2 $Header: /cvsroot/pgsql/doc/src/sgml/syntax.sgml,v 1.59 2002/03/22 19:20:31 petere Exp $
5 <chapter id="sql-syntax">
6 <title>SQL Syntax</title>
8 <indexterm zone="sql-syntax">
9 <primary>syntax</primary>
10 <secondary>SQL</secondary>
15 This chapter describes the syntax of SQL.
19 <sect1 id="sql-syntax-lexical">
20 <title>Lexical Structure</title>
23 SQL input consists of a sequence of
24 <firstterm>commands</firstterm>. A command is composed of a
25 sequence of <firstterm>tokens</firstterm>, terminated by a
26 semicolon (<quote>;</quote>). The end of the input stream also
27 terminates a command. Which tokens are valid depends on the syntax
28 of the particular command.
32 A token can be a <firstterm>key word</firstterm>, an
33 <firstterm>identifier</firstterm>, a <firstterm>quoted
34 identifier</firstterm>, a <firstterm>literal</firstterm> (or
35 constant), or a special character symbol. Tokens are normally
36 separated by whitespace (space, tab, newline), but need not be if
37 there is no ambiguity (which is generally only the case if a
38 special character is adjacent to some other token type).
42 Additionally, <firstterm>comments</firstterm> can occur in SQL
43 input. They are not tokens, they are effectively equivalent to
47 <informalexample id="sql-syntax-ex-commands">
49 For example, the following is (syntactically) valid SQL input:
51 SELECT * FROM MY_TABLE;
52 UPDATE MY_TABLE SET A = 5;
53 INSERT INTO MY_TABLE VALUES (3, 'hi there');
55 This is a sequence of three commands, one per line (although this
56 is not required; more than one command can be on a line, and
57 commands can usefully be split across lines).
62 The SQL syntax is not very consistent regarding what tokens
63 identify commands and which are operands or parameters. The first
64 few tokens are generally the command name, so in the above example
65 we would usually speak of a <quote>SELECT</quote>, an
66 <quote>UPDATE</quote>, and an <quote>INSERT</quote> command. But
67 for instance the <command>UPDATE</command> command always requires
68 a <token>SET</token> token to appear in a certain position, and
69 this particular variation of <command>INSERT</command> also
70 requires a <token>VALUES</token> in order to be complete. The
71 precise syntax rules for each command are described in the
72 <citetitle>Reference Manual</citetitle>.
75 <sect2 id="sql-syntax-identifiers">
76 <title>Identifiers and Key Words</title>
78 <indexterm zone="sql-syntax-identifiers">
79 <primary>identifiers</primary>
82 <indexterm zone="sql-syntax-identifiers">
83 <primary>key words</primary>
84 <secondary>syntax</secondary>
88 Tokens such as <token>SELECT</token>, <token>UPDATE</token>, or
89 <token>VALUES</token> in the example above are examples of
90 <firstterm>key words</firstterm>, that is, words that have a fixed
91 meaning in the SQL language. The tokens <token>MY_TABLE</token>
92 and <token>A</token> are examples of
93 <firstterm>identifiers</firstterm>. They identify names of
94 tables, columns, or other database objects, depending on the
95 command they are used in. Therefore they are sometimes simply
96 called <quote>names</quote>. Key words and identifiers have the
97 same lexical structure, meaning that one cannot know whether a
98 token is an identifier or a key word without knowing the language.
99 A complete list of key words can be found in <xref
100 linkend="sql-keywords-appendix">.
104 SQL identifiers and key words must begin with a letter
105 (<literal>a</literal>-<literal>z</literal>, but also letters with
106 diacritical marks and non-Latin letters) or an underscore
107 (<literal>_</literal>). Subsequent characters in an identifier or
108 key word can be letters, digits
109 (<literal>0</literal>-<literal>9</literal>), or underscores,
110 although the SQL standard will not define a key word that contains
111 digits or starts or ends with an underscore.
115 The system uses no more than <symbol>NAMEDATALEN</symbol>-1
116 characters of an identifier; longer names can be written in
117 commands, but they will be truncated. By default,
118 <symbol>NAMEDATALEN</symbol> is 32 so the maximum identifier length
119 is 31 (but at the time the system is built,
120 <symbol>NAMEDATALEN</symbol> can be changed in
121 <filename>src/include/postgres_ext.h</filename>).
126 <primary>case sensitivity</primary>
127 <secondary>SQL commands</secondary>
129 Identifier and key word names are case insensitive. Therefore
131 UPDATE MY_TABLE SET A = 5;
133 can equivalently be written as
135 uPDaTE my_TabLE SeT a = 5;
137 A convention often used is to write key words in upper
138 case and names in lower case, e.g.,
140 UPDATE my_table SET a = 5;
146 <primary>quotes</primary>
147 <secondary>and identifiers</secondary>
149 There is a second kind of identifier: the <firstterm>delimited
150 identifier</firstterm> or <firstterm>quoted
151 identifier</firstterm>. It is formed by enclosing an arbitrary
152 sequence of characters in double-quotes
153 (<literal>"</literal>). <!-- " font-lock mania --> A delimited
154 identifier is always an identifier, never a key word. So
155 <literal>"select"</literal> could be used to refer to a column or
156 table named <quote>select</quote>, whereas an unquoted
157 <literal>select</literal> would be taken as a key word and
158 would therefore provoke a parse error when used where a table or
159 column name is expected. The example can be written with quoted
160 identifiers like this:
162 UPDATE "my_table" SET "a" = 5;
167 Quoted identifiers can contain any character other than a double
168 quote itself. This allows constructing table or column names that
169 would otherwise not be possible, such as ones containing spaces or
170 ampersands. The length limitation still applies.
174 Quoting an identifier also makes it case-sensitive, whereas
175 unquoted names are always folded to lower case. For example, the
176 identifiers <literal>FOO</literal>, <literal>foo</literal> and
177 <literal>"foo"</literal> are considered the same by
178 <productname>PostgreSQL</productname>, but <literal>"Foo"</literal>
179 and <literal>"FOO"</literal> are different from these three and
183 The folding of unquoted names to lower case in <productname>PostgreSQL</>
184 is incompatible with the SQL standard, which says that unquoted
185 names should be folded to upper case. Thus, <literal>foo</literal>
186 should be equivalent to <literal>"FOO"</literal> not
187 <literal>"foo"</literal> according to the standard. If you want to
188 write portable applications you are advised to always quote a particular
189 name or never quote it.
196 <sect2 id="sql-syntax-constants">
197 <title>Constants</title>
199 <indexterm zone="sql-syntax-constants">
200 <primary>constants</primary>
204 There are four kinds of <firstterm>implicitly-typed
205 constants</firstterm> in <productname>PostgreSQL</productname>:
206 strings, bit strings, integers, and floating-point numbers.
207 Constants can also be specified with explicit types, which can
208 enable more accurate representation and more efficient handling by
209 the system. The implicit constants are described below; explicit
210 constants are discussed afterwards.
213 <sect3 id="sql-syntax-strings">
214 <title>String Constants</title>
216 <indexterm zone="sql-syntax-strings">
217 <primary>character strings</primary>
218 <secondary>constants</secondary>
223 <primary>quotes</primary>
224 <secondary>escaping</secondary>
226 A string constant in SQL is an arbitrary sequence of characters
227 bounded by single quotes (<quote>'</quote>), e.g., <literal>'This
228 is a string'</literal>. SQL allows single quotes to be embedded
229 in strings by typing two adjacent single quotes (e.g.,
230 <literal>'Dianne''s horse'</literal>). In
231 <productname>PostgreSQL</productname> single quotes may
232 alternatively be escaped with a backslash (<quote>\</quote>,
233 e.g., <literal>'Dianne\'s horse'</literal>).
237 C-style backslash escapes are also available:
238 <literal>\b</literal> is a backspace, <literal>\f</literal> is a
239 form feed, <literal>\n</literal> is a newline,
240 <literal>\r</literal> is a carriage return, <literal>\t</literal>
241 is a tab, and <literal>\<replaceable>xxx</replaceable></literal>,
242 where <replaceable>xxx</replaceable> is an octal number, is the
243 character with the corresponding ASCII code. Any other character
244 following a backslash is taken literally. Thus, to include a
245 backslash in a string constant, type two backslashes.
249 The character with the code zero cannot be in a string constant.
253 Two string constants that are only separated by whitespace
254 <emphasis>with at least one newline</emphasis> are concatenated
255 and effectively treated as if the string had been written in one
256 constant. For example:
269 is not valid syntax, and <productname>PostgreSQL</productname> is
270 consistent with <acronym>SQL9x</acronym> in this regard.
274 <sect3 id="sql-syntax-bit-strings">
275 <title>Bit-String Constants</title>
277 <indexterm zone="sql-syntax-bit-strings">
278 <primary>bit strings</primary>
279 <secondary>constants</secondary>
283 Bit-string constants look like string constants with a
284 <literal>B</literal> (upper or lower case) immediately before the
285 opening quote (no intervening whitespace), e.g.,
286 <literal>B'1001'</literal>. The only characters allowed within
287 bit-string constants are <literal>0</literal> and
288 <literal>1</literal>. Bit-string constants can be continued
289 across lines in the same way as regular string constants.
294 <title>Integer Constants</title>
297 Integer constants in SQL are sequences of decimal digits (0
298 though 9) with no decimal point and no exponent. The range of legal values
299 depends on which integer data type is used, but the plain
300 <type>integer</type> type accepts values ranging from -2147483648
301 to +2147483647. (The optional plus or minus sign is actually a
302 separate unary operator and not part of the integer constant.)
307 <title>Floating-Point Constants</title>
310 <primary>floating point</primary>
311 <secondary>constants</secondary>
315 Floating-point constants are accepted in these general forms:
317 <replaceable>digits</replaceable>.<optional><replaceable>digits</replaceable></optional><optional>e<optional>+-</optional><replaceable>digits</replaceable></optional>
318 <optional><replaceable>digits</replaceable></optional>.<replaceable>digits</replaceable><optional>e<optional>+-</optional><replaceable>digits</replaceable></optional>
319 <replaceable>digits</replaceable>e<optional>+-</optional><replaceable>digits</replaceable>
321 where <replaceable>digits</replaceable> is one or more decimal
322 digits. At least one digit must be before or after the decimal
323 point. At least one digit must follow the exponent delimiter
324 (<literal>e</literal>) if that field is present.
325 Thus, a floating-point constant is distinguished from an integer
326 constant by the presence of either the decimal point or the
327 exponent clause (or both). There must not be a space or other
328 characters embedded in the constant.
333 These are some examples of valid floating-point constants:
345 Floating-point constants are of type <type>DOUBLE
346 PRECISION</type>. <type>REAL</type> can be specified explicitly
347 by using <acronym>SQL</acronym> string notation or
348 <productname>PostgreSQL</productname> type notation:
351 REAL '1.23' -- string style
352 '1.23'::REAL -- PostgreSQL (historical) style
357 <sect3 id="sql-syntax-constants-generic">
358 <title>Constants of Other Types</title>
361 <primary>data types</primary>
362 <secondary>constants</secondary>
366 A constant of an <emphasis>arbitrary</emphasis> type can be
367 entered using any one of the following notations:
369 <replaceable>type</replaceable> '<replaceable>string</replaceable>'
370 '<replaceable>string</replaceable>'::<replaceable>type</replaceable>
371 CAST ( '<replaceable>string</replaceable>' AS <replaceable>type</replaceable> )
373 The string's text is passed to the input conversion
374 routine for the type called <replaceable>type</replaceable>. The
375 result is a constant of the indicated type. The explicit type
376 cast may be omitted if there is no ambiguity as to the type the
377 constant must be (for example, when it is passed as an argument
378 to a non-overloaded function), in which case it is automatically
383 It is also possible to specify a type coercion using a function-like
386 <replaceable>typename</replaceable> ( '<replaceable>string</replaceable>' )
388 but not all type names may be used in this way; see <xref
389 linkend="sql-syntax-type-casts"> for details.
393 The <literal>::</literal>, <literal>CAST()</literal>, and
394 function-call syntaxes can also be used to specify run-time type
395 conversions of arbitrary expressions, as discussed in <xref
396 linkend="sql-syntax-type-casts">. But the form
397 <replaceable>type</replaceable> '<replaceable>string</replaceable>'
398 can only be used to specify the type of a literal constant.
399 Another restriction on
400 <replaceable>type</replaceable> '<replaceable>string</replaceable>'
401 is that it does not work for array types; use <literal>::</literal>
402 or <literal>CAST()</literal> to specify the type of an array constant.
407 <title>Array constants</title>
410 <primary>arrays</primary>
411 <secondary>constants</secondary>
415 The general format of an array constant is the following:
417 '{ <replaceable>val1</replaceable> <replaceable>delim</replaceable> <replaceable>val2</replaceable> <replaceable>delim</replaceable> ... }'
419 where <replaceable>delim</replaceable> is the delimiter character
420 for the type, as recorded in its <literal>pg_type</literal>
421 entry. (For all built-in types, this is the comma character
422 <quote><literal>,</literal></>.) Each <replaceable>val</replaceable> is either a constant
423 of the array element type, or a subarray. An example of an
426 '{{1,2,3},{4,5,6},{7,8,9}}'
428 This constant is a two-dimensional, 3-by-3 array consisting of three
429 subarrays of integers.
433 Individual array elements can be placed between double-quote
434 marks (<literal>"</literal>) <!-- " --> to avoid ambiguity
435 problems with respect to whitespace. Without quote marks, the
436 array-value parser will skip leading whitespace.
440 (Array constants are actually only a special case of the generic
441 type constants discussed in the previous section. The constant
442 is initially treated as a string and passed to the array input
443 conversion routine. An explicit type specification might be
450 <sect2 id="sql-syntax-operators">
451 <title>Operators</title>
453 <indexterm zone="sql-syntax-operators">
454 <primary>operators</primary>
455 <secondary>syntax</secondary>
459 An operator is a sequence of up to <symbol>NAMEDATALEN</symbol>-1
460 (31 by default) characters from the following list:
462 + - * / < > = ~ ! @ # % ^ & | ` ? $
465 There are a few restrictions on operator names, however:
469 <literal>$</> (dollar) cannot be a single-character operator, although it
470 can be part of a multiple-character operator name.
476 <literal>--</literal> and <literal>/*</literal> cannot appear
477 anywhere in an operator name, since they will be taken as the
484 A multiple-character operator name cannot end in <literal>+</> or <literal>-</>,
485 unless the name also contains at least one of these characters:
487 ~ ! @ # % ^ & | ` ? $
489 For example, <literal>@-</literal> is an allowed operator name,
490 but <literal>*-</literal> is not. This restriction allows
491 <productname>PostgreSQL</productname> to parse SQL-compliant
492 queries without requiring spaces between tokens.
499 When working with non-SQL-standard operator names, you will usually
500 need to separate adjacent operators with spaces to avoid ambiguity.
501 For example, if you have defined a left unary operator named <literal>@</literal>,
502 you cannot write <literal>X*@Y</literal>; you must write
503 <literal>X* @Y</literal> to ensure that
504 <productname>PostgreSQL</productname> reads it as two operator names
510 <title>Special Characters</title>
513 Some characters that are not alphanumeric have a special meaning
514 that is different from being an operator. Details on the usage can
515 be found at the location where the respective syntax element is
516 described. This section only exists to advise the existence and
517 summarize the purposes of these characters.
522 A dollar sign (<literal>$</literal>) followed by digits is used
523 to represent the positional parameters in the body of a function
524 definition. In other contexts the dollar sign may be part of an
531 Parentheses (<literal>()</literal>) have their usual meaning to
532 group expressions and enforce precedence. In some cases
533 parentheses are required as part of the fixed syntax of a
534 particular SQL command.
540 Brackets (<literal>[]</literal>) are used to select the elements
541 of an array. See <xref linkend="arrays"> for more information
548 Commas (<literal>,</literal>) are used in some syntactical
549 constructs to separate the elements of a list.
555 The semicolon (<literal>;</literal>) terminates an SQL command.
556 It cannot appear anywhere within a command, except within a
557 string constant or quoted identifier.
563 The colon (<literal>:</literal>) is used to select
564 <quote>slices</quote> from arrays. (See <xref
565 linkend="arrays">.) In certain SQL dialects (such as Embedded
566 SQL), the colon is used to prefix variable names.
572 The asterisk (<literal>*</literal>) has a special meaning when
573 used in the <command>SELECT</command> command or with the
574 <function>COUNT</function> aggregate function.
580 The period (<literal>.</literal>) is used in floating-point
581 constants, and to separate table and column names.
589 <sect2 id="sql-syntax-comments">
590 <title>Comments</title>
592 <indexterm zone="sql-syntax-comments">
593 <primary>comments</primary>
594 <secondary>in SQL</secondary>
598 A comment is an arbitrary sequence of characters beginning with
599 double dashes and extending to the end of the line, e.g.:
601 -- This is a standard SQL92 comment
606 Alternatively, C-style block comments can be used:
609 * with nesting: /* nested block comment */
612 where the comment begins with <literal>/*</literal> and extends to
613 the matching occurrence of <literal>*/</literal>. These block
614 comments nest, as specified in SQL99 but unlike C, so that one can
615 comment out larger blocks of code that may contain existing block
620 A comment is removed from the input stream before further syntax
621 analysis and is effectively replaced by whitespace.
627 <sect1 id="sql-syntax-columns">
628 <title>Columns</title>
631 A <firstterm>column</firstterm>
632 is either a user-defined column of a given table or one of the
633 following system-defined columns:
636 <primary>columns</primary>
637 <secondary>system columns</secondary>
642 <term><structfield>oid</></term>
646 <primary>OID</primary>
648 The object identifier (object ID) of a row. This is a serial number
649 that is automatically added by <productname>PostgreSQL</productname> to all table rows (unless
650 the table was created WITHOUT OIDS, in which case this column is
657 <term><structfield>tableoid</></term>
660 The OID of the table containing this row. This attribute is
661 particularly handy for queries that select from inheritance
662 hierarchies, since without it, it's difficult to tell which
663 individual table a row came from. The
664 <structfield>tableoid</structfield> can be joined against the
665 <structfield>oid</structfield> column of
666 <classname>pg_class</classname> to obtain the table name.
672 <term><structfield>xmin</></term>
675 The identity (transaction ID) of the inserting transaction for
676 this tuple. (Note: A tuple is an individual state of a row;
677 each update of a row creates a new tuple for the same logical row.)
683 <term><structfield>cmin</></term>
686 The command identifier (starting at zero) within the inserting
693 <term><structfield>xmax</></term>
696 The identity (transaction ID) of the deleting transaction,
697 or zero for an undeleted tuple. It is possible for this field
698 to be nonzero in a visible tuple: that usually indicates that the
699 deleting transaction hasn't committed yet, or that an attempted
700 deletion was rolled back.
706 <term><structfield>cmax</></term>
709 The command identifier within the deleting transaction, or zero.
715 <term><structfield>ctid</></term>
718 The tuple ID of the tuple within its table. This is a pair
719 (block number, tuple index within block) that identifies the
720 physical location of the tuple. Note that although the <structfield>ctid</structfield>
721 can be used to locate the tuple very quickly, a row's <structfield>ctid</structfield>
722 will change each time it is updated or moved by <command>VACUUM
724 Therefore <structfield>ctid</structfield> is useless as a long-term row identifier.
725 The OID, or even better a user-defined serial number, should
726 be used to identify logical rows.
734 OIDs are 32-bit quantities and are assigned from a single cluster-wide
735 counter. In a large or long-lived database, it is possible for the
736 counter to wrap around. Hence, it is bad practice to assume that OIDs
737 are unique, unless you take steps to ensure that they are unique.
738 Recommended practice when using OIDs for row identification is to create
739 a unique constraint on the OID column of each table for which the OID will be
740 used. Never assume that OIDs are unique across tables; use the
741 combination of <structfield>tableoid</> and row OID if you need a database-wide
742 identifier. (Future releases of <productname>PostgreSQL</productname> are likely to use a separate
743 OID counter for each table, so that <structfield>tableoid</> <emphasis>must</> be
744 included to arrive at a globally unique identifier.)
748 Transaction identifiers are 32-bit quantities. In a long-lived
749 database it is possible for transaction IDs to wrap around. This
750 is not a fatal problem given appropriate maintenance procedures;
751 see the <citetitle>Administrator's Guide</> for details. However, it is
752 unwise to depend on uniqueness of transaction IDs over the long term
753 (more than one billion transactions).
757 Command identifiers are also 32-bit quantities. This creates a hard
758 limit of 2<superscript>32</> (4 billion) SQL commands within a single transaction.
759 In practice this limit is not a problem --- note that the limit is on
760 number of SQL queries, not number of tuples processed.
765 <sect1 id="sql-expressions">
766 <title>Value Expressions</title>
769 Value expressions are used in a variety of contexts, such
770 as in the target list of the <command>SELECT</command> command, as
771 new column values in <command>INSERT</command> or
772 <command>UPDATE</command>, or in search conditions in a number of
773 commands. The result of a value expression is sometimes called a
774 <firstterm>scalar</firstterm>, to distinguish it from the result of
775 a table expression (which is a table). Value expressions are
776 therefore also called <firstterm>scalar expressions</firstterm> (or
777 even simply <firstterm>expressions</firstterm>). The expression
778 syntax allows the calculation of values from primitive parts using
779 arithmetic, logical, set, and other operations.
783 A value expression is one of the following:
788 A constant or literal value; see <xref linkend="sql-syntax-constants">.
800 A positional parameter reference, in the body of a function declaration.
806 An operator invocation.
818 An aggregate expression.
835 <synopsis>( <replaceable>expression</replaceable> )</synopsis>
837 Parentheses are used to group subexpressions and override precedence.
844 In addition to this list, there are a number of constructs that can
845 be classified as an expression but do not follow any general syntax
846 rules. These generally have the semantics of a function or
847 operator and are explained in the appropriate location in <xref
848 linkend="functions">. An example is the <literal>IS NULL</literal>
853 We have already discussed constants in <xref
854 linkend="sql-syntax-constants">. The following sections discuss
855 the remaining options.
859 <title>Column References</title>
862 A column can be referenced in the form:
864 <replaceable>correlation</replaceable>.<replaceable>columnname</replaceable> `['<replaceable>subscript</replaceable>`]'
867 <replaceable>correlation</replaceable> is either the name of a
868 table, an alias for a table defined by means of a FROM clause, or
869 the key words <literal>NEW</literal> or <literal>OLD</literal>.
870 (NEW and OLD can only appear in the action portion of a rule,
871 while other correlation names can be used in any SQL statement.)
872 The correlation name and separating dot may be omitted if the column name
874 across all the tables being used in the current query. If
875 <replaceable>column</replaceable> is of an array type, then the
876 optional <replaceable>subscript</replaceable> selects a specific
877 element or elements in the array. If no subscript is provided, then the
878 whole array is selected. (See <xref linkend="arrays"> for more about
884 <title>Positional Parameters</title>
887 A positional parameter reference is used to indicate a parameter
888 in an SQL function. Typically this is used in SQL function
889 definition statements. The form of a parameter is:
891 $<replaceable>number</replaceable>
896 For example, consider the definition of a function,
897 <function>dept</function>, as
900 CREATE FUNCTION dept (text) RETURNS dept
901 AS 'SELECT * FROM dept WHERE name = $1'
905 Here the <literal>$1</literal> will be replaced by the first
906 function argument when the function is invoked.
911 <title>Operator Invocations</title>
914 There are three possible syntaxes for an operator invocation:
916 <member><replaceable>expression</replaceable> <replaceable>operator</replaceable> <replaceable>expression</replaceable> (binary infix operator)</member>
917 <member><replaceable>operator</replaceable> <replaceable>expression</replaceable> (unary prefix operator)</member>
918 <member><replaceable>expression</replaceable> <replaceable>operator</replaceable> (unary postfix operator)</member>
920 where the <replaceable>operator</replaceable> token follows the syntax
921 rules of <xref linkend="sql-syntax-operators"> or is one of the
922 tokens <token>AND</token>, <token>OR</token>, and
923 <token>NOT</token>. Which particular operators exist and whether
924 they are unary or binary depends on what operators have been
925 defined by the system or the user. <xref linkend="functions">
926 describes the built-in operators.
931 <title>Function Calls</title>
934 The syntax for a function call is the name of a function
935 (which is subject to the syntax rules for identifiers of <xref
936 linkend="sql-syntax-identifiers">), followed by its argument list
937 enclosed in parentheses:
940 <replaceable>function</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional> )
945 For example, the following computes the square root of 2:
952 The list of built-in functions is in <xref linkend="functions">.
953 Other functions may be added by the user.
957 <sect2 id="syntax-aggregates">
958 <title>Aggregate Expressions</title>
960 <indexterm zone="syntax-aggregates">
961 <primary>aggregate functions</primary>
965 An <firstterm>aggregate expression</firstterm> represents the
966 application of an aggregate function across the rows selected by a
967 query. An aggregate function reduces multiple inputs to a single
968 output value, such as the sum or average of the inputs. The
969 syntax of an aggregate expression is one of the following:
972 <member><replaceable>aggregate_name</replaceable> (<replaceable>expression</replaceable>)</member>
973 <member><replaceable>aggregate_name</replaceable> (ALL <replaceable>expression</replaceable>)</member>
974 <member><replaceable>aggregate_name</replaceable> (DISTINCT <replaceable>expression</replaceable>)</member>
975 <member><replaceable>aggregate_name</replaceable> ( * )</member>
978 where <replaceable>aggregate_name</replaceable> is a previously
979 defined aggregate, and <replaceable>expression</replaceable> is
980 any value expression that does not itself contain an aggregate
985 The first form of aggregate expression invokes the aggregate
986 across all input rows for which the given expression yields a
987 non-NULL value. (Actually, it is up to the aggregate function
988 whether to ignore NULLs or not --- but all the standard ones do.)
989 The second form is the same as the first, since
990 <literal>ALL</literal> is the default. The third form invokes the
991 aggregate for all distinct non-NULL values of the expression found
992 in the input rows. The last form invokes the aggregate once for
993 each input row regardless of NULL or non-NULL values; since no
994 particular input value is specified, it is generally only useful
995 for the <function>count()</function> aggregate function.
999 For example, <literal>count(*)</literal> yields the total number
1000 of input rows; <literal>count(f1)</literal> yields the number of
1001 input rows in which <literal>f1</literal> is non-NULL;
1002 <literal>count(distinct f1)</literal> yields the number of
1003 distinct non-NULL values of <literal>f1</literal>.
1007 The predefined aggregate functions are described in <xref
1008 linkend="functions-aggregate">. Other aggregate functions may be added
1013 <sect2 id="sql-syntax-type-casts">
1014 <title>Type Casts</title>
1017 <primary>data types</primary>
1018 <secondary>type casts</secondary>
1022 A type cast specifies a conversion from one data type to another.
1023 <productname>PostgreSQL</productname> accepts two equivalent syntaxes
1026 CAST ( <replaceable>expression</replaceable> AS <replaceable>type</replaceable> )
1027 <replaceable>expression</replaceable>::<replaceable>type</replaceable>
1029 The <literal>CAST</> syntax conforms to SQL92; the syntax with
1030 <literal>::</literal> is historical <productname>PostgreSQL</productname>
1035 When a cast is applied to a value expression of a known type, it
1036 represents a run-time type conversion. The cast will succeed only
1037 if a suitable type conversion function is available. Notice that this
1038 is subtly different from the use of casts with constants, as shown in
1039 <xref linkend="sql-syntax-constants-generic">. A cast applied to an
1040 unadorned string literal represents the initial assignment of a type
1041 to a literal constant value, and so it will succeed for any type
1042 (if the contents of the string literal are acceptable input syntax for the
1047 An explicit type cast may be omitted if there is no ambiguity as to the
1048 type that a value expression must produce (for example, when it is
1049 assigned to a table column); the system will automatically apply a
1050 type cast in such cases.
1054 It is also possible to specify a type cast using a function-like
1057 <replaceable>typename</replaceable> ( <replaceable>expression</replaceable> )
1059 However, this only works for types whose names are also valid as
1060 function names. For example, <literal>double precision</literal>
1061 can't be used this way, but the equivalent <literal>float8</literal>
1062 can. Also, the names <literal>interval</>, <literal>time</>, and
1063 <literal>timestamp</> can only be used in this fashion if they are
1064 double-quoted, because of parser conflicts. Therefore, the use of
1065 the function-like cast syntax leads to inconsistencies and should
1066 probably be avoided in new applications.
1071 <title>Scalar Subqueries</title>
1074 A scalar subquery is an ordinary
1075 <command>SELECT</command> in parentheses that returns exactly one
1076 row with one column. The <command>SELECT</command> query is executed
1077 and the single returned value is used in the surrounding value expression.
1078 It is an error to use a query that
1079 returns more than one row or more than one column as a scalar subquery.
1080 (But if, during a particular execution, the subquery returns no rows,
1081 there is no error; the scalar result is taken to be NULL.)
1082 The subquery can refer to variables from the surrounding query,
1083 which will act as constants during any one evaluation of the subquery.
1084 See also <xref linkend="functions-subquery">.
1088 For example, the following finds the largest city population in each
1091 SELECT name, (SELECT max(pop) FROM cities WHERE cities.state = states.name)
1100 <sect1 id="sql-precedence">
1101 <title>Lexical Precedence</title>
1103 <indexterm zone="sql-precedence">
1104 <primary>operators</primary>
1105 <secondary>precedence</secondary>
1109 The precedence and associativity of the operators is hard-wired
1110 into the parser. Most operators have the same precedence and are
1111 left-associative. This may lead to non-intuitive behavior; for
1112 example the Boolean operators <literal><</> and <literal>></> have a different
1113 precedence than the Boolean operators <literal><=</> and <literal>>=</>. Also,
1114 you will sometimes need to add parentheses when using combinations
1115 of binary and unary operators. For instance
1123 because the parser has no idea -- until it is too late -- that
1124 <token>!</token> is defined as a postfix operator, not an infix one.
1125 To get the desired behavior in this case, you must write
1129 This is the price one pays for extensibility.
1132 <table tocentry="1">
1133 <title>Operator Precedence (decreasing)</title>
1138 <entry>Operator/Element</entry>
1139 <entry>Associativity</entry>
1140 <entry>Description</entry>
1146 <entry><token>::</token></entry>
1148 <entry><productname>PostgreSQL</productname>-style typecast</entry>
1152 <entry><token>[</token> <token>]</token></entry>
1154 <entry>array element selection</entry>
1158 <entry><token>.</token></entry>
1160 <entry>table/column name separator</entry>
1164 <entry><token>-</token></entry>
1165 <entry>right</entry>
1166 <entry>unary minus</entry>
1170 <entry><token>^</token></entry>
1172 <entry>exponentiation</entry>
1176 <entry><token>*</token> <token>/</token> <token>%</token></entry>
1178 <entry>multiplication, division, modulo</entry>
1182 <entry><token>+</token> <token>-</token></entry>
1184 <entry>addition, subtraction</entry>
1188 <entry><token>IS</token></entry>
1190 <entry>test for TRUE, FALSE, UNKNOWN, NULL</entry>
1194 <entry><token>ISNULL</token></entry>
1196 <entry>test for NULL</entry>
1200 <entry><token>NOTNULL</token></entry>
1202 <entry>test for NOT NULL</entry>
1206 <entry>(any other)</entry>
1208 <entry>all other native and user-defined operators</entry>
1212 <entry><token>IN</token></entry>
1214 <entry>set membership</entry>
1218 <entry><token>BETWEEN</token></entry>
1220 <entry>containment</entry>
1224 <entry><token>OVERLAPS</token></entry>
1226 <entry>time interval overlap</entry>
1230 <entry><token>LIKE</token> <token>ILIKE</token></entry>
1232 <entry>string pattern matching</entry>
1236 <entry><token><</token> <token>></token></entry>
1238 <entry>less than, greater than</entry>
1242 <entry><token>=</token></entry>
1243 <entry>right</entry>
1244 <entry>equality, assignment</entry>
1248 <entry><token>NOT</token></entry>
1249 <entry>right</entry>
1250 <entry>logical negation</entry>
1254 <entry><token>AND</token></entry>
1256 <entry>logical conjunction</entry>
1260 <entry><token>OR</token></entry>
1262 <entry>logical disjunction</entry>
1269 Note that the operator precedence rules also apply to user-defined
1270 operators that have the same names as the built-in operators
1271 mentioned above. For example, if you define a
1272 <quote>+</quote> operator for some custom data type it will have
1273 the same precedence as the built-in <quote>+</quote> operator, no
1274 matter what yours does.
1280 <!-- Keep this comment at the end of the file
1285 sgml-minimize-attributes:nil
1286 sgml-always-quote-attributes:t
1289 sgml-parent-document:nil
1290 sgml-default-dtd-file:"./reference.ced"
1291 sgml-exposed-tags:nil
1292 sgml-local-catalogs:("/usr/lib/sgml/catalog")
1293 sgml-local-ecat-files:nil