string with a fixed length of n [0-9a-zA-Z]{n} "foo"
SMALLINT A 16 bits signed integer \-?[0-9]+ 584
INTEGER A 32 bits signed integer \-?[0-9]+ -8748
FLOAT A decimal floating point \-?[0-9]+[\.[0-9]+]? 48.96
NUMBER(n,[d]) A number with n digits (and d decimal digits if mentioned) \-?[0-9]+[\.[0-9]+]? 484.65
DATE A date (YYYY-MM-DD) [0-9][0-9][0-9][0-9]\-[0-1][0-9]\-[0-3][0-9] 2009-03-24
TIME A time period of sixty minutes; one twenty-fourth of a day [0-2][0-9]\:[0-5][0-9]\:[0-5][0-9] 11:24:56
TIMESTAMP A date and hour [0-9]+ 18648689595962
BLOB Any binary data Any

There is no boolean type. Integers are used instead.

SELECT queryEdit

The exhaustive syntax of the SELECT query is as follows:

SELECT[ ALL| DISTINCT] <COLUMN name>[[ AS] <alias>][,[ ALL| DISTINCT] <COLUMN name>[[ AS] <alias>]]*
FROM <table>[[ AS] <alias>|[[ FULL| LEFT| RIGHT] OUTER| INNER] JOIN <table> ON <expression>]
 [, <table>[[ AS] <alias>|[[ FULL| LEFT| RIGHT] OUTER| INNER] JOIN <table> ON <expression>]]*
 
[WHERE <predicate>[{ AND| OR} <predicate>]*]
[GROUP BY <COLUMN name>[, <COLUMN name>]*
 [HAVING <predicate>[{ AND| OR} <predicate>]]*]
]
[ORDER BY <COLUMN name>[ ASC| DESC][, <COLUMN name>[ ASC| DESC]]*]
[FETCH FIRST <count> ROWS ONLY];

First queryEdit

Let's create the table reunion with many columns:

reunion
id_reunion INTEGER
name VARCHAR(20)
description VARCHAR(255)
priority CHAR(1)
planned SMALLINT
date DATE
hour TIME
duration INTEGER
# id_office INTEGER
pdf_report BLOB

...and let's fill it:

reunion
id_reunion name description priority planned date hour duration # id_office pdf_report
1 Planning We need to plan the project. A 1 2008-03-24 10:30:00 60 35 48644...846348
2 Progress What we have done. C 1 2008-05-12 14:00:00 30 13 9862...15676
3 Change What we need to change in the project. B 1 2008-06-03 9:30:00 90 41 34876...4846548
4 Presentation Presentation of the project. D 0 2008-09-11 15:30:00 120 27
5 Reporting Explanation to the new beginner. B 1 2009-03-15 14:00:00 60 7 19739...37718
6 Learning A new software version has been installed. B 1 2009-09-21 16:00:00 120 11 785278...37528

Let's do a simple query. The following query just returns the content of the reunion table:

  • Query:
SELECT *
FROM reunion;
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|4          |Presentation |Presentation of the project.               |D        |0       |2008-09-11 |15:30:00 |120      |27        |NULL            |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|6          |Learning     |A new software version has been installed. |B        |1       |2009-09-21 |16:00:00 |120      |11        |785278...37528  |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

The form of the result depends on the client application. It can be returned as a text output (backend), a HTML page (thin client), a program object (middleware) etc... The statements, queries, clauses (SELECT, FROM...), instructions and operators are not case sensitive but they are commonly written in uppercase for readability.

The SELECT and FROM clauses are the two required clauses of a SELECT query:

  • FROM : list the tables the query uses to return the data,
  • SELECT : list the data to return.

WHERE clauseEdit

The WHERE clause doesn't influence the columns the query returns but the rows. It filters the rows applying predicates on it. A predicate specifies conditions that can be true or false. SQL can handle conditions whose result is unknown. For example, the following query returns the reunions which have a B priority level:

  • Query:
SELECT *
FROM reunion
WHERE reunion.priority = 'B';
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|6          |Learning     |A new software version has been installed. |B        |1       |2009-09-21 |16:00:00 |120      |11        |785278...37528  |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

The table name can be omitted if it is not ambiguous.

PredicateEdit

Compared to the second operand, the first operand can be :

  • equal : =
  • different : <>
  • lesser : <
  • lesser or equal : <=
  • greater : >
  • greater or equal : >=

The following query returns the reunions which have another priority level than B:

  • Query:
SELECT *
FROM reunion
WHERE priority <> 'B';
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|4          |Presentation |Presentation of the project.               |D        |0       |2008-09-11 |15:30:00 |120      |27        |NULL            |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

OperatorsEdit

The WHERE clause can have several conditions using the operators AND (all the conditions must be true) and OR (only one condition needs to be true). The operator OR is inclusive (several conditions can be true). The order of evaluation can be indicated with brackets. NOT inverts a condition. The following query returns the reunions which have a B priority level and last more than an hour or which take place on 2008/05/12:

  • Query:
SELECT *
FROM reunion
WHERE (priority = 'B' AND NOT duration <= 60) OR DATE = '2008-05-12';
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|6          |Learning     |A new software version has been installed. |B        |1       |2009-09-21 |16:00:00 |120      |11        |785278...37528  |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

LIKEEdit

LIKE allows simplified regular expression matching. It can be applied on the text columns (CHAR, VARCHAR,...).

  • Alphanumerical characters only match identical text,
  • % is a wildcard that matches any text,
  • _ is a wildcard that matches any single character,

The following query returns the reunions which end with "ing" and which contain " the " in its description:

  • Query:
SELECT *
FROM reunion
WHERE name LIKE '%ing' AND description LIKE '% the %';
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

BETWEEN and INEdit

BETWEEN matches a range of values that can be numbers, dates or times. IN matches a list of allowed values. The following query returns the reunions which take place between 2008-04-01 and 2009-04-01 and have an A, B or D priority level:

  • Query:
SELECT *
FROM reunion
WHERE DATE BETWEEN '2008-04-01' AND '2009-04-01' AND priority IN ('A', 'B', 'D');
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|4          |Presentation |Presentation of the project.               |D        |0       |2008-09-11 |15:30:00 |120      |27        |NULL            |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

EXISTSEdit

EXISTS is usually used with a subselect. This predicate is true if the list (i.e. the result set of a subselect) is not empty. This keyword allows to filter the returned rows using data that are not directly associated to the returned rows (i.e. they are not joined, not linked, not related... to the returned rows) so you can not use junction in this case. For instance, we want to retrieve all the reunions for which there is at least one reunion two times longer:

  • Query:
SELECT *
FROM reunion r1
WHERE EXISTS (
  SELECT r2.id_reunion
  FROM reunion r2
  WHERE r2.duration = r1.duration * 2
);
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

The duration of another reunion is used in this query whereas there is no join, no link and no relationship between the two rows. This condition can not be done without EXISTS. Note that the subselect uses the alias r1 whereas this alias is defined in the main query.

EXISTS is also used to match a lack of data. Let's remember the employee table and the members table:

employee
id_employee firstname lastname phone mail
1 Big BOSS 936854270 big.boss@company.com
2 John DOE 936854271 john.doe@company.com
3 Linus TORVALDS 936854272 linus.torvalds@company.com
4 Jimmy WALES 936854273 jimmy.wales@company.com
5 Larry PAGE 936854274 larry.page@company.com
6 Max THE GOOGLER 936854275 max.the-googler@company.com
7 Jenny THE WIKIPEDIAN 936854276 jenny.the-wikipedian@company.com
members
# id_employee # id_project
3 2
2 1
4 3
5 1
2 3
6 1
7 3

The following query returns the employees who are not linked to any project (i.e. the ones there is no relationship for them in the members table):

  • Query:
SELECT *
FROM employees e
WHERE NOT EXISTS (
  SELECT m.id_employee
  FROM members m
  WHERE m.id_employee = e.id_employee
);
  • Result:

|------------------------------------------------------------------|
|id_employee |firstname |lastname |phone     |mail                 |
|------------|----------|---------|----------|---------------------|
|1           |Big       |BOSS     |936854270 |big.boss@company.com |
|------------------------------------------------------------------|

IS NULLEdit

IS NULL tests if a column is filled. It is often used for foreign key columns.

FROM clauseEdit

The FROM clause defines the tables that are used for the query but it can also join tables. A JOIN builds a super table with the columns of two tables to be used for the query. To explain what a join is, we consider two archaic tables without primary keys nor foreign keys:

table_1
common_value specific_value_1
red 9999
grey 6666
white 0000
purple 7777
purple 2222
black 8888
table_2
common_value specific_value_2
green HHHHHH
yellow PPPPPP
black FFFFFF
red OOOOOO
red LLLLLL
blue RRRRRR

We want to associate values from columns of different tables matching values on a given column in each table.

FULL OUTER JOINEdit

A JOIN is made matching a column on a table to a column on the other table. After a FULL OUTER JOIN, for a given value (red), for a given row with this value on one table ([ red | 9999 ]), one row is created for each row that matches on the other table ([ red | OOOOOO ] and [ red | LLLLLL ]). If a value exists in only one table, then a row is created and is completed with NULL columns.

FROM table_1 FULL OUTER JOIN table_2 ON table_1.common_value = table_2.common_value
common_value specific_value_1 specific_value_2
red 9999 OOOOOO
red 9999 LLLLLL
grey 6666 NULL
white 0000 NULL
purple 7777 NULL
purple 2222 NULL
black 8888 FFFFFF
green NULL HHHHHH
yellow NULL PPPPPP
blue NULL RRRRRR

RIGHT OUTER JOINEdit

The RIGHT OUTER JOIN is like the FULL OUTER JOIN but it doesn't create row for values that don't exist on the left table.

FROM table_1 RIGHT OUTER JOIN table_2 ON table_1.common_value = table_2.common_value
common_value specific_value_1 specific_value_2
red 9999 OOOOOO
red 9999 LLLLLL
black 8888 FFFFFF
green NULL HHHHHH
yellow NULL PPPPPP
blue NULL RRRRRR

LEFT OUTER JOINEdit

The LEFT OUTER JOIN is like the FULL OUTER JOIN but it doesn't create row for values that don't exist on the right table.

FROM table_1 LEFT OUTER JOIN table_2 ON table_1.common_value = table_2.common_value
common_value specific_value_1 specific_value_2
red 9999 OOOOOO
red 9999 LLLLLL
grey 6666 NULL
white 0000 NULL
purple 7777 NULL
purple 2222 NULL
black 8888 FFFFFF

INNER JOINEdit

The INNER JOIN is like the FULL OUTER JOIN but it creates row only for values that exist on both the left table and the right table.

FROM table_1 INNER JOIN table_2 ON table_1.common_value = table_2.common_value
common_value specific_value_1 specific_value_2
red 9999 OOOOOO
red 9999 LLLLLL
black 8888 FFFFFF

AliasEdit

The FROM clause can declare several tables, separated by , and aliases can be defined for table name with the keyword AS, which allows the user to make several joins with the same tables. The following query is equivalent to the INNER JOIN above:

  • Query:
SELECT *
FROM table_1 AS t1, table_2 AS t2
WHERE t1.common_value = t2.common_value

The keyword AS can be omitted.

SELECT clauseEdit

The SELECT clause doesn't influence the data processed by the query but the data returned to the user. * return all the data processed after joining and filtering. Otherwise, the SELECT clause lists expressions separated by ,.

The expressions can be a table name, a table name and a column name separated by a dot or simply a column name if it is not ambiguous. The SELECT clause also allows evaluated expressions like addition, subtraction, concatenation, ... An expression can be followed by an alias with the keyword AS. The keyword AS can be omitted.

Here is an example:

  • Query:
SELECT reunion.id_reunion, concat(name, ' : ', reunion.description) n, priority AS p, planned * 10 AS plan, duration + 10 AS reunion_length
FROM reunion;
  • Result:
|-------------------------------------------------------------------------------------------|
|id_reunion |n                                                     |p |plan |reunion_length |
|-----------|------------------------------------------------------|--|-----|---------------|
|1          |Planning : We need to plan the project.               |A |10   |70             |
|2          |Progress : What we have done.                         |C |10   |40             |
|3          |Change : What we need to change in the project.       |B |10   |100            |
|4          |Presentation : Presentation of the project.           |D |0    |130            |
|5          |Reporting : Explanation to the new beginner.          |B |10   |70             |
|6          |Learning : A new software version has been install... |B |10   |130            |
|-------------------------------------------------------------------------------------------|

The expressions can be also the following aggregation functions:

  • count(*): the count of rows returned,
  • max(<column_name>): the greatest value of the column,
  • min(<column_name>): the lowest value of the column.

Here is a new example:

  • Query:
SELECT COUNT(*) * 10 AS c, MAX(DATE) AS latest_date, MIN(reunion.DATE) oldest_date
FROM reunion;
  • Result:
|-----------------------------|
|c  |latest_date |oldest_date |
|---|------------|------------|
|60 |2009-09-21  |2008-03-24  |
|-----------------------------|

ORDER BY clauseEdit

The ORDER BY clause sorts the rows returned by the query by one or several columns. The sort is done with the first column mentioned. The second column is used to sort the rows which have the same value in the first column and so on. The keywords ASC or DESC can be added after each column. ASC indicates an ascending sort. DESC indicates a descending sort. Default is a descending sort. Let's do two simple requests, the first sorting by only one column and the second sorting by two columns:

  • Query:
SELECT *
FROM reunion
ORDER BY priority ASC;
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|6          |Learning     |A new software version has been installed. |B        |1       |2009-09-21 |16:00:00 |120      |11        |785278...37528  |
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|4          |Presentation |Presentation of the project.               |D        |0       |2008-09-11 |15:30:00 |120      |27        |NULL            |
|----------------------------------------------------------------------------------------------------------------------------------------------------|
  • Query:
SELECT *
FROM reunion
ORDER BY priority ASC, duration DESC;
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|6          |Learning     |A new software version has been installed. |B        |1       |2009-09-21 |16:00:00 |120      |11        |785278...37528  |
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|5          |Reporting    |Explanation to the new beginner.           |B        |1       |2009-03-15 |14:00:00 |60       |7         |19739...37718   |
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|4          |Presentation |Presentation of the project.               |D        |0       |2008-09-11 |15:30:00 |120      |27        |NULL            |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

GROUP BY clauseEdit

The GROUP BY clause is used for aggregation operations. It gathers the rows into groups, for instance, all the rows that have the same value in a given column. After gathering rows into groups, any aggregation operation is applied on each group instead of a unique big group of rows. As a consequence, an aggregation operation will return as many result as the number of groups. Groups can be formed with all the rows that have the same value for a given column or the same combination of values for several given columns. For instance, we want to know the number of reunions for each type of priority:

  • Query:
SELECT COUNT(*) AS NUMBER, priority
FROM reunion
GROUP BY priority;
  • Result:
|-----------------|
|number |priority |
|-------|---------|
|1      |A        |
|3      |B        |
|1      |C        |
|1      |D        |
|-----------------|

Due to the GROUP BY clause, the aggregation function count(*) doesn't return a global count but a count for each priority level (A, B, C and D).

  • Query:
SELECT COUNT(*) AS NUMBER, planned, duration
FROM reunion
GROUP BY planned, duration;
  • Result:
|--------------------------|
|number |planned |duration |
|-------|--------|---------|
|1      |0       |120      |
|1      |1       |30       |
|2      |1       |60       |
|1      |1       |90       |
|1      |1       |120      |
|--------------------------|

Note that there are four groups with 1 for the column planned and there are two groups with 120 for the column duration. However, you can see that there is no group with the same combination of values from the two columns.

HAVING clauseEdit

The HAVING clause is used with the GROUP BY clause. The HAVING clause contains a predicate and removes from the returned rows the groups for which the predicate is false. For example, we want to retrieve only the priorities for which there are at least two reunions with the same priority level:

  • Query:
SELECT priority
FROM reunion
GROUP BY priority
HAVING COUNT(*) > 1;
  • Result:
|---------|
|priority |
|---------|
|B        |
|---------|

FETCH FIRST clauseEdit

The FETCH FIRST clause is used to limit the number of returned rows. Only the first rows are returned. The number of returned rows is the number indicated in the clause.

  • Query:
SELECT *
FROM reunion
FETCH FIRST 4 ROWS ONLY;
  • Result:
|----------------------------------------------------------------------------------------------------------------------------------------------------|
|id_reunion |name         |description                                |priority |planned |date       |hour     |duration |id_office |pdf_report      |
|-----------|-------------|-------------------------------------------|---------|--------|-----------|---------|---------|----------|----------------|
|1          |Planning     |We need to plan the project.               |A        |1       |2008-03-24 |10:30:00 |60       |35        |48644...846348  |
|2          |Progress     |What we have done.                         |C        |1       |2008-05-12 |14:00:00 |30       |13        |9862...15676    |
|3          |Change       |What we need to change in the project.     |B        |1       |2008-06-03 |9:30:00  |90       |41        |34876...4846548 |
|4          |Presentation |Presentation of the project.               |D        |0       |2008-09-11 |15:30:00 |120      |27        |NULL            |
|----------------------------------------------------------------------------------------------------------------------------------------------------|

This clause is often used not to return useless rows for test or to improve the performance.

Now you can explore all the data of an already existing database.

SQL FunctionsEdit

  • COUNT
  • AVG
  • MIN
  • MAX
  • SUM

Eg:

SELECT '''COUNT(*)''' FROM reunion

returns the number of rows in the table reunion.

---

  • See also: [[1]]