path: root/doc/src/sgml/syntax.sgml

<!-- doc/src/sgml/syntax.sgml -->

<chapter id="sql-syntax">
 <title>SQL Syntax</title>

 <indexterm zone="sql-syntax">
  <primary>syntax</primary>
  <secondary>SQL</secondary>
 </indexterm>

 <para>
  This chapter describes the syntax of SQL.  It forms the foundation
  for understanding the following chapters which will go into detail
  about how SQL commands are applied to define and modify data.
 </para>

 <para>
  We also advise users who are already familiar with SQL to read this
  chapter carefully because it contains several rules and concepts that
  are implemented inconsistently among SQL databases or that are
  specific to <productname>PostgreSQL</productname>.
 </para>

 <sect1 id="sql-syntax-lexical">
  <title>Lexical Structure</title>

  <indexterm>
   <primary>token</primary>
  </indexterm>

  <para>
   SQL input consists of a sequence of
   <firstterm>commands</firstterm>.  A command is composed of a
   sequence of <firstterm>tokens</firstterm>, terminated by a
   semicolon (<quote>;</quote>).  The end of the input stream also
   terminates a command.  Which tokens are valid depends on the syntax
   of the particular command.
  </para>

  <para>
   A token can be a <firstterm>key word</firstterm>, an
   <firstterm>identifier</firstterm>, a <firstterm>quoted
   identifier</firstterm>, a <firstterm>literal</firstterm> (or
   constant), or a special character symbol.  Tokens are normally
   separated by whitespace (space, tab, newline), but need not be if
   there is no ambiguity (which is generally only the case if a
   special character is adjacent to some other token type).
  </para>

   <para>
    For example, the following is (syntactically) valid SQL input:
<programlisting>
SELECT * FROM MY_TABLE;
UPDATE MY_TABLE SET A = 5;
INSERT INTO MY_TABLE VALUES (3, 'hi there');
</programlisting>
    This is a sequence of three commands, one per line (although this
    is not required; more than one command can be on a line, and
    commands can usefully be split across lines).
   </para>

  <para>
   Additionally, <firstterm>comments</firstterm> can occur in SQL
   input.  They are not tokens, they are effectively equivalent to
   whitespace.
  </para>

  <para>
   The SQL syntax is not very consistent regarding what tokens
   identify commands and which are operands or parameters.  The first
   few tokens are generally the command name, so in the above example
   we would usually speak of a <quote>SELECT</quote>, an
   <quote>UPDATE</quote>, and an <quote>INSERT</quote> command.  But
   for instance the <command>UPDATE</command> command always requires
   a <token>SET</token> token to appear in a certain position, and
   this particular variation of <command>INSERT</command> also
   requires a <token>VALUES</token> in order to be complete.  The
   precise syntax rules for each command are described in <xref linkend="reference"/>.
  </para>

  <sect2 id="sql-syntax-identifiers">
   <title>Identifiers and Key Words</title>

   <indexterm zone="sql-syntax-identifiers">
    <primary>identifier</primary>
    <secondary>syntax of</secondary>
   </indexterm>

   <indexterm zone="sql-syntax-identifiers">
    <primary>name</primary>
    <secondary>syntax of</secondary>
   </indexterm>

   <indexterm zone="sql-syntax-identifiers">
    <primary>key word</primary>
    <secondary>syntax of</secondary>
   </indexterm>

   <para>
    Tokens such as <token>SELECT</token>, <token>UPDATE</token>, or
    <token>VALUES</token> in the example above are examples of
    <firstterm>key words</firstterm>, that is, words that have a fixed
    meaning in the SQL language.  The tokens <token>MY_TABLE</token>
    and <token>A</token> are examples of
    <firstterm>identifiers</firstterm>.  They identify names of
    tables, columns, or other database objects, depending on the
    command they are used in.  Therefore they are sometimes simply
    called <quote>names</quote>.  Key words and identifiers have the
    same lexical structure, meaning that one cannot know whether a
    token is an identifier or a key word without knowing the language.
    A complete list of key words can be found in <xref
    linkend="sql-keywords-appendix"/>.
   </para>

   <para>
    SQL identifiers and key words must begin with a letter
    (<literal>a</literal>-<literal>z</literal>, but also letters with
    diacritical marks and non-Latin letters) or an underscore
    (<literal>_</literal>).  Subsequent characters in an identifier or
    key word can be letters, underscores, digits
    (<literal>0</literal>-<literal>9</literal>), or dollar signs
    (<literal>$</literal>).  Note that dollar signs are not allowed in identifiers
    according to the letter of the SQL standard, so their use might render
    applications less portable.
    The SQL standard will not define a key word that contains
    digits or starts or ends with an underscore, so identifiers of this
    form are safe against possible conflict with future extensions of the
    standard.
   </para>

   <para>
    <indexterm><primary>identifier</primary><secondary>length</secondary></indexterm>
    The system uses no more than <symbol>NAMEDATALEN</symbol>-1
    bytes of an identifier; longer names can be written in
    commands, but they will be truncated.  By default,
    <symbol>NAMEDATALEN</symbol> is 64 so the maximum identifier
    length is 63 bytes. If this limit is problematic, it can be raised by
    changing the <symbol>NAMEDATALEN</symbol> constant in
    <filename>src/include/pg_config_manual.h</filename>.
   </para>

   <para>
    <indexterm>
     <primary>case sensitivity</primary>
     <secondary>of SQL commands</secondary>
    </indexterm>
    Key words and unquoted identifiers are case insensitive.  Therefore:
<programlisting>
UPDATE MY_TABLE SET A = 5;
</programlisting>
    can equivalently be written as:
<programlisting>
uPDaTE my_TabLE SeT a = 5;
</programlisting>
    A convention often used is to write key words in upper
    case and names in lower case, e.g.:
<programlisting>
UPDATE my_table SET a = 5;
</programlisting>
   </para>

   <para>
    <indexterm>
     <primary>quotation marks</primary>
     <secondary>and identifiers</secondary>
    </indexterm>
    There is a second kind of identifier:  the <firstterm>delimited
    identifier</firstterm> or <firstterm>quoted
    identifier</firstterm>.  It is formed by enclosing an arbitrary
    sequence of characters in double-quotes
    (<literal>"</literal>). <!-- " font-lock mania --> A delimited
    identifier is always an identifier, never a key word.  So
    <literal>"select"</literal> could be used to refer to a column or
    table named <quote>select</quote>, whereas an unquoted
    <literal>select</literal> would be taken as a key word and
    would therefore provoke a parse error when used where a table or
    column name is expected.  The example can be written with quoted
    identifiers like this:
<programlisting>
UPDATE "my_table" SET "a" = 5;
</programlisting>
   </para>

   <para>
    Quoted identifiers can contain any character, except the character
    with code zero.  (To include a double quote, write two double quotes.)
    This allows constructing table or column names that would
    otherwise not be possible, such as ones containing spaces or
    ampersands.  The length limitation still applies.
   </para>

   <para>
    Quoting an identifier also makes it case-sensitive, whereas
    unquoted names are always folded to lower case.  For example, the
    identifiers <literal>FOO</literal>, <literal>foo</literal>, and
    <literal>"foo"</literal> are considered the same by
    <productname>PostgreSQL</productname>, but
    <literal>"Foo"</literal> and <literal>"FOO"</literal> are
    different from these three and each other.  (The folding of
    unquoted names to lower case in <productname>PostgreSQL</productname> is
    incompatible with the SQL standard, which says that unquoted names
    should be folded to upper case.  Thus, <literal>foo</literal>
    should be equivalent to <literal>"FOO"</literal> not
    <literal>"foo"</literal> according to the standard.  If you want
    to write portable applications you are advised to always quote a
    particular name or never quote it.)
   </para>

   <indexterm>
     <primary>Unicode escape</primary>
     <secondary>in identifiers</secondary>
   </indexterm>

   <para>
    A variant of quoted
    identifiers allows including escaped Unicode characters identified
    by their code points.  This variant starts
    with <literal>U&amp;</literal> (upper or lower case U followed by
    ampersand) immediately before the opening double quote, without
    any spaces in between, for example <literal>U&amp;"foo"</literal>.
    (Note that this creates an ambiguity with the
    operator <literal>&amp;</literal>.  Use spaces around the operator to
    avoid this problem.)  Inside the quotes, Unicode characters can be
    specified in escaped form by writing a backslash followed by the
    four-digit hexadecimal code point number or alternatively a
    backslash followed by a plus sign followed by a six-digit
    hexadecimal code point number.  For example, the
    identifier <literal>"data"</literal> could be written as
<programlisting>
U&amp;"d\0061t\+000061"
</programlisting>
    The following less trivial example writes the Russian
    word <quote>slon</quote> (elephant) in Cyrillic letters:
<programlisting>
U&amp;"\0441\043B\043E\043D"
</programlisting>
   </para>

   <para>
    If a different escape character than backslash is desired, it can
    be specified using
    the <literal>UESCAPE</literal><indexterm><primary>UESCAPE</primary></indexterm>
    clause after the string, for example:
<programlisting>
U&amp;"d!0061t!+000061" UESCAPE '!'
</programlisting>
    The escape character can be any single character other than a
    hexadecimal digit, the plus sign, a single quote, a double quote,
    or a whitespace character.  Note that the escape character is
    written in single quotes, not double quotes,
    after <literal>UESCAPE</literal>.
   </para>

   <para>
    To include the escape character in the identifier literally, write
    it twice.
   </para>

   <para>
    Either the 4-digit or the 6-digit escape form can be used to
    specify UTF-16 surrogate pairs to compose characters with code
    points larger than U+FFFF, although the availability of the
    6-digit form technically makes this unnecessary.  (Surrogate
    pairs are not stored directly, but are combined into a single
    code point.)
   </para>

   <para>
    If the server encoding is not UTF-8, the Unicode code point identified
    by one of these escape sequences is converted to the actual server
    encoding; an error is reported if that's not possible.
   </para>
  </sect2>


  <sect2 id="sql-syntax-constants">
   <title>Constants</title>

   <indexterm zone="sql-syntax-constants">
    <primary>constant</primary>
   </indexterm>

   <para>
    There are three kinds of <firstterm>implicitly-typed
    constants</firstterm> in <productname>PostgreSQL</productname>:
    strings, bit strings, and numbers.
    Constants can also be specified with explicit types, which can
    enable more accurate representation and more efficient handling by
    the system. These alternatives are discussed in the following
    subsections.
   </para>

   <sect3 id="sql-syntax-strings">
    <title>String Constants</title>

    <indexterm zone="sql-syntax-strings">
     <primary>character string</primary>
     <secondary>constant</secondary>
    </indexterm>

    <para>
     <indexterm>
      <primary>quotation marks</primary>
      <secondary>escaping</secondary>
     </indexterm>
     A string constant in SQL is an arbitrary sequence of characters
     bounded by single quotes (<literal>'</literal>), for example
     <literal>'This is a string'</literal>.  To include
     a single-quote character within a string constant,
     write two adjacent single quotes, e.g.,
     <literal>'Dianne''s horse'</literal>.
     Note that this is <emphasis>not</emphasis> the same as a double-quote
     character (<literal>"</literal>). <!-- font-lock sanity: " -->
    </para>

    <para>
     Two string constants that are only separated by whitespace
     <emphasis>with at least one newline</emphasis> are concatenated
     and effectively treated as if the string had been written as one
     constant.  For example:
<programlisting>
SELECT 'foo'
'bar';
</programlisting>
     is equivalent to:
<programlisting>
SELECT 'foobar';
</programlisting>
     but:
<programlisting>
SELECT 'foo'      'bar';
</programlisting>
     is not valid syntax.  (This slightly bizarre behavior is specified
     by <acronym>SQL</acronym>; <productname>PostgreSQL</productname> is
     following the standard.)
    </para>
   </sect3>

   <sect3 id="sql-syntax-strings-escape">
    <title>String Constants with C-Style Escapes</title>

     <indexterm zone="sql-syntax-strings-escape">
      <primary>escape string syntax</primary>
     </indexterm>
     <indexterm zone="sql-syntax-strings-escape">
      <primary>backslash escapes</primary>
     </indexterm>

    <para>
     <productname>PostgreSQL</productname> also accepts <quote>escape</quote>
     string constants, which are an extension to the SQL standard.
     An escape string constant is specified by writing the letter
     <literal>E</literal> (upper or lower case) just before the opening single
     quote, e.g., <literal>E'foo'</literal>.  (When continuing an escape string
     constant across lines, write <literal>E</literal> only before the first opening
     quote.)
     Within an escape string, a backslash character (<literal>\</literal>) begins a
     C-like <firstterm>backslash escape</firstterm> sequence, in which the combination
     of backslash and following character(s) represent a special byte
     value, as shown in <xref linkend="sql-backslash-table"/>.
    </para>

     <table id="sql-backslash-table">
      <title>Backslash Escape Sequences</title>
      <tgroup cols="2">
      <thead>
       <row>
        <entry>Backslash Escape Sequence</entry>
        <entry>Interpretation</entry>
       </row>
      </thead>

      <tbody>
       <row>
        <entry><literal>\b</literal></entry>
        <entry>backspace</entry>
       </row>
       <row>
        <entry><literal>\f</literal></entry>
        <entry>form feed</entry>
       </row>
       <row>
        <entry><literal>\n</literal></entry>
        <entry>newline</entry>
       </row>
       <row>
        <entry><literal>\r</literal></entry>
        <entry>carriage return</entry>
       </row>
       <row>
        <entry><literal>\t</literal></entry>
        <entry>tab</entry>
       </row>
       <row>
        <entry>
         <literal>\<replaceable>o</replaceable></literal>,
         <literal>\<replaceable>oo</replaceable></literal>,
         <literal>\<replaceable>ooo</replaceable></literal>
         (<replaceable>o</replaceable> = 0&ndash;7)
        </entry>
        <entry>octal byte value</entry>
       </row>
       <row>
        <entry>
         <literal>\x<replaceable>h</replaceable></literal>,
         <literal>\x<replaceable>hh</replaceable></literal>
         (<replaceable>h</replaceable> = 0&ndash;9, A&ndash;F)
        </entry>
        <entry>hexadecimal byte value</entry>
       </row>
       <row>
        <entry>
         <literal>\u<replaceable>xxxx</replaceable></literal>,
         <literal>\U<replaceable>xxxxxxxx</replaceable></literal>
         (<replaceable>x</replaceable> = 0&ndash;9, A&ndash;F)
        </entry>
        <entry>16 or 32-bit hexadecimal Unicode character value</entry>
       </row>
      </tbody>
      </tgroup>
     </table>

    <para>
     Any other
     character following a backslash is taken literally. Thus, to
     include a backslash character, write two backslashes (<literal>\\</literal>).
     Also, a single quote can be included in an escape string by writing
     <literal>\'</literal>, in addition to the normal way of <literal>''</literal>.
    </para>

    <para>
     It is your responsibility that the byte sequences you create,
     especially when using the octal or hexadecimal escapes, compose
     valid characters in the server character set encoding.
     A useful alternative is to use Unicode escapes or the
     alternative Unicode escape syntax, explained
     in <xref linkend="sql-syntax-strings-uescape"/>; then the server
     will check that the character conversion is possible.
    </para>

    <caution>
    <para>
     If the configuration parameter
     <xref linkend="guc-standard-conforming-strings"/> is <literal>off</literal>,
     then <productname>PostgreSQL</productname> recognizes backslash escapes
     in both regular and escape string constants.  However, as of
     <productname>PostgreSQL</productname> 9.1, the default is <literal>on</literal>, meaning
     that backslash escapes are recognized only in escape string constants.
     This behavior is more standards-compliant, but might break applications
     which rely on the historical behavior, where backslash escapes
     were always recognized.  As a workaround, you can set this parameter
     to <literal>off</literal>, but it is better to migrate away from using backslash
     escapes.  If you need to use a backslash escape to represent a special
     character, write the string constant with an <literal>E</literal>.
    </para>

    <para>
     In addition to <varname>standard_conforming_strings</varname>, the configuration
     parameters <xref linkend="guc-escape-string-warning"/> and
     <xref linkend="guc-backslash-quote"/> govern treatment of backslashes
     in string constants.
    </para>
    </caution>

    <para>
     The character with the code zero cannot be in a string constant.
    </para>
   </sect3>

   <sect3 id="sql-syntax-strings-uescape">
    <title>String Constants with Unicode Escapes</title>

    <indexterm  zone="sql-syntax-strings-uescape">
     <primary>Unicode escape</primary>
     <secondary>in string constants</secondary>
    </indexterm>

    <para>
     <productname>PostgreSQL</productname> also supports another type
     of escape syntax for strings that allows specifying arbitrary
     Unicode characters by code point.  A Unicode escape string
     constant starts with <literal>U&amp;</literal> (upper or lower case
     letter U followed by ampersand) immediately before the opening
     quote, without any spaces in between, for
     example <literal>U&amp;'foo'</literal>.  (Note that this creates an
     ambiguity with the operator <literal>&amp;</literal>.  Use spaces
     around the operator to avoid this problem.)  Inside the quotes,
     Unicode characters can be specified in escaped form by writing a
     backslash followed by the four-digit hexadecimal code point
     number or alternatively a backslash followed by a plus sign
     followed by a six-digit hexadecimal code point number.  For
     example, the string <literal>'data'</literal> could be written as
<programlisting>
U&amp;'d\0061t\+000061'
</programlisting>
     The following less trivial example writes the Russian
     word <quote>slon</quote> (elephant) in Cyrillic letters:
<programlisting>
U&amp;'\0441\043B\043E\043D'
</programlisting>
    </para>

    <para>
     If a different escape character than backslash is desired, it can
     be specified using
     the <literal>UESCAPE</literal><indexterm><primary>UESCAPE</primary></indexterm>
     clause after the string, for example:
<programlisting>
U&amp;'d!0061t!+000061' UESCAPE '!'
</programlisting>
     The escape character can be any single character other than a
     hexadecimal digit, the plus sign, a single quote, a double quote,
     or a whitespace character.
    </para>

    <para>
     To include the escape character in the string literally, write
     it twice.
    </para>

    <para>
     Either the 4-digit or the 6-digit escape form can be used to
     specify UTF-16 surrogate pairs to compose characters with code
     points larger than U+FFFF, although the availability of the
     6-digit form technically makes this unnecessary.  (Surrogate
     pairs are not stored directly, but are combined into a single
     code point.)
    </para>

    <para>
     If the server encoding is not UTF-8, the Unicode code point identified
     by one of these escape sequences is converted to the actual server
     encoding; an error is reported if that's not possible.
    </para>

    <para>
     Also, the Unicode escape syntax for string constants only works
     when the configuration
     parameter <xref linkend="guc-standard-conforming-strings"/> is
     turned on.  This is because otherwise this syntax could confuse
     clients that parse the SQL statements to the point that it could
     lead to SQL injections and similar security issues.  If the
     parameter is set to off, this syntax will be rejected with an
     error message.
    </para>
   </sect3>

   <sect3 id="sql-syntax-dollar-quoting">
    <title>Dollar-Quoted String Constants</title>

     <indexterm>
      <primary>dollar quoting</primary>
     </indexterm>

    <para>
     While the standard syntax for specifying string constants is usually
     convenient, it can be difficult to understand when the desired string
     contains many single quotes, since each of those must
     be doubled. To allow more readable queries in such situations,
     <productname>PostgreSQL</productname> provides another way, called
     <quote>dollar quoting</quote>, to write string constants.
     A dollar-quoted string constant
     consists of a dollar sign (<literal>$</literal>), an optional
     <quote>tag</quote> of zero or more characters, another dollar
     sign, an arbitrary sequence of characters that makes up the
     string content, a dollar sign, the same tag that began this
     dollar quote, and a dollar sign. For example, here are two
     different ways to specify the string <quote>Dianne's horse</quote>
     using dollar quoting:
<programlisting>
$$Dianne's horse$$
$SomeTag$Dianne's horse$SomeTag$
</programlisting>
     Notice that inside the dollar-quoted string, single quotes can be
     used without needing to be escaped.  Indeed, no characters inside
     a dollar-quoted string are ever escaped: the string content is always
     written literally.  Backslashes are not special, and neither are
     dollar signs, unless they are part of a sequence matching the opening
     tag.
    </para>

    <para>
     It is possible to nest dollar-quoted string constants by choosing
     different tags at each nesting level.  This is most commonly used in
     writing function definitions.  For example:
<programlisting>
$function$
BEGIN
    RETURN ($1 ~ $q$[\t\r\n\v\\]$q$);
END;
$function$
</programlisting>
     Here, the sequence <literal>$q$[\t\r\n\v\\]$q$</literal> represents a
     dollar-quoted literal string <literal>[\t\r\n\v\\]</literal>, which will
     be recognized when the function body is executed by
     <productname>PostgreSQL</productname>.  But since the sequence does not match
     the outer dollar quoting delimiter <literal>$function$</literal>, it is
     just some more characters within the constant so far as the outer
     string is concerned.
    </para>

    <para>
     The tag, if any, of a dollar-quoted string follows the same rules
     as an unquoted identifier, except that it cannot contain a dollar sign.
     Tags are case sensitive, so <literal>$tag$String content$tag$</literal>
     is correct, but <literal>$TAG$String content$tag$</literal> is not.
    </para>

    <para>
     A dollar-quoted string that follows a keyword or identifier must
     be separated from it by whitespace; otherwise the dollar quoting
     delimiter would be taken as part of the preceding identifier.
    </para>

    <para>
     Dollar quoting is not part of the SQL standard, but it is often a more
     convenient way to write complicated string literals than the
     standard-compliant single quote syntax.  It is particularly useful when
     representing string constants inside other constants, as is often needed
     in procedural function definitions.  With single-quote syntax, each
     backslash in the above example would have to be written as four
     backslashes, which would be reduced to two backslashes in parsing the
     original string constant, and then to one when the inner string constant
     is re-parsed during function execution.
    </para>
   </sect3>

   <sect3 id="sql-syntax-bit-strings">
    <title>Bit-String Constants</title>

    <indexterm zone="sql-syntax-bit-strings">
     <primary>bit string</primary>
     <secondary>constant</secondary>
    </indexterm>

    <para>
     Bit-string constants look like regular string constants with a
     <literal>B</literal> (upper or lower case) immediately before the
     opening quote (no intervening whitespace), e.g.,
     <literal>B'1001'</literal>.  The only characters allowed within
     bit-string constants are <literal>0</literal> and
     <literal>1</literal>.
    </para>

    <para>
     Alternatively, bit-string constants can be specified in hexadecimal
     notation, using a leading <literal>X</literal> (upper or lower case),
     e.g., <literal>X'1FF'</literal>.  This notation is equivalent to
     a bit-string constant with four binary digits for each hexadecimal digit.
    </para>

    <para>
     Both forms of bit-string constant can be continued
     across lines in the same way as regular string constants.
     Dollar quoting cannot be used in a bit-string constant.
    </para>
   </sect3>

   <sect3 id="sql-syntax-constants-numeric">
    <title>Numeric Constants</title>

    <indexterm>
     <primary>number</primary>
     <secondary>constant</secondary>
    </indexterm>

    <para>
     Numeric constants are accepted in these general forms:
<synopsis>
<replaceable>digits</replaceable>
<replaceable>digits</replaceable>.<optional><replaceable>digits</replaceable></optional><optional>e<optional>+-</optional><replaceable>digits</replaceable></optional>
<optional><replaceable>digits</replaceable></optional>.<replaceable>digits</replaceable><optional>e<optional>+-</optional><replaceable>digits</replaceable></optional>
<replaceable>digits</replaceable>e<optional>+-</optional><replaceable>digits</replaceable>
</synopsis>
     where <replaceable>digits</replaceable> is one or more decimal
     digits (0 through 9).  At least one digit must be before or after the
     decimal point, if one is used.  At least one digit must follow the
     exponent marker (<literal>e</literal>), if one is present.
     There cannot be any spaces or other characters embedded in the
     constant.  Note that any leading plus or minus sign is not actually
     considered part of the constant; it is an operator applied to the
     constant.
    </para>

    <para>
     These are some examples of valid numeric constants:
<literallayout>
42
3.5
4.
.001
5e2
1.925e-3
</literallayout>
    </para>

    <para>
     <indexterm><primary>integer</primary></indexterm>
     <indexterm><primary>bigint</primary></indexterm>
     <indexterm><primary>numeric</primary></indexterm>
     A numeric constant that contains neither a decimal point nor an
     exponent is initially presumed to be type <type>integer</type> if its
     value fits in type <type>integer</type> (32 bits); otherwise it is
     presumed to be type <type>bigint</type> if its
     value fits in type <type>bigint</type> (64 bits); otherwise it is
     taken to be type <type>numeric</type>.  Constants that contain decimal
     points and/or exponents are always initially presumed to be type
     <type>numeric</type>.
    </para>

    <para>
     The initially assigned data type of a numeric constant is just a
     starting point for the type resolution algorithms.  In most cases
     the constant will be automatically coerced to the most
     appropriate type depending on context.  When necessary, you can
     force a numeric value to be interpreted as a specific data type
     by casting it.<indexterm><primary>type cast</primary></indexterm>
     For example, you can force a numeric value to be treated as type
     <type>real</type> (<type>float4</type>) by writing:

<programlisting>
REAL '1.23'  -- string style
1.23::REAL   -- PostgreSQL (historical) style
</programlisting>

     These are actually just special cases of the general casting
     notations discussed next.
    </para>
   </sect3>

   <sect3 id="sql-syntax-constants-generic">
    <title>Constants of Other Types</title>

    <indexterm>
     <primary>data type</primary>
     <secondary>constant</secondary>
    </indexterm>

    <para>
     A constant of an <emphasis>arbitrary</emphasis> type can be
     entered using any one of the following notations:
<synopsis>
<replaceable>type</replaceable> '<replaceable>string</replaceable>'
'<replaceable>string</replaceable>'::<replaceable>type</replaceable>
CAST ( '<replaceable>string</replaceable>' AS <replaceable>type</replaceable> )
</synopsis>
     The string constant's text is passed to the input conversion
     routine for the type called <replaceable>type</replaceable>. The
     result is a constant of the indicated type.  The explicit type
     cast can be omitted if there is no ambiguity as to the type the
     constant must be (for example, when it is assigned directly to a
     table column), in which case it is automatically coerced.
    </para>

    <para>
     The string constant can be written using either regular SQL
     notation or dollar-quoting.
    </para>

    <para>
     It is also possible to specify a type coercion using a function-like
     syntax:
<synopsis>
<replaceable>typename</replaceable> ( '<replaceable>string</replaceable>' )
</synopsis>
     but not all type names can be used in this way; see <xref
     linkend="sql-syntax-type-casts"/> for details.
    </para>

    <para>
     The <literal>::</literal>, <literal>CAST()</literal>, and
     function-call syntaxes can also be used to specify run-time type
     conversions of arbitrary expressions, as discussed in <xref
     linkend="sql-syntax-type-casts"/>.  To avoid syntactic ambiguity, the
     <literal><replaceable>type</replaceable> '<replaceable>string</replaceable>'</literal>
     syntax can only be used to specify the type of a simple literal constant.
     Another restriction on the
     <literal><replaceable>type</replaceable> '<replaceable>string</replaceable>'</literal>
     syntax is that it does not work for array types; use <literal>::</literal>
     or <literal>CAST()</literal> to specify the type of an array constant.
    </para>

    <para>
     The <literal>CAST()</literal> syntax conforms to SQL.  The
     <literal><replaceable>type</replaceable> '<replaceable>string</replaceable>'</literal>
     syntax is a generalization of the standard: SQL specifies this syntax only
     for a few data types, but <productname>PostgreSQL</productname> allows it
     for all types.  The syntax with
     <literal>::</literal> is historical <productname>PostgreSQL</productname>
     usage, as is the function-call syntax.
    </para>
   </sect3>
  </sect2>

  <sect2 id="sql-syntax-operators">
   <title>Operators</title>

   <indexterm zone="sql-syntax-operators">
    <primary>operator</primary>
    <secondary>syntax</secondary>
   </indexterm>

   <para>
    An operator name is a sequence of up to <symbol>NAMEDATALEN</symbol>-1
    (63 by default) characters from the following list:
<literallayout>
+ - * / &lt; &gt; = ~ ! @ # % ^ &amp; | ` ?
</literallayout>

    There are a few restrictions on operator names, however:
    <itemizedlist>
     <listitem>
      <para>
       <literal>--</literal> and <literal>/*</literal> cannot appear
       anywhere in an operator name, since they will be taken as the
       start of a comment.
      </para>
     </listitem>

     <listitem>
      <para>
       A multiple-character operator name cannot end in <literal>+</literal> or <literal>-</literal>,
       unless the name also contains at least one of these characters:
<literallayout>
~ ! @ # % ^ &amp; | ` ?
</literallayout>
       For example, <literal>@-</literal> is an allowed operator name,
       but <literal>*-</literal> is not.  This restriction allows
       <productname>PostgreSQL</productname> to parse SQL-compliant
       queries without requiring spaces between tokens.
      </para>
     </listitem>
    </itemizedlist>
   </para>

   <para>
    When working with non-SQL-standard operator names, you will usually
    need to separate adjacent operators with spaces to avoid ambiguity.
    For example, if you have defined a prefix operator named <literal>@</literal>,
    you cannot write <literal>X*@Y</literal>; you must write
    <literal>X* @Y</literal> to ensure that
    <productname>PostgreSQL</productname> reads it as two operator names
    not one.
   </para>
  </sect2>

  <sect2 id="sql-syntax-special-chars">
   <title>Special Characters</title>

  <para>
   Some characters that are not alphanumeric have a special meaning
   that is different from being an operator.  Details on the usage can
   be found at the location where the respective syntax element is
   described.  This section only exists to advise the existence and
   summarize the purposes of these characters.

   <itemizedlist>
    <listitem>
     <para>
      A dollar sign (<literal>$</literal>) followed by digits is used
      to represent a positional parameter in the body of a function
      definition or a prepared statement.  In other contexts the
      dollar sign can be part of an identifier or a dollar-quoted string
      constant.
     </para>
    </listitem>

    <listitem>
     <para>
      Parentheses (<literal>()</literal>) have their usual meaning to
      group expressions and enforce precedence.  In some cases
      parentheses are required as part of the fixed syntax of a
      particular SQL command.
     </para>
    </listitem>

    <listitem>
     <para>
      Brackets (<literal>[]</literal>) are used to select the elements
      of an array.  See <xref linkend="arrays"/> for more information
      on arrays.
     </para>
    </listitem>

    <listitem>
     <para>
      Commas (<literal>,</literal>) are used in some syntactical
      constructs to separate the elements of a list.
     </para>
    </listitem>

    <listitem>
     <para>
      The semicolon (<literal>;</literal>) terminates an SQL command.
      It cannot appear anywhere within a command, except within a
      string constant or quoted identifier.
     </para>
    </listitem>

    <listitem>
     <para>
      The colon (<literal>:</literal>) is used to select
      <quote>slices</quote> from arrays. (See <xref
      linkend="arrays"/>.)  In certain SQL dialects (such as Embedded
      SQL), the colon is used to prefix variable names.
     </para>
    </listitem>

    <listitem>
     <para>
      The asterisk (<literal>*</literal>) is used in some contexts to denote
      all the fields of a table row or composite value.  It also
      has a special meaning when used as the argument of an
      aggregate function, namely that the aggregate does not require
      any explicit parameter.
     </para>
    </listitem>

    <listitem>
     <para>
      The period (<literal>.</literal>) is used in numeric
      constants, and to separate schema, table, and column names.
     </para>
    </listitem>
   </itemizedlist>

   </para>
  </sect2>

  <sect2 id="sql-syntax-comments">
   <title>Comments</title>

   <indexterm zone="sql-syntax-comments">
    <primary>comment</primary>
    <secondary sortas="SQL">in SQL</secondary>
   </indexterm>

   <para>
    A comment is a sequence of characters beginning with
    double dashes and extending to the end of the line, e.g.:
<programlisting>
-- This is a standard SQL comment
</programlisting>
   </para>

   <para>
    Alternatively, C-style block comments can be used:
<programlisting>
/* multiline comment
 * with nesting: /* nested block comment */
 */
</programlisting>
    where the comment begins with <literal>/*</literal> and extends to
    the matching occurrence of <literal>*/</literal>. These block
    comments nest, as specified in the SQL standard but unlike C, so that one can
    comment out larger blocks of code that might contain existing block
    comments.
   </para>

   <para>
    A comment is removed from the input stream before further syntax
    analysis and is effectively replaced by whitespace.
   </para>
  </sect2>

  <sect2 id="sql-precedence">
   <title>Operator Precedence</title>

   <indexterm zone="sql-precedence">
    <primary>operator</primary>
    <secondary>precedence</secondary>
   </indexterm>

   <para>
    <xref linkend="sql-precedence-table"/> shows the precedence and
    associativity of the operators in <productname>PostgreSQL</productname>.
    Most operators have the same precedence and are left-associative.
    The precedence and associativity of the operators is hard-wired
    into the parser.
    Add parentheses if you want an expression with multiple operators
    to be parsed in some other way than what the precedence rules imply.
   </para>

   <table id="sql-precedence-table">
    <title>Operator Precedence (highest to lowest)</title>

    <tgroup cols="3">
     <colspec colname="col1" colwidth="2*"/>
     <colspec colname="col2" colwidth="1*"/>
     <colspec colname="col3" colwidth="2*"/>
     <thead>
      <row>
       <entry>Operator/Element</entry>
       <entry>Associativity</entry>
       <entry>Description</entry>
      </row>
     </thead>

     <tbody>
      <row>
       <entry><token>.</token></entry>
       <entry>left</entry>
       <entry>table/column name separator</entry>
      </row>

      <row>
       <entry><token>::</token></entry>
       <entry>left</entry>
       <entry><productname>PostgreSQL</productname>-style typecast</entry>
      </row>

      <row>
       <entry><token>[</token> <token>]</token></entry>
       <entry>left</entry>
       <entry>array element selection</entry>
      </row>

      <row>
       <entry><token>+</token> <token>-</token></entry>
       <entry>right</entry>
       <entry>unary plus, unary minus</entry>
      </row>

      <row>
       <entry><token>COLLATE</token></entry>
       <entry>left</entry>
       <entry>collation selection</entry>
      </row>

      <row>
       <entry><token>AT</token></entry>
       <entry>left</entry>
       <entry><literal>AT TIME ZONE</literal></entry>
      </row>

      <row>
       <entry><token>^</token></entry>
       <entry>left</entry>
       <entry>exponentiation</entry>
      </row>

      <row>
       <entry><token>*</token> <token>/</token> <token>%</token></entry>
       <entry>left</entry>
       <entry>multiplication, division, modulo</entry>
      </row>

      <row>
       <entry><token>+</token> <token>-</token></entry>
       <entry>left</entry>
       <entry>addition, subtraction</entry>
      </row>

      <row>
       <entry>(any other operator)</entry>
       <entry>left</entry>
       <entry>all other native and user-defined operators</entry>
      </row>

      <row>
       <entry><token>BETWEEN</token> <token>IN</token> <token>LIKE</token> <token>ILIKE</token> <token>SIMILAR</token></entry>
       <entry></entry>
       <entry>range containment, set membership, string matching</entry>
      </row>

      <row>
       <entry><token>&lt;</token> <token>&gt;</token> <token>=</token> <token>&lt;=</token> <token>&gt;=</token> <token>&lt;&gt;</token>
</entry>
       <entry></entry>
       <entry>comparison operators</entry>
      </row>

      <row>
       <entry><token>IS</token> <token>ISNULL</token> <token>NOTNULL</token></entry>
       <entry></entry>
       <entry><literal>IS TRUE</literal>, <literal>IS FALSE</literal>, <literal>IS
       NULL</literal>, <literal>IS DISTINCT FROM</literal>, etc.</entry>
      </row>

      <row>
       <entry><token>NOT</token></entry>
       <entry>right</entry>
       <entry>logical negation</entry>
      </row>

      <row>
       <entry><token>AND</token></entry>
       <entry>left</entry>
       <entry>logical conjunction</entry>
      </row>

      <row>
       <entry><token>OR</token></entry>
       <entry>left</entry>
       <entry>logical disjunction</entry>
      </row>
     </tbody>
    </tgroup>
   </table>

   <para>
    Note that the operator precedence rules also apply to user-defined
    operators that have the same names as the built-in operators
    mentioned above.  For example, if you define a
    <quote>+</quote> operator for some custom data type it will have
    the same precedence as the built-in <quote>+</quote> operator, no
    matter what yours does.
   </para>

   <para>
    When a schema-qualified operator name is used in the
    <literal>OPERATOR</literal> syntax, as for example in:
<programlisting>
SELECT 3 OPERATOR(pg_catalog.+) 4;
</programlisting>
    the <literal>OPERATOR</literal> construct is taken to have the default precedence
    shown in <xref linkend="sql-precedence-table"/> for
    <quote>any other operator</quote>.  This is true no matter
    which specific operator appears inside <literal>OPERATOR()</literal>.
   </para>

   <note>
    <para>
     <productname>PostgreSQL</productname> versions before 9.5 used slightly different
     operator precedence rules.  In particular, <token>&lt;=</token>
     <token>&gt;=</token> and <token>&lt;&gt;</token> used to be treated as
     generic operators; <literal>IS</literal> tests used to have higher priority;
     and <literal>NOT BETWEEN</literal> and related constructs acted inconsistently,
     being taken in some cases as having the precedence of <literal>NOT</literal>
     rather than <literal>BETWEEN</literal>.  These rules were changed for better
     compliance with the SQL standard and to reduce confusion from
     inconsistent treatment of logically equivalent constructs.  In most
     cases, these changes will result in no behavioral change, or perhaps
     in <quote>no such operator</quote> failures which can be resolved by adding
     parentheses.  However there are corner cases in which a query might
     change behavior without any parsing error being reported.
    </para>
   </note>
  </sect2>
 </sect1>

 <sect1 id="sql-expressions">
  <title>Value Expressions</title>

  <indexterm zone="sql-expressions">
   <primary>expression</primary>
   <secondary>syntax</secondary>
  </indexterm>

  <indexterm zone="sql-expressions">
   <primary>value expression</primary>
  </indexterm>

  <indexterm>
   <primary>scalar</primary>
   <see>expression</see>
  </indexterm>

  <para>
   Value expressions are used in a variety of contexts, such
   as in the target list of the <command>SELECT</command> command, as
   new column values in <command>INSERT</command> or
   <command>UPDATE</command>, or in search conditions in a number of
   commands.  The result of a value expression is sometimes called a
   <firstterm>scalar</firstterm>, to distinguish it from the result of
   a table expression (which is a table).  Value expressions are
   therefore also called <firstterm>scalar expressions</firstterm> (or
   even simply <firstterm>expressions</firstterm>).  The expression
   syntax allows the calculation of values from primitive parts using
   arithmetic, logical, set, and other operations.
  </para>

  <para>
   A value expression is one of the following:

   <itemizedlist>
    <listitem>
     <para>
      A constant or literal value
     </para>
    </listitem>

    <listitem>
     <para>
      A column reference
     </para>
    </listitem>

    <listitem>
     <para>
      A positional parameter reference, in the body of a function definition
      or prepared statement
     </para>
    </listitem>

    <listitem>
     <para>
      A subscripted expression
     </para>
    </listitem>

    <listitem>
     <para>
      A field selection expression
     </para>
    </listitem>

    <listitem>
     <para>
      An operator invocation
     </para>
    </listitem>

    <listitem>
     <para>
      A function call
     </para>
    </listitem>

    <listitem>
     <para>
      An aggregate expression
     </para>
    </listitem>

    <listitem>
     <para>
      A window function call
     </para>
    </listitem>

    <listitem>
     <para>
      A type cast
     </para>
    </listitem>

    <listitem>
     <para>
      A collation expression
     </para>
    </listitem>

    <listitem>
     <para>
      A scalar subquery
     </para>
    </listitem>

    <listitem>
     <para>
      An array constructor
     </para>
    </listitem>

    <listitem>
     <para>
      A row constructor
     </para>
    </listitem>

    <listitem>
     <para>
      Another value expression in parentheses (used to group
      subexpressions and override
      precedence<indexterm><primary>parenthesis</primary></indexterm>)
     </para>
    </listitem>
   </itemizedlist>
  </para>

  <para>
   In addition to this list, there are a number of constructs that can
   be classified as an expression but do not follow any general syntax
   rules.  These generally have the semantics of a function or
   operator and are explained in the appropriate location in <xref
   linkend="functions"/>.  An example is the <literal>IS NULL</literal>
   clause.
  </para>

  <para>
   We have already discussed constants in <xref
   linkend="sql-syntax-constants"/>.  The following sections discuss
   the remaining options.
  </para>

  <sect2 id="sql-expressions-column-refs">
   <title>Column References</title>

   <indexterm>
    <primary>column reference</primary>
   </indexterm>

   <para>
    A column can be referenced in the form:
<synopsis>
<replaceable>correlation</replaceable>.<replaceable>columnname</replaceable>
</synopsis>
   </para>

   <para>
    <replaceable>correlation</replaceable> is the name of a
    table (possibly qualified with a schema name), or an alias for a table
    defined by means of a <literal>FROM</literal> clause.
    The correlation name and separating dot can be omitted if the column name
    is unique across all the tables being used in the current query.  (See also <xref linkend="queries"/>.)
   </para>
  </sect2>

  <sect2 id="sql-expressions-parameters-positional">
   <title>Positional Parameters</title>

   <indexterm>
    <primary>parameter</primary>
    <secondary>syntax</secondary>
   </indexterm>

   <indexterm>
    <primary>$</primary>
   </indexterm>

   <para>
    A positional parameter reference is used to indicate a value
    that is supplied externally to an SQL statement.  Parameters are
    used in SQL function definitions and in prepared queries.  Some
    client libraries also support specifying data values separately
    from the SQL command string, in which case parameters are used to
    refer to the out-of-line data values.
    The form of a parameter reference is:
<synopsis>
$<replaceable>number</replaceable>
</synopsis>
   </para>

   <para>
    For example, consider the definition of a function,
    <function>dept</function>, as:

<programlisting>
CREATE FUNCTION dept(text) RETURNS dept
    AS $$ SELECT * FROM dept WHERE name = $1 $$
    LANGUAGE SQL;
</programlisting>

    Here the <literal>$1</literal> references the value of the first
    function argument whenever the function is invoked.
   </para>
  </sect2>

  <sect2 id="sql-expressions-subscripts">
   <title>Subscripts</title>

   <indexterm>
    <primary>subscript</primary>
   </indexterm>

   <para>
    If an expression yields a value of an array type, then a specific
    element of the array value can be extracted by writing
<synopsis>
<replaceable>expression</replaceable>[<replaceable>subscript</replaceable>]
</synopsis>
    or multiple adjacent elements (an <quote>array slice</quote>) can be extracted
    by writing
<synopsis>
<replaceable>expression</replaceable>[<replaceable>lower_subscript</replaceable>:<replaceable>upper_subscript</replaceable>]
</synopsis>
    (Here, the brackets <literal>[ ]</literal> are meant to appear literally.)
    Each <replaceable>subscript</replaceable> is itself an expression,
    which will be rounded to the nearest integer value.
   </para>

   <para>
    In general the array <replaceable>expression</replaceable> must be
    parenthesized, but the parentheses can be omitted when the expression
    to be subscripted is just a column reference or positional parameter.
    Also, multiple subscripts can be concatenated when the original array
    is multidimensional.
    For example:

<programlisting>
mytable.arraycolumn[4]
mytable.two_d_column[17][34]
$1[10:42]
(arrayfunction(a,b))[42]
</programlisting>

    The parentheses in the last example are required.
    See <xref linkend="arrays"/> for more about arrays.
   </para>
  </sect2>

  <sect2 id="field-selection">
   <title>Field Selection</title>

   <indexterm>
    <primary>field selection</primary>
   </indexterm>

   <para>
    If an expression yields a value of a composite type (row type), then a
    specific field of the row can be extracted by writing
<synopsis>
<replaceable>expression</replaceable>.<replaceable>fieldname</replaceable>
</synopsis>
   </para>

   <para>
    In general the row <replaceable>expression</replaceable> must be
    parenthesized, but the parentheses can be omitted when the expression
    to be selected from is just a table reference or positional parameter.
    For example:

<programlisting>
mytable.mycolumn
$1.somecolumn
(rowfunction(a,b)).col3
</programlisting>

    (Thus, a qualified column reference is actually just a special case
    of the field selection syntax.)  An important special case is
    extracting a field from a table column that is of a composite type:

<programlisting>
(compositecol).somefield
(mytable.compositecol).somefield
</programlisting>

    The parentheses are required here to show that
    <structfield>compositecol</structfield> is a column name not a table name,
    or that <structname>mytable</structname> is a table name not a schema name
    in the second case.
   </para>

   <para>
    You can ask for all fields of a composite value by
    writing <literal>.*</literal>:
<programlisting>
(compositecol).*
</programlisting>
    This notation behaves differently depending on context;
    see <xref linkend="rowtypes-usage"/> for details.
   </para>
  </sect2>

  <sect2 id="sql-expressions-operator-calls">
   <title>Operator Invocations</title>

   <indexterm>
    <primary>operator</primary>
    <secondary>invocation</secondary>
   </indexterm>

   <para>
    There are two possible syntaxes for an operator invocation:
    <simplelist>
     <member><replaceable>expression</replaceable> <replaceable>operator</replaceable> <replaceable>expression</replaceable> (binary infix operator)</member>
     <member><replaceable>operator</replaceable> <replaceable>expression</replaceable> (unary prefix operator)</member>
    </simplelist>
    where the <replaceable>operator</replaceable> token follows the syntax
    rules of <xref linkend="sql-syntax-operators"/>, or is one of the
    key words <token>AND</token>, <token>OR</token>, and
    <token>NOT</token>, or is a qualified operator name in the form:
<synopsis>
<literal>OPERATOR(</literal><replaceable>schema</replaceable><literal>.</literal><replaceable>operatorname</replaceable><literal>)</literal>
</synopsis>
    Which particular operators exist and whether
    they are unary or binary depends on what operators have been
    defined by the system or the user.  <xref linkend="functions"/>
    describes the built-in operators.
   </para>
  </sect2>

  <sect2 id="sql-expressions-function-calls">
   <title>Function Calls</title>

   <indexterm>
    <primary>function</primary>
    <secondary>invocation</secondary>
   </indexterm>

   <para>
    The syntax for a function call is the name of a function
    (possibly qualified with a schema name), followed by its argument list
    enclosed in parentheses:

<synopsis>
<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional> )
</synopsis>
   </para>

   <para>
    For example, the following computes the square root of 2:
<programlisting>
sqrt(2)
</programlisting>
   </para>

   <para>
    The list of built-in functions is in <xref linkend="functions"/>.
    Other functions can be added by the user.
   </para>

   <para>
    When issuing queries in a database where some users mistrust other users,
    observe security precautions from <xref linkend="typeconv-func"/> when
    writing function calls.
   </para>

   <para>
    The arguments can optionally have names attached.
    See <xref linkend="sql-syntax-calling-funcs"/> for details.
   </para>

   <note>
    <para>
     A function that takes a single argument of composite type can
     optionally be called using field-selection syntax, and conversely
     field selection can be written in functional style.  That is, the
     notations <literal>col(table)</literal> and <literal>table.col</literal> are
     interchangeable.  This behavior is not SQL-standard but is provided
     in <productname>PostgreSQL</productname> because it allows use of functions to
     emulate <quote>computed fields</quote>.  For more information see
     <xref linkend="rowtypes-usage"/>.
    </para>
   </note>
  </sect2>

  <sect2 id="syntax-aggregates">
   <title>Aggregate Expressions</title>

   <indexterm zone="syntax-aggregates">
    <primary>aggregate function</primary>
    <secondary>invocation</secondary>
   </indexterm>

   <indexterm zone="syntax-aggregates">
    <primary>ordered-set aggregate</primary>
   </indexterm>

   <indexterm zone="syntax-aggregates">
    <primary>WITHIN GROUP</primary>
   </indexterm>

   <indexterm zone="syntax-aggregates">
    <primary>FILTER</primary>
   </indexterm>

   <para>
    An <firstterm>aggregate expression</firstterm> represents the
    application of an aggregate function across the rows selected by a
    query.  An aggregate function reduces multiple inputs to a single
    output value, such as the sum or average of the inputs.  The
    syntax of an aggregate expression is one of the following:

<synopsis>
<replaceable>aggregate_name</replaceable> (<replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
<replaceable>aggregate_name</replaceable> (ALL <replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
<replaceable>aggregate_name</replaceable> (DISTINCT <replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
<replaceable>aggregate_name</replaceable> ( * ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
<replaceable>aggregate_name</replaceable> ( [ <replaceable>expression</replaceable> [ , ... ] ] ) WITHIN GROUP ( <replaceable>order_by_clause</replaceable> ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
</synopsis>

    where <replaceable>aggregate_name</replaceable> is a previously
    defined aggregate (possibly qualified with a schema name) and
    <replaceable>expression</replaceable> is
    any value expression that does not itself contain an aggregate
    expression or a window function call.  The optional
    <replaceable>order_by_clause</replaceable> and
    <replaceable>filter_clause</replaceable> are described below.
   </para>

   <para>
    The first form of aggregate expression invokes the aggregate
    once for each input row.
    The second form is the same as the first, since
    <literal>ALL</literal> is the default.
    The third form invokes the aggregate once for each distinct value
    of the expression (or distinct set of values, for multiple expressions)
    found in the input rows.
    The fourth form invokes the aggregate once for each input row; since no
    particular input value is specified, it is generally only useful
    for the <function>count(*)</function> aggregate function.
    The last form is used with <firstterm>ordered-set</firstterm> aggregate
    functions, which are described below.
   </para>

   <para>
    Most aggregate functions ignore null inputs, so that rows in which
    one or more of the expression(s) yield null are discarded.  This
    can be assumed to be true, unless otherwise specified, for all
    built-in aggregates.
   </para>

   <para>
    For example, <literal>count(*)</literal> yields the total number
    of input rows; <literal>count(f1)</literal> yields the number of
    input rows in which <literal>f1</literal> is non-null, since
    <function>count</function> ignores nulls; and
    <literal>count(distinct f1)</literal> yields the number of
    distinct non-null values of <literal>f1</literal>.
   </para>

   <para>
    Ordinarily, the input rows are fed to the aggregate function in an
    unspecified order.  In many cases this does not matter; for example,
    <function>min</function> produces the same result no matter what order it
    receives the inputs in.  However, some aggregate functions
    (such as <function>array_agg</function> and <function>string_agg</function>) produce
    results that depend on the ordering of the input rows.  When using
    such an aggregate, the optional <replaceable>order_by_clause</replaceable> can be
    used to specify the desired ordering.  The <replaceable>order_by_clause</replaceable>
    has the same syntax as for a query-level <literal>ORDER BY</literal> clause, as
    described in <xref linkend="queries-order"/>, except that its expressions
    are always just expressions and cannot be output-column names or numbers.
    For example:
<programlisting>
SELECT array_agg(a ORDER BY b DESC) FROM table;
</programlisting>
   </para>

   <para>
    When dealing with multiple-argument aggregate functions, note that the
    <literal>ORDER BY</literal> clause goes after all the aggregate arguments.
    For example, write this:
<programlisting>
SELECT string_agg(a, ',' ORDER BY a) FROM table;
</programlisting>
    not this:
<programlisting>
SELECT string_agg(a ORDER BY a, ',') FROM table;  -- incorrect
</programlisting>
    The latter is syntactically valid, but it represents a call of a
    single-argument aggregate function with two <literal>ORDER BY</literal> keys
    (the second one being rather useless since it's a constant).
   </para>

   <para>
    If <literal>DISTINCT</literal> is specified in addition to an
    <replaceable>order_by_clause</replaceable>, then all the <literal>ORDER BY</literal>
    expressions must match regular arguments of the aggregate; that is,
    you cannot sort on an expression that is not included in the
    <literal>DISTINCT</literal> list.
   </para>

   <note>
    <para>
     The ability to specify both <literal>DISTINCT</literal> and <literal>ORDER BY</literal>
     in an aggregate function is a <productname>PostgreSQL</productname> extension.
    </para>
   </note>

   <para>
    Placing <literal>ORDER BY</literal> within the aggregate's regular argument
    list, as described so far, is used when ordering the input rows for
    general-purpose and statistical aggregates, for which ordering is
    optional.  There is a
    subclass of aggregate functions called <firstterm>ordered-set
    aggregates</firstterm> for which an <replaceable>order_by_clause</replaceable>
    is <emphasis>required</emphasis>, usually because the aggregate's computation is
    only sensible in terms of a specific ordering of its input rows.
    Typical examples of ordered-set aggregates include rank and percentile
    calculations.  For an ordered-set aggregate,
    the <replaceable>order_by_clause</replaceable> is written
    inside <literal>WITHIN GROUP (...)</literal>, as shown in the final syntax
    alternative above.  The expressions in
    the <replaceable>order_by_clause</replaceable> are evaluated once per
    input row just like regular aggregate arguments, sorted as per
    the <replaceable>order_by_clause</replaceable>'s requirements, and fed
    to the aggregate function as input arguments.  (This is unlike the case
    for a non-<literal>WITHIN GROUP</literal> <replaceable>order_by_clause</replaceable>,
    which is not treated as argument(s) to the aggregate function.)  The
    argument expressions preceding <literal>WITHIN GROUP</literal>, if any, are
    called <firstterm>direct arguments</firstterm> to distinguish them from
    the <firstterm>aggregated arguments</firstterm> listed in
    the <replaceable>order_by_clause</replaceable>.  Unlike regular aggregate
    arguments, direct arguments are evaluated only once per aggregate call,
    not once per input row.  This means that they can contain variables only
    if those variables are grouped by <literal>GROUP BY</literal>; this restriction
    is the same as if the direct arguments were not inside an aggregate
    expression at all.  Direct arguments are typically used for things like
    percentile fractions, which only make sense as a single value per
    aggregation calculation.  The direct argument list can be empty; in this
    case, write just <literal>()</literal> not <literal>(*)</literal>.
    (<productname>PostgreSQL</productname> will actually accept either spelling, but
    only the first way conforms to the SQL standard.)
   </para>

   <para>
    <indexterm>
     <primary>median</primary>
     <seealso>percentile</seealso>
    </indexterm>
    An example of an ordered-set aggregate call is:

<programlisting>
SELECT percentile_cont(0.5) WITHIN GROUP (ORDER BY income) FROM households;
 percentile_cont
-----------------
           50489
</programlisting>

   which obtains the 50th percentile, or median, value of
   the <structfield>income</structfield> column from table <structname>households</structname>.
   Here, <literal>0.5</literal> is a direct argument; it would make no sense
   for the percentile fraction to be a value varying across rows.
   </para>

   <para>
    If <literal>FILTER</literal> is specified, then only the input
    rows for which the <replaceable>filter_clause</replaceable>
    evaluates to true are fed to the aggregate function; other rows
    are discarded.  For example:
<programlisting>
SELECT
    count(*) AS unfiltered,
    count(*) FILTER (WHERE i &lt; 5) AS filtered
FROM generate_series(1,10) AS s(i);
 unfiltered | filtered
------------+----------
         10 |        4
(1 row)
</programlisting>
   </para>

   <para>
    The predefined aggregate functions are described in <xref
    linkend="functions-aggregate"/>.  Other aggregate functions can be added
    by the user.
   </para>

   <para>
    An aggregate expression can only appear in the result list or
    <literal>HAVING</literal> clause of a <command>SELECT</command> command.
    It is forbidden in other clauses, such as <literal>WHERE</literal>,
    because those clauses are logically evaluated before the results
    of aggregates are formed.
   </para>

   <para>
    When an aggregate expression appears in a subquery (see
    <xref linkend="sql-syntax-scalar-subqueries"/> and
    <xref linkend="functions-subquery"/>), the aggregate is normally
    evaluated over the rows of the subquery.  But an exception occurs
    if the aggregate's arguments (and <replaceable>filter_clause</replaceable>
    if any) contain only outer-level variables:
    the aggregate then belongs to the nearest such outer level, and is
    evaluated over the rows of that query.  The aggregate expression
    as a whole is then an outer reference for the subquery it appears in,
    and acts as a constant over any one evaluation of that subquery.
    The restriction about
    appearing only in the result list or <literal>HAVING</literal> clause
    applies with respect to the query level that the aggregate belongs to.
   </para>
  </sect2>

  <sect2 id="syntax-window-functions">
   <title>Window Function Calls</title>

   <indexterm zone="syntax-window-functions">
    <primary>window function</primary>
    <secondary>invocation</secondary>
   </indexterm>

   <indexterm zone="syntax-window-functions">
    <primary>OVER clause</primary>
   </indexterm>

   <para>
    A <firstterm>window function call</firstterm> represents the application
    of an aggregate-like function over some portion of the rows selected
    by a query.  Unlike non-window aggregate calls, this is not tied
    to grouping of the selected rows into a single output row &mdash; each
    row remains separate in the query output.  However the window function
    has access to all the rows that would be part of the current row's
    group according to the grouping specification (<literal>PARTITION BY</literal>
    list) of the window function call.
    The syntax of a window function call is one of the following:

<synopsis>
<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional>) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER <replaceable>window_name</replaceable>
<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional>) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER ( <replaceable class="parameter">window_definition</replaceable> )
<replaceable>function_name</replaceable> ( * ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER <replaceable>window_name</replaceable>
<replaceable>function_name</replaceable> ( * ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER ( <replaceable class="parameter">window_definition</replaceable> )
</synopsis>
    where <replaceable class="parameter">window_definition</replaceable>
    has the syntax
<synopsis>
[ <replaceable class="parameter">existing_window_name</replaceable> ]
[ PARTITION BY <replaceable class="parameter">expression</replaceable> [, ...] ]
[ ORDER BY <replaceable class="parameter">expression</replaceable> [ ASC | DESC | USING <replaceable class="parameter">operator</replaceable> ] [ NULLS { FIRST | LAST } ] [, ...] ]
[ <replaceable class="parameter">frame_clause</replaceable> ]
</synopsis>
    The optional <replaceable class="parameter">frame_clause</replaceable>
    can be one of
<synopsis>
{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
</synopsis>
    where <replaceable>frame_start</replaceable>
    and <replaceable>frame_end</replaceable> can be one of
<synopsis>
UNBOUNDED PRECEDING
<replaceable>offset</replaceable> PRECEDING
CURRENT ROW
<replaceable>offset</replaceable> FOLLOWING
UNBOUNDED FOLLOWING
</synopsis>
    and <replaceable>frame_exclusion</replaceable> can be one of
<synopsis>
EXCLUDE CURRENT ROW
EXCLUDE GROUP
EXCLUDE TIES
EXCLUDE NO OTHERS
</synopsis>
   </para>

   <para>
    Here, <replaceable>expression</replaceable> represents any value
    expression that does not itself contain window function calls.
   </para>

   <para>
    <replaceable>window_name</replaceable> is a reference to a named window
    specification defined in the query's <literal>WINDOW</literal> clause.
    Alternatively, a full <replaceable>window_definition</replaceable> can
    be given within parentheses, using the same syntax as for defining a
    named window in the <literal>WINDOW</literal> clause; see the
    <xref linkend="sql-select"/> reference page for details.  It's worth
    pointing out that <literal>OVER wname</literal> is not exactly equivalent to
    <literal>OVER (wname ...)</literal>; the latter implies copying and modifying the
    window definition, and will be rejected if the referenced window
    specification includes a frame clause.
   </para>

   <para>
    The <literal>PARTITION BY</literal> clause groups the rows of the query into
    <firstterm>partitions</firstterm>, which are processed separately by the window
    function.  <literal>PARTITION BY</literal> works similarly to a query-level
    <literal>GROUP BY</literal> clause, except that its expressions are always just
    expressions and cannot be output-column names or numbers.
    Without <literal>PARTITION BY</literal>, all rows produced by the query are
    treated as a single partition.
    The <literal>ORDER BY</literal> clause determines the order in which the rows
    of a partition are processed by the window function.  It works similarly
    to a query-level <literal>ORDER BY</literal> clause, but likewise cannot use
    output-column names or numbers.  Without <literal>ORDER BY</literal>, rows are
    processed in an unspecified order.
   </para>

   <para>
    The <replaceable class="parameter">frame_clause</replaceable> specifies
    the set of rows constituting the <firstterm>window frame</firstterm>, which is a
    subset of the current partition, for those window functions that act on
    the frame instead of the whole partition.  The set of rows in the frame
    can vary depending on which row is the current row.  The frame can be
    specified in <literal>RANGE</literal>, <literal>ROWS</literal>
    or <literal>GROUPS</literal> mode; in each case, it runs from
    the <replaceable>frame_start</replaceable> to
    the <replaceable>frame_end</replaceable>.
    If <replaceable>frame_end</replaceable> is omitted, the end defaults
    to <literal>CURRENT ROW</literal>.
   </para>

   <para>
    A <replaceable>frame_start</replaceable> of <literal>UNBOUNDED PRECEDING</literal> means
    that the frame starts with the first row of the partition, and similarly
    a <replaceable>frame_end</replaceable> of <literal>UNBOUNDED FOLLOWING</literal> means
    that the frame ends with the last row of the partition.
   </para>

   <para>
    In <literal>RANGE</literal> or <literal>GROUPS</literal> mode,
    a <replaceable>frame_start</replaceable> of
    <literal>CURRENT ROW</literal> means the frame starts with the current
    row's first <firstterm>peer</firstterm> row (a row that the
    window's <literal>ORDER BY</literal> clause sorts as equivalent to the
    current row), while a <replaceable>frame_end</replaceable> of
    <literal>CURRENT ROW</literal> means the frame ends with the current
    row's last peer row.
    In <literal>ROWS</literal> mode, <literal>CURRENT ROW</literal> simply
    means the current row.
   </para>

   <para>
    In the <replaceable>offset</replaceable> <literal>PRECEDING</literal>
    and <replaceable>offset</replaceable> <literal>FOLLOWING</literal> frame
    options, the <replaceable>offset</replaceable> must be an expression not
    containing any variables, aggregate functions, or window functions.
    The meaning of the <replaceable>offset</replaceable> depends on the
    frame mode:
    <itemizedlist>
     <listitem>
      <para>
       In <literal>ROWS</literal> mode,
       the <replaceable>offset</replaceable> must yield a non-null,
       non-negative integer, and the option means that the frame starts or
       ends the specified number of rows before or after the current row.
      </para>
     </listitem>
     <listitem>
      <para>
       In <literal>GROUPS</literal> mode,
       the <replaceable>offset</replaceable> again must yield a non-null,
       non-negative integer, and the option means that the frame starts or
       ends the specified number of <firstterm>peer groups</firstterm>
       before or after the current row's peer group, where a peer group is a
       set of rows that are equivalent in the <literal>ORDER BY</literal>
       ordering.  (There must be an <literal>ORDER BY</literal> clause
       in the window definition to use <literal>GROUPS</literal> mode.)
      </para>
     </listitem>
     <listitem>
      <para>
       In <literal>RANGE</literal> mode, these options require that
       the <literal>ORDER BY</literal> clause specify exactly one column.
       The <replaceable>offset</replaceable> specifies the maximum
       difference between the value of that column in the current row and
       its value in preceding or following rows of the frame.  The data type
       of the <replaceable>offset</replaceable> expression varies depending
       on the data type of the ordering column.  For numeric ordering
       columns it is typically of the same type as the ordering column,
       but for datetime ordering columns it is an <type>interval</type>.
       For example, if the ordering column is of type <type>date</type>
       or <type>timestamp</type>, one could write <literal>RANGE BETWEEN
       '1 day' PRECEDING AND '10 days' FOLLOWING</literal>.
       The <replaceable>offset</replaceable> is still required to be
       non-null and non-negative, though the meaning
       of <quote>non-negative</quote> depends on its data type.
      </para>
     </listitem>
    </itemizedlist>
    In any case, the distance to the end of the frame is limited by the
    distance to the end of the partition, so that for rows near the partition
    ends the frame might contain fewer rows than elsewhere.
   </para>

   <para>
    Notice that in both <literal>ROWS</literal> and <literal>GROUPS</literal>
    mode, <literal>0 PRECEDING</literal> and <literal>0 FOLLOWING</literal>
    are equivalent to <literal>CURRENT ROW</literal>.  This normally holds
    in <literal>RANGE</literal> mode as well, for an appropriate
    data-type-specific meaning of <quote>zero</quote>.
   </para>

   <para>
    The <replaceable>frame_exclusion</replaceable> option allows rows around
    the current row to be excluded from the frame, even if they would be
    included according to the frame start and frame end options.
    <literal>EXCLUDE CURRENT ROW</literal> excludes the current row from the
    frame.
    <literal>EXCLUDE GROUP</literal> excludes the current row and its
    ordering peers from the frame.
    <literal>EXCLUDE TIES</literal> excludes any peers of the current
    row from the frame, but not the current row itself.
    <literal>EXCLUDE NO OTHERS</literal> simply specifies explicitly the
    default behavior of not excluding the current row or its peers.
   </para>

   <para>
    The default framing option is <literal>RANGE UNBOUNDED PRECEDING</literal>,
    which is the same as <literal>RANGE BETWEEN UNBOUNDED PRECEDING AND
    CURRENT ROW</literal>.  With <literal>ORDER BY</literal>, this sets the frame to be
    all rows from the partition start up through the current row's last
    <literal>ORDER BY</literal> peer.  Without <literal>ORDER BY</literal>,
    this means all rows of the partition are included in the window frame,
    since all rows become peers of the current row.
   </para>

   <para>
    Restrictions are that
    <replaceable>frame_start</replaceable> cannot be <literal>UNBOUNDED FOLLOWING</literal>,
    <replaceable>frame_end</replaceable> cannot be <literal>UNBOUNDED PRECEDING</literal>,
    and the <replaceable>frame_end</replaceable> choice cannot appear earlier in the
    above list of <replaceable>frame_start</replaceable>
    and <replaceable>frame_end</replaceable> options than
    the <replaceable>frame_start</replaceable> choice does &mdash; for example
    <literal>RANGE BETWEEN CURRENT ROW AND <replaceable>offset</replaceable>
    PRECEDING</literal> is not allowed.
    But, for example, <literal>ROWS BETWEEN 7 PRECEDING AND 8
    PRECEDING</literal> is allowed, even though it would never select any
    rows.
   </para>

   <para>
    If <literal>FILTER</literal> is specified, then only the input
    rows for which the <replaceable>filter_clause</replaceable>
    evaluates to true are fed to the window function; other rows
    are discarded.  Only window functions that are aggregates accept
    a <literal>FILTER</literal> clause.
   </para>

   <para>
    The built-in window functions are described in <xref
    linkend="functions-window-table"/>.  Other window functions can be added by
    the user.  Also, any built-in or user-defined general-purpose or
    statistical aggregate can be used as a window function.  (Ordered-set
    and hypothetical-set aggregates cannot presently be used as window functions.)
   </para>

   <para>
    The syntaxes using <literal>*</literal> are used for calling parameter-less
    aggregate functions as window functions, for example
    <literal>count(*) OVER (PARTITION BY x ORDER BY y)</literal>.
    The asterisk (<literal>*</literal>) is customarily not used for
    window-specific functions.  Window-specific functions do not
    allow <literal>DISTINCT</literal> or <literal>ORDER BY</literal> to be used within the
    function argument list.
   </para>

   <para>
    Window function calls are permitted only in the <literal>SELECT</literal>
    list and the <literal>ORDER BY</literal> clause of the query.
   </para>

   <para>
    More information about window functions can be found in
    <xref linkend="tutorial-window"/>,
    <xref linkend="functions-window"/>, and
    <xref linkend="queries-window"/>.
   </para>
  </sect2>

  <sect2 id="sql-syntax-type-casts">
   <title>Type Casts</title>

   <indexterm>
    <primary>data type</primary>
    <secondary>type cast</secondary>
   </indexterm>

   <indexterm>
    <primary>type cast</primary>
   </indexterm>

   <indexterm>
    <primary>::</primary>
   </indexterm>

   <para>
    A type cast specifies a conversion from one data type to another.
    <productname>PostgreSQL</productname> accepts two equivalent syntaxes
    for type casts:
<synopsis>
CAST ( <replaceable>expression</replaceable> AS <replaceable>type</replaceable> )
<replaceable>expression</replaceable>::<replaceable>type</replaceable>
</synopsis>
    The <literal>CAST</literal> syntax conforms to SQL; the syntax with
    <literal>::</literal> is historical <productname>PostgreSQL</productname>
    usage.
   </para>

   <para>
    When a cast is applied to a value expression of a known type, it
    represents a run-time type conversion.  The cast will succeed only
    if a suitable type conversion operation has been defined.  Notice that this
    is subtly different from the use of casts with constants, as shown in
    <xref linkend="sql-syntax-constants-generic"/>.  A cast applied to an
    unadorned string literal represents the initial assignment of a type
    to a literal constant value, and so it will succeed for any type
    (if the contents of the string literal are acceptable input syntax for the
    data type).
   </para>

   <para>
    An explicit type cast can usually be omitted if there is no ambiguity as
    to the type that a value expression must produce (for example, when it is
    assigned to a table column); the system will automatically apply a
    type cast in such cases.  However, automatic casting is only done for
    casts that are marked <quote>OK to apply implicitly</quote>
    in the system catalogs.  Other casts must be invoked with
    explicit casting syntax.  This restriction is intended to prevent
    surprising conversions from being applied silently.
   </para>

   <para>
    It is also possible to specify a type cast using a function-like
    syntax:
<synopsis>
<replaceable>typename</replaceable> ( <replaceable>expression</replaceable> )
</synopsis>
    However, this only works for types whose names are also valid as
    function names.  For example, <literal>double precision</literal>
    cannot be used this way, but the equivalent <literal>float8</literal>
    can.  Also, the names <literal>interval</literal>, <literal>time</literal>, and
    <literal>timestamp</literal> can only be used in this fashion if they are
    double-quoted, because of syntactic conflicts.  Therefore, the use of
    the function-like cast syntax leads to inconsistencies and should
    probably be avoided.
   </para>

   <note>
    <para>
     The function-like syntax is in fact just a function call.  When
     one of the two standard cast syntaxes is used to do a run-time
     conversion, it will internally invoke a registered function to
     perform the conversion.  By convention, these conversion functions
     have the same name as their output type, and thus the <quote>function-like
     syntax</quote> is nothing more than a direct invocation of the underlying
     conversion function.  Obviously, this is not something that a portable
     application should rely on.  For further details see
     <xref linkend="sql-createcast"/>.
    </para>
   </note>
  </sect2>

  <sect2 id="sql-syntax-collate-exprs">
   <title>Collation Expressions</title>

   <indexterm>
    <primary>COLLATE</primary>
   </indexterm>

   <para>
    The <literal>COLLATE</literal> clause overrides the collation of
    an expression.  It is appended to the expression it applies to:
<synopsis>
<replaceable>expr</replaceable> COLLATE <replaceable>collation</replaceable>
</synopsis>
    where <replaceable>collation</replaceable> is a possibly
    schema-qualified identifier.  The <literal>COLLATE</literal>
    clause binds tighter than operators; parentheses can be used when
    necessary.
   </para>

   <para>
    If no collation is explicitly specified, the database system
    either derives a collation from the columns involved in the
    expression, or it defaults to the default collation of the
    database if no column is involved in the expression.
   </para>

   <para>
    The two common uses of the <literal>COLLATE</literal> clause are
    overriding the sort order in an <literal>ORDER BY</literal> clause, for
    example:
<programlisting>
SELECT a, b, c FROM tbl WHERE ... ORDER BY a COLLATE "C";
</programlisting>
    and overriding the collation of a function or operator call that
    has locale-sensitive results, for example:
<programlisting>
SELECT * FROM tbl WHERE a &gt; 'foo' COLLATE "C";
</programlisting>
    Note that in the latter case the <literal>COLLATE</literal> clause is
    attached to an input argument of the operator we wish to affect.
    It doesn't matter which argument of the operator or function call the
    <literal>COLLATE</literal> clause is attached to, because the collation that is
    applied by the operator or function is derived by considering all
    arguments, and an explicit <literal>COLLATE</literal> clause will override the
    collations of all other arguments.  (Attaching non-matching
    <literal>COLLATE</literal> clauses to more than one argument, however, is an
    error.  For more details see <xref linkend="collation"/>.)
    Thus, this gives the same result as the previous example:
<programlisting>
SELECT * FROM tbl WHERE a COLLATE "C" &gt; 'foo';
</programlisting>
    But this is an error:
<programlisting>
SELECT * FROM tbl WHERE (a &gt; 'foo') COLLATE "C";
</programlisting>
    because it attempts to apply a collation to the result of the
    <literal>&gt;</literal> operator, which is of the non-collatable data type
    <type>boolean</type>.
   </para>
  </sect2>

  <sect2 id="sql-syntax-scalar-subqueries">
   <title>Scalar Subqueries</title>

   <indexterm>
    <primary>subquery</primary>
   </indexterm>

   <para>
    A scalar subquery is an ordinary
    <command>SELECT</command> query in parentheses that returns exactly one
    row with one column.  (See <xref linkend="queries"/> for information about writing queries.)
    The <command>SELECT</command> query is executed
    and the single returned value is used in the surrounding value expression.
    It is an error to use a query that
    returns more than one row or more than one column as a scalar subquery.
    (But if, during a particular execution, the subquery returns no rows,
    there is no error; the scalar result is taken to be null.)
    The subquery can refer to variables from the surrounding query,
    which will act as constants during any one evaluation of the subquery.
    See also <xref linkend="functions-subquery"/> for other expressions involving subqueries.
   </para>

   <para>
    For example, the following finds the largest city population in each
    state:
<programlisting>
SELECT name, (SELECT max(pop) FROM cities WHERE cities.state = states.name)
    FROM states;
</programlisting>
   </para>
  </sect2>

  <sect2 id="sql-syntax-array-constructors">
   <title>Array Constructors</title>

   <indexterm>
    <primary>array</primary>
    <secondary>constructor</secondary>
   </indexterm>

   <indexterm>
    <primary>ARRAY</primary>
   </indexterm>

   <para>
    An array constructor is an expression that builds an
    array value using values for its member elements.  A simple array
    constructor
    consists of the key word <literal>ARRAY</literal>, a left square bracket
    <literal>[</literal>, a list of expressions (separated by commas) for the
    array element values, and finally a right square bracket <literal>]</literal>.
    For example:
<programlisting>
SELECT ARRAY[1,2,3+4];
  array
---------
 {1,2,7}
(1 row)
</programlisting>
    By default,
    the array element type is the common type of the member expressions,
    determined using the same rules as for <literal>UNION</literal> or
    <literal>CASE</literal> constructs (see <xref linkend="typeconv-union-case"/>).
    You can override this by explicitly casting the array constructor to the
    desired type, for example:
<programlisting>
SELECT ARRAY[1,2,22.7]::integer[];
  array
----------
 {1,2,23}
(1 row)
</programlisting>
    This has the same effect as casting each expression to the array
    element type individually.
    For more on casting, see <xref linkend="sql-syntax-type-casts"/>.
   </para>

   <para>
    Multidimensional array values can be built by nesting array
    constructors.
    In the inner constructors, the key word <literal>ARRAY</literal> can
    be omitted.  For example, these produce the same result:

<programlisting>
SELECT ARRAY[ARRAY[1,2], ARRAY[3,4]];
     array
---------------
 {{1,2},{3,4}}
(1 row)

SELECT ARRAY[[1,2],[3,4]];
     array
---------------
 {{1,2},{3,4}}
(1 row)
</programlisting>

    Since multidimensional arrays must be rectangular, inner constructors
    at the same level must produce sub-arrays of identical dimensions.
    Any cast applied to the outer <literal>ARRAY</literal> constructor propagates
    automatically to all the inner constructors.
  </para>

  <para>
    Multidimensional array constructor elements can be anything yielding
    an array of the proper kind, not only a sub-<literal>ARRAY</literal> construct.
    For example:
<programlisting>
CREATE TABLE arr(f1 int[], f2 int[]);

INSERT INTO arr VALUES (ARRAY[[1,2],[3,4]], ARRAY[[5,6],[7,8]]);

SELECT ARRAY[f1, f2, '{{9,10},{11,12}}'::int[]] FROM arr;
                     array
------------------------------------------------
 {{{1,2},{3,4}},{{5,6},{7,8}},{{9,10},{11,12}}}
(1 row)
</programlisting>
  </para>

  <para>
   You can construct an empty array, but since it's impossible to have an
   array with no type, you must explicitly cast your empty array to the
   desired type.  For example:
<programlisting>
SELECT ARRAY[]::integer[];
 array
-------
 {}
(1 row)
</programlisting>
  </para>

  <para>
   It is also possible to construct an array from the results of a
   subquery.  In this form, the array constructor is written with the
   key word <literal>ARRAY</literal> followed by a parenthesized (not
   bracketed) subquery. For example:
<programlisting>
SELECT ARRAY(SELECT oid FROM pg_proc WHERE proname LIKE 'bytea%');
                              array
------------------------------------------------------------------
 {2011,1954,1948,1952,1951,1244,1950,2005,1949,1953,2006,31,2412}
(1 row)

SELECT ARRAY(SELECT ARRAY[i, i*2] FROM generate_series(1,5) AS a(i));
              array
----------------------------------
 {{1,2},{2,4},{3,6},{4,8},{5,10}}
(1 row)
</programlisting>
   The subquery must return a single column.
   If the subquery's output column is of a non-array type, the resulting
   one-dimensional array will have an element for each row in the
   subquery result, with an element type matching that of the
   subquery's output column.
   If the subquery's output column is of an array type, the result will be
   an array of the same type but one higher dimension; in this case all
   the subquery rows must yield arrays of identical dimensionality, else
   the result would not be rectangular.
  </para>

  <para>
   The subscripts of an array value built with <literal>ARRAY</literal>
   always begin with one.  For more information about arrays, see
   <xref linkend="arrays"/>.
  </para>

  </sect2>

  <sect2 id="sql-syntax-row-constructors">
   <title>Row Constructors</title>

   <indexterm>
    <primary>composite type</primary>
    <secondary>constructor</secondary>
   </indexterm>

   <indexterm>
    <primary>row type</primary>
    <secondary>constructor</secondary>
   </indexterm>

   <indexterm>
    <primary>ROW</primary>
   </indexterm>

   <para>
    A row constructor is an expression that builds a row value (also
    called a composite value) using values
    for its member fields.  A row constructor consists of the key word
    <literal>ROW</literal>, a left parenthesis, zero or more
    expressions (separated by commas) for the row field values, and finally
    a right parenthesis.  For example:
<programlisting>
SELECT ROW(1,2.5,'this is a test');
</programlisting>
    The key word <literal>ROW</literal> is optional when there is more than one
    expression in the list.
   </para>

   <para>
    A row constructor can include the syntax
    <replaceable>rowvalue</replaceable><literal>.*</literal>,
    which will be expanded to a list of the elements of the row value,
    just as occurs when the <literal>.*</literal> syntax is used at the top level
    of a <command>SELECT</command> list (see <xref linkend="rowtypes-usage"/>).
    For example, if table <literal>t</literal> has
    columns <literal>f1</literal> and <literal>f2</literal>, these are the same:
<programlisting>
SELECT ROW(t.*, 42) FROM t;
SELECT ROW(t.f1, t.f2, 42) FROM t;
</programlisting>
   </para>

   <note>
    <para>
     Before <productname>PostgreSQL</productname> 8.2, the
     <literal>.*</literal> syntax was not expanded in row constructors, so
     that writing <literal>ROW(t.*, 42)</literal> created a two-field row whose first
     field was another row value.  The new behavior is usually more useful.
     If you need the old behavior of nested row values, write the inner
     row value without <literal>.*</literal>, for instance
     <literal>ROW(t, 42)</literal>.
    </para>
   </note>

   <para>
    By default, the value created by a <literal>ROW</literal> expression is of
    an anonymous record type.  If necessary, it can be cast to a named
    composite type &mdash; either the row type of a table, or a composite type
    created with <command>CREATE TYPE AS</command>.  An explicit cast might be needed
    to avoid ambiguity.  For example:
<programlisting>
CREATE TABLE mytable(f1 int, f2 float, f3 text);

CREATE FUNCTION getf1(mytable) RETURNS int AS 'SELECT $1.f1' LANGUAGE SQL;

-- No cast needed since only one getf1() exists
SELECT getf1(ROW(1,2.5,'this is a test'));
 getf1
-------
     1
(1 row)

CREATE TYPE myrowtype AS (f1 int, f2 text, f3 numeric);

CREATE FUNCTION getf1(myrowtype) RETURNS int AS 'SELECT $1.f1' LANGUAGE SQL;

-- Now we need a cast to indicate which function to call:
SELECT getf1(ROW(1,2.5,'this is a test'));
ERROR:  function getf1(record) is not unique

SELECT getf1(ROW(1,2.5,'this is a test')::mytable);
 getf1
-------
     1
(1 row)

SELECT getf1(CAST(ROW(11,'this is a test',2.5) AS myrowtype));
 getf1
-------
    11
(1 row)
</programlisting>
  </para>

  <para>
   Row constructors can be used to build composite values to be stored
   in a composite-type table column, or to be passed to a function that
   accepts a composite parameter.  Also,
   it is possible to compare two row values or test a row with
   <literal>IS NULL</literal> or <literal>IS NOT NULL</literal>, for example:
<programlisting>
SELECT ROW(1,2.5,'this is a test') = ROW(1, 3, 'not the same');

SELECT ROW(table.*) IS NULL FROM table;  -- detect all-null rows
</programlisting>
   For more detail see <xref linkend="functions-comparisons"/>.
   Row constructors can also be used in connection with subqueries,
   as discussed in <xref linkend="functions-subquery"/>.
  </para>

  </sect2>

  <sect2 id="syntax-express-eval">
   <title>Expression Evaluation Rules</title>

   <indexterm>
    <primary>expression</primary>
    <secondary>order of evaluation</secondary>
   </indexterm>

   <para>
    The order of evaluation of subexpressions is not defined.  In
    particular, the inputs of an operator or function are not necessarily
    evaluated left-to-right or in any other fixed order.
   </para>

   <para>
    Furthermore, if the result of an expression can be determined by
    evaluating only some parts of it, then other subexpressions
    might not be evaluated at all.  For instance, if one wrote:
<programlisting>
SELECT true OR somefunc();
</programlisting>
    then <literal>somefunc()</literal> would (probably) not be called
    at all. The same would be the case if one wrote:
<programlisting>
SELECT somefunc() OR true;
</programlisting>
    Note that this is not the same as the left-to-right
    <quote>short-circuiting</quote> of Boolean operators that is found
    in some programming languages.
   </para>

   <para>
    As a consequence, it is unwise to use functions with side effects
    as part of complex expressions.  It is particularly dangerous to
    rely on side effects or evaluation order in <literal>WHERE</literal> and <literal>HAVING</literal> clauses,
    since those clauses are extensively reprocessed as part of
    developing an execution plan.  Boolean
    expressions (<literal>AND</literal>/<literal>OR</literal>/<literal>NOT</literal> combinations) in those clauses can be reorganized
    in any manner allowed by the laws of Boolean algebra.
   </para>

   <para>
    When it is essential to force evaluation order, a <literal>CASE</literal>
    construct (see <xref linkend="functions-conditional"/>) can be
    used.  For example, this is an untrustworthy way of trying to
    avoid division by zero in a <literal>WHERE</literal> clause:
<programlisting>
SELECT ... WHERE x &gt; 0 AND y/x &gt; 1.5;
</programlisting>
    But this is safe:
<programlisting>
SELECT ... WHERE CASE WHEN x &gt; 0 THEN y/x &gt; 1.5 ELSE false END;
</programlisting>
    A <literal>CASE</literal> construct used in this fashion will defeat optimization
    attempts, so it should only be done when necessary.  (In this particular
    example, it would be better to sidestep the problem by writing
    <literal>y &gt; 1.5*x</literal> instead.)
   </para>

   <para>
    <literal>CASE</literal> is not a cure-all for such issues, however.
    One limitation of the technique illustrated above is that it does not
    prevent early evaluation of constant subexpressions.
    As described in <xref linkend="xfunc-volatility"/>, functions and
    operators marked <literal>IMMUTABLE</literal> can be evaluated when
    the query is planned rather than when it is executed.  Thus for example
<programlisting>
SELECT CASE WHEN x &gt; 0 THEN x ELSE 1/0 END FROM tab;
</programlisting>
    is likely to result in a division-by-zero failure due to the planner
    trying to simplify the constant subexpression,
    even if every row in the table has <literal>x &gt; 0</literal> so that the
    <literal>ELSE</literal> arm would never be entered at run time.
   </para>

   <para>
    While that particular example might seem silly, related cases that don't
    obviously involve constants can occur in queries executed within
    functions, since the values of function arguments and local variables
    can be inserted into queries as constants for planning purposes.
    Within <application>PL/pgSQL</application> functions, for example, using an
    <literal>IF</literal>-<literal>THEN</literal>-<literal>ELSE</literal> statement to protect
    a risky computation is much safer than just nesting it in a
    <literal>CASE</literal> expression.
   </para>

   <para>
    Another limitation of the same kind is that a <literal>CASE</literal> cannot
    prevent evaluation of an aggregate expression contained within it,
    because aggregate expressions are computed before other
    expressions in a <literal>SELECT</literal> list or <literal>HAVING</literal> clause
    are considered.  For example, the following query can cause a
    division-by-zero error despite seemingly having protected against it:
<programlisting>
SELECT CASE WHEN min(employees) > 0
            THEN avg(expenses / employees)
       END
    FROM departments;
</programlisting>
    The <function>min()</function> and <function>avg()</function> aggregates are computed
    concurrently over all the input rows, so if any row
    has <structfield>employees</structfield> equal to zero, the division-by-zero error
    will occur before there is any opportunity to test the result of
    <function>min()</function>.  Instead, use a <literal>WHERE</literal>
    or <literal>FILTER</literal> clause to prevent problematic input rows from
    reaching an aggregate function in the first place.
   </para>
  </sect2>
 </sect1>

 <sect1 id="sql-syntax-calling-funcs">
  <title>Calling Functions</title>

   <indexterm zone="sql-syntax-calling-funcs">
    <primary>notation</primary>
    <secondary>functions</secondary>
   </indexterm>

   <para>
    <productname>PostgreSQL</productname> allows functions that have named
    parameters to be called using either <firstterm>positional</firstterm> or
    <firstterm>named</firstterm> notation.  Named notation is especially
    useful for functions that have a large number of parameters, since it
    makes the associations between parameters and actual arguments more
    explicit and reliable.
    In positional notation, a function call is written with
    its argument values in the same order as they are defined in the function
    declaration.  In named notation, the arguments are matched to the
    function parameters by name and can be written in any order.
    For each notation, also consider the effect of function argument types,
    documented in <xref linkend="typeconv-func"/>.
   </para>

   <para>
    In either notation, parameters that have default values given in the
    function declaration need not be written in the call at all.  But this
    is particularly useful in named notation, since any combination of
    parameters can be omitted; while in positional notation parameters can
    only be omitted from right to left.
   </para>

   <para>
    <productname>PostgreSQL</productname> also supports
    <firstterm>mixed</firstterm> notation, which combines positional and
    named notation.  In this case, positional parameters are written first
    and named parameters appear after them.
   </para>

   <para>
    The following examples will illustrate the usage of all three
    notations, using the following function definition:
<programlisting>
CREATE FUNCTION concat_lower_or_upper(a text, b text, uppercase boolean DEFAULT false)
RETURNS text
AS
$$
 SELECT CASE
        WHEN $3 THEN UPPER($1 || ' ' || $2)
        ELSE LOWER($1 || ' ' || $2)
        END;
$$
LANGUAGE SQL IMMUTABLE STRICT;
</programlisting>
    Function <function>concat_lower_or_upper</function> has two mandatory
    parameters, <literal>a</literal> and <literal>b</literal>.  Additionally
    there is one optional parameter <literal>uppercase</literal> which defaults
    to <literal>false</literal>.  The <literal>a</literal> and
    <literal>b</literal> inputs will be concatenated, and forced to either
    upper or lower case depending on the <literal>uppercase</literal>
    parameter.  The remaining details of this function
    definition are not important here (see <xref linkend="extend"/> for
    more information).
   </para>

   <sect2 id="sql-syntax-calling-funcs-positional">
    <title>Using Positional Notation</title>

    <indexterm>
     <primary>function</primary>
     <secondary>positional notation</secondary>
    </indexterm>

    <para>
     Positional notation is the traditional mechanism for passing arguments
     to functions in <productname>PostgreSQL</productname>.  An example is:
<screen>
SELECT concat_lower_or_upper('Hello', 'World', true);
 concat_lower_or_upper
-----------------------
 HELLO WORLD
(1 row)
</screen>
     All arguments are specified in order.  The result is upper case since
     <literal>uppercase</literal> is specified as <literal>true</literal>.
     Another example is:
<screen>
SELECT concat_lower_or_upper('Hello', 'World');
 concat_lower_or_upper
-----------------------
 hello world
(1 row)
</screen>
     Here, the <literal>uppercase</literal> parameter is omitted, so it
     receives its default value of <literal>false</literal>, resulting in
     lower case output.  In positional notation, arguments can be omitted
     from right to left so long as they have defaults.
    </para>
   </sect2>

   <sect2 id="sql-syntax-calling-funcs-named">
    <title>Using Named Notation</title>

    <indexterm>
     <primary>function</primary>
     <secondary>named notation</secondary>
    </indexterm>

    <para>
     In named notation, each argument's name is specified using
     <literal>=&gt;</literal> to separate it from the argument expression.
     For example:
<screen>
SELECT concat_lower_or_upper(a =&gt; 'Hello', b =&gt; 'World');
 concat_lower_or_upper
-----------------------
 hello world
(1 row)
</screen>
     Again, the argument <literal>uppercase</literal> was omitted
     so it is set to <literal>false</literal> implicitly.  One advantage of
     using named notation is that the arguments may be specified in any
     order, for example:
<screen>
SELECT concat_lower_or_upper(a =&gt; 'Hello', b =&gt; 'World', uppercase =&gt; true);
 concat_lower_or_upper
-----------------------
 HELLO WORLD
(1 row)

SELECT concat_lower_or_upper(a =&gt; 'Hello', uppercase =&gt; true, b =&gt; 'World');
 concat_lower_or_upper
-----------------------
 HELLO WORLD
(1 row)
</screen>
    </para>

    <para>
      An older syntax based on ":=" is supported for backward compatibility:
<screen>
SELECT concat_lower_or_upper(a := 'Hello', uppercase := true, b := 'World');
 concat_lower_or_upper
-----------------------
 HELLO WORLD
(1 row)
</screen>
    </para>
   </sect2>

  <sect2 id="sql-syntax-calling-funcs-mixed">
   <title>Using Mixed Notation</title>

   <indexterm>
    <primary>function</primary>
    <secondary>mixed notation</secondary>
   </indexterm>

   <para>
    The mixed notation combines positional and named notation. However, as
    already mentioned, named arguments cannot precede positional arguments.
    For example:
<screen>
SELECT concat_lower_or_upper('Hello', 'World', uppercase =&gt; true);
 concat_lower_or_upper
-----------------------
 HELLO WORLD
(1 row)
</screen>
    In the above query, the arguments <literal>a</literal> and
    <literal>b</literal> are specified positionally, while
    <literal>uppercase</literal> is specified by name.  In this example,
    that adds little except documentation.  With a more complex function
    having numerous parameters that have default values, named or mixed
    notation can save a great deal of writing and reduce chances for error.
   </para>

   <note>
    <para>
     Named and mixed call notations currently cannot be used when calling an
     aggregate function (but they do work when an aggregate function is used
     as a window function).
    </para>
   </note>
  </sect2>
 </sect1>

</chapter>