php-doc-en/language/functions.xml

<?xml version="1.0" encoding="utf-8"?>
<!-- $Revision$ -->
 <chapter xml:id="language.functions" xmlns="http://docbook.org/ns/docbook">
  <title>Functions</title>

  <sect1 xml:id="functions.user-defined">
   <title>User-defined functions</title>
 
   <para>
    A function may be defined using syntax such as the following:
   </para>
   <para>
    <example>
     <title>Pseudo code to demonstrate function uses</title>
     <programlisting role="php">
<![CDATA[
<?php
function foo($arg_1, $arg_2, /* ..., */ $arg_n)
{
    echo "Example function.\n";
    return $retval;
}
?>
]]>
     </programlisting>
    </example>
   </para>
   
   <simpara>
    Any valid PHP code may appear inside a function, even other
    functions and <link linkend="language.oop5.basic.class">class</link>
    definitions.
   </simpara>
   <para>
    Function names follow the same rules as other labels in PHP. A
    valid function name starts with a letter or underscore, followed
    by any number of letters, numbers, or underscores. As a regular
    expression, it would be expressed thus:
    <code>^[a-zA-Z_\x80-\xff][a-zA-Z0-9_\x80-\xff]*$</code>.
   </para>
   &tip.userlandnaming;
   <simpara>
    Functions need not be defined before they are referenced,
    <emphasis>except</emphasis> when a function is conditionally defined as
    shown in the two examples below.
   </simpara>
   <para>
    When a function is defined in a conditional manner such as the two
    examples shown. Its definition must be processed <emphasis>prior</emphasis>
    to being called.
   </para>
   <para>
    <example>
     <title>Conditional functions</title>
     <programlisting role="php">
<![CDATA[
<?php

$makefoo = true;

/* We can't call foo() from here 
   since it doesn't exist yet,
   but we can call bar() */

bar();

if ($makefoo) {
  function foo()
  {
    echo "I don't exist until program execution reaches me.\n";
  }
}

/* Now we can safely call foo()
   since $makefoo evaluated to true */

if ($makefoo) foo();

function bar() 
{
  echo "I exist immediately upon program start.\n";
}

?>
]]>
     </programlisting>
    </example>
   </para>
   <para>
    <example>
     <title>Functions within functions</title>
     <programlisting role="php">
<![CDATA[
<?php
function foo() 
{
  function bar() 
  {
    echo "I don't exist until foo() is called.\n";
  }
}

/* We can't call bar() yet
   since it doesn't exist. */

foo();

/* Now we can call bar(),
   foo()'s processing has
   made it accessible. */

bar();

?>  
]]>
     </programlisting>
    </example>
   </para>
   <para>
    All functions and classes in PHP have the global scope - they can be
    called outside a function even if they were defined inside and vice versa.
   </para>
   <simpara>
    PHP does not support function overloading, nor is it possible to
    undefine or redefine previously-declared functions.
   </simpara>
   <note>
    <simpara>
     Function names are case-insensitive for the ASCII characters <literal>A</literal> to <literal>Z</literal>, though it is usually good form
     to call functions as they appear in their declaration.
    </simpara>
   </note>   
   <simpara>
    Both <link linkend="functions.variable-arg-list">variable number of
    arguments</link> and <link linkend="functions.arguments.default">default
    arguments</link> are supported in functions. See also the function
    references for
    <function>func_num_args</function>,
    <function>func_get_arg</function>, and
    <function>func_get_args</function> for more information.
   </simpara>
   
   <para>
    It is possible to call recursive functions in PHP.
    <example>
     <title>Recursive functions</title>
     <programlisting role="php">
<![CDATA[
<?php
function recursion($a)
{
    if ($a < 20) {
        echo "$a\n";
        recursion($a + 1);
    }
}
?>
]]>
     </programlisting>
    </example>
   <note>
    <simpara>
     Recursive function/method calls with over 100-200 recursion levels can
     smash the stack and cause a termination of the current script. Especially,
     infinite recursion is considered a programming error.
    </simpara>
   </note>
   </para>

  </sect1>
 
  <sect1 xml:id="functions.arguments">
   <title>Function arguments</title>
 
   <simpara>
    Information may be passed to functions via the argument list,
    which is a comma-delimited list of expressions. The arguments are
    evaluated from left to right, before the function is actually called
    (<emphasis>eager</emphasis> evaluation).
   </simpara>

   <para>
    PHP supports passing arguments by value (the default), <link
    linkend="functions.arguments.by-reference">passing by
    reference</link>, and <link
    linkend="functions.arguments.default">default argument
    values</link>. <link linkend="functions.variable-arg-list">Variable-length
    argument lists</link> and <link linkend="functions.named-arguments">Named Arguments</link>
    are also supported.
   </para>
   <para>
    <example>
     <title>Passing arrays to functions</title>
     <programlisting role="php">
<![CDATA[
<?php
function takes_array($input)
{
    echo "$input[0] + $input[1] = ", $input[0]+$input[1];
}
?>
]]>
     </programlisting>
    </example>
   </para>
   <para>
    As of PHP 8.0.0, the list of function arguments may include a trailing comma, which
    will be ignored.  That is particularly useful in cases where the list of arguments is
    long or contains long variable names, making it convenient to list arguments vertically.
   </para>
   <example>
    <title>Function Argument List with trailing Comma</title>
    <programlisting role="php">
<![CDATA[
<?php
function takes_many_args(
    $first_arg,
    $second_arg,
    $a_very_long_argument_name,
    $arg_with_default = 5,
    $again = 'a default string', // This trailing comma was not permitted before 8.0.0.
)
{
    // ...
}
?>
]]>
    </programlisting>
   </example>
   <para>
    As of PHP 8.0.0, declaring mandatory arguments after optional arguments
    is deprecated. This can generally be resolved by dropping the default value.
    One exception to this rule are arguments of the form
    <code>Type $param = null</code>, where the &null; default makes the type implicitly
    nullable. This usage remains allowed, though it is recommended to use an
    explicit nullable type instead.
   </para>
   <example>
    <title>Declaring optional arguments after mandatory arguments</title>
    <programlisting role="php">
<![CDATA[
<?php
function foo($a = [], $b) {} // Before
function foo($a, $b) {}      // After

function bar(A $a = null, $b) {} // Still allowed
function bar(?A $a, $b) {}       // Recommended
?>
]]>
    </programlisting>
   </example>

   <sect2 xml:id="functions.arguments.by-reference">
    <title>Passing arguments by reference</title>
 
    <simpara>
     By default, function arguments are passed by value (so that if
     the value of the argument within the function is changed, it does
     not get changed outside of the function). To allow a function to modify its
     arguments, they must be passed by reference.
    </simpara>
    <para>
     To have an argument to a function always passed by reference, prepend an
     ampersand (&amp;) to the argument name in the function definition:
    </para>
    <para>
     <example>
      <title>Passing function parameters by reference</title>
      <programlisting role="php">
<![CDATA[
<?php
function add_some_extra(&$string)
{
    $string .= 'and something extra.';
}
$str = 'This is a string, ';
add_some_extra($str);
echo $str;    // outputs 'This is a string, and something extra.'
?>
]]>
      </programlisting>
     </example>
    </para>
    <para>
     It is an error to pass a value as argument which is supposed to be passed by reference.
    </para>
   </sect2>
   <sect2 xml:id="functions.arguments.default">
    <title>Default argument values</title>
 
    <para>
     A function may define C++-style default values for scalar
     arguments as follows:
    </para>
    <para>
     <example>
      <title>Use of default parameters in functions</title>
      <programlisting role="php">
<![CDATA[
<?php
function makecoffee($type = "cappuccino")
{
    return "Making a cup of $type.\n";
}
echo makecoffee();
echo makecoffee(null);
echo makecoffee("espresso");
?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
Making a cup of cappuccino.
Making a cup of .
Making a cup of espresso.
]]>
      </screen>
     </example>
    </para>
    <para>
     PHP also allows the use of <type>array</type>s and the special type &null;
     as default values, for example:
    </para>
    <para>
     <example>
      <title>Using non-scalar types as default values</title>
      <programlisting role="php">
<![CDATA[
<?php
function makecoffee($types = array("cappuccino"), $coffeeMaker = NULL)
{
    $device = is_null($coffeeMaker) ? "hands" : $coffeeMaker;
    return "Making a cup of ".join(", ", $types)." with $device.\n";
}
echo makecoffee();
echo makecoffee(array("cappuccino", "lavazza"), "teapot");
?>
]]>
      </programlisting>
     </example>
    
    </para>
    <simpara>
     The default value must be a constant expression, not (for
     example) a variable, a class member or a function call.
    </simpara>
    <para>
     Note that when using default arguments, any defaults should be on
     the right side of any non-default arguments; otherwise, things
     will not work as expected. Consider the following code snippet:
    </para>
    <para>
     <example>
      <title>Incorrect usage of default function arguments</title>
      <programlisting role="php">
<![CDATA[
<?php
function makeyogurt($type = "acidophilus", $flavour)
{
    return "Making a bowl of $type $flavour.\n";
}
 
echo makeyogurt("raspberry");   // won't work as expected
?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
Warning: Missing argument 2 in call to makeyogurt() in 
/usr/local/etc/httpd/htdocs/phptest/functest.html on line 41
Making a bowl of raspberry .
]]>
      </screen>
     </example>
    </para>
    <para>
     Now, compare the above with this:
    </para>
    <para>
     <example>
      <title>Correct usage of default function arguments</title>
      <programlisting role="php">
<![CDATA[
<?php
function makeyogurt($flavour, $type = "acidophilus")
{
    return "Making a bowl of $type $flavour.\n";
}
 
echo makeyogurt("raspberry");   // works as expected
?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
Making a bowl of acidophilus raspberry.
]]>
      </screen>
     </example>
    </para>
    <note>
     <simpara>
      Arguments that are passed by reference may have a default value.
     </simpara>
    </note>
   </sect2>

   <sect2 xml:id="functions.variable-arg-list">
    <title>Variable-length argument lists</title>

    <simpara>
     PHP has support for variable-length argument lists in
     user-defined functions by using the
     <literal>...</literal> token.
    </simpara>
    
    <note>
     <simpara>
      It is also possible to achieve variable-length arguments by using
      <function>func_num_args</function>,
      <function>func_get_arg</function>, and
      <function>func_get_args</function> functions.
      This technique is not recommended as it was used prior to the introduction
      of the <literal>...</literal> token.
     </simpara>
    </note>

    <para>
     Argument lists may include the
     <literal>...</literal> token to denote that the function accepts a
     variable number of arguments. The arguments will be passed into the
     given variable as an array; for example:

     <example>
      <title>Using <literal>...</literal> to access variable arguments</title>
      <programlisting role="php">
<![CDATA[
<?php
function sum(...$numbers) {
    $acc = 0;
    foreach ($numbers as $n) {
        $acc += $n;
    }
    return $acc;
}

echo sum(1, 2, 3, 4);
?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
10
]]>
      </screen>
     </example>
    </para>

    <para>
     <literal>...</literal> can also be used when calling functions to unpack
     an <type>array</type> or <classname>Traversable</classname> variable or
     literal into the argument list:

     <example>
      <title>Using <literal>...</literal> to provide arguments</title>
      <programlisting role="php">
<![CDATA[
<?php
function add($a, $b) {
    return $a + $b;
}

echo add(...[1, 2])."\n";

$a = [1, 2];
echo add(...$a);
?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
3
3
]]>
      </screen>
     </example>
    </para>

    <para>
     You may specify normal positional arguments before the
     <literal>...</literal> token. In this case, only the trailing arguments
     that don't match a positional argument will be added to the array
     generated by <literal>...</literal>.
    </para>

    <para>
     It is also possible to add a
     <link linkend="language.types.declarations">type declaration</link> before the
     <literal>...</literal> token. If this is present, then all arguments
     captured by <literal>...</literal> must match that parameter type.

     <example>
      <title>Type declared variable arguments</title>
      <programlisting role="php">
<![CDATA[
<?php
function total_intervals($unit, DateInterval ...$intervals) {
    $time = 0;
    foreach ($intervals as $interval) {
        $time += $interval->$unit;
    }
    return $time;
}

$a = new DateInterval('P1D');
$b = new DateInterval('P2D');
echo total_intervals('d', $a, $b).' days';

// This will fail, since null isn't a DateInterval object.
echo total_intervals('d', null);
?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
3 days
Catchable fatal error: Argument 2 passed to total_intervals() must be an instance of DateInterval, null given, called in - on line 14 and defined in - on line 2
]]>
      </screen>
     </example>
    </para>

    <para>
     Finally, variable arguments can also be passed
     <link linkend="functions.arguments.by-reference">by reference</link> by
     prefixing the <literal>...</literal> with an ampersand
     (<literal>&amp;</literal>).
    </para>

    <sect3 xml:id="functions.variable-arg-list.old">
     <title>Older versions of PHP</title>

     <para>
      No special syntax is required to note that a function is variadic;
      however access to the function's arguments must use
      <function>func_num_args</function>, <function>func_get_arg</function>
      and <function>func_get_args</function>.
     </para>

     <para>
      The first example above would be implemented as follows in old versions of PHP:

      <example>
       <title>Accessing variable arguments in old PHP versions</title>
       <programlisting role="php">
<![CDATA[
<?php
function sum() {
    $acc = 0;
    foreach (func_get_args() as $n) {
        $acc += $n;
    }
    return $acc;
}

echo sum(1, 2, 3, 4);
?>
]]>
       </programlisting>
       &example.outputs;
       <screen>
<![CDATA[
10
]]>
       </screen>
      </example>
     </para>
    </sect3>

   </sect2>

   <sect2 xml:id="functions.named-arguments">
    <title>Named Arguments</title>

    <para>
     PHP 8.0.0 introduced named arguments as an extension of the existing
     positional parameters. Named arguments allow passing arguments to a
     function based on the parameter name, rather than the parameter position.
     This makes the meaning of the argument self-documenting, makes the
     arguments order-independent and allows skipping default values arbitrarily.
    </para>

    <para>
     Named arguments are passed by prefixing the value with the parameter name
     followed by a colon. Using reserved keywords as parameter names is allowed.
     The parameter name must be an identifier, specifying dynamically
     is not allowed.
    </para>

    <example>
     <title>Named argument syntax</title>
     <programlisting role="php">
      <![CDATA[
<?php
myFunction(paramName: $value);
array_foobar(array: $value);

// NOT supported.
function_name($variableStoringParamName: $value);
?>
]]>
     </programlisting>
    </example>

    <example>
     <title>Positional arguments versus named arguments</title>
     <programlisting role="php">
      <![CDATA[
<?php
// Using positional arguments:
array_fill(0, 100, 50);

// Using named arguments:
array_fill(start_index: 0, count: 100, value: 50);
?>
]]>
     </programlisting>
    </example>

    <para>
     The order in which the named arguments are passed does not matter.
    </para>

    <example>
     <title>Same example as above with a different order of parameters</title>
     <programlisting role="php">
      <![CDATA[
<?php
array_fill(value: 50, count: 100, start_index: 0);
?>
]]>
     </programlisting>
    </example>

    <para>
     Named arguments can be combined with positional arguments. In this case,
     the named arguments must come after the positional arguments.
     It is also possible to specify only some of the optional arguments of a
     function, regardless of their order.
    </para>

    <example>
     <title>Combining named arguments with positional arguments</title>
     <programlisting role="php">
      <![CDATA[
<?php
htmlspecialchars($string, double_encode: false);
// Same as
htmlspecialchars($string, ENT_QUOTES | ENT_SUBSTITUTE | ENT_HTML401, 'UTF-8', false);
?>
]]>
     </programlisting>
    </example>

    <para>
     Passing the same parameter multiple times results in an Error exception.
    </para>

    <example>
     <title>Error exception when passing the same parameter multiple times</title>
     <programlisting role="php">
      <![CDATA[
<?php
function foo($param) { ... }

foo(param: 1, param: 2);
// Error: Named parameter $param overwrites previous argument
foo(1, param: 2);
// Error: Named parameter $param overwrites previous argument
?>
]]>
     </programlisting>
    </example>

   </sect2>
  </sect1>
 
  <sect1 xml:id="functions.returning-values">
   <title>Returning values</title>
 
   <para>
    Values are returned by using the optional return statement. Any
    type may be returned, including arrays and objects. This causes the
    function to end its execution immediately and pass control back to
    the line from which it was called. See <function>return</function>
    for more information.
   </para>

   <note>
    <para>
     If the <function>return</function> is omitted the value &null; will be
     returned.
    </para>
   </note>

   <sect2>
    <title>Use of return</title>
    <para>
     <example>
      <title>Use of <function>return</function></title>
      <programlisting role="php">
<![CDATA[
<?php
function square($num)
{
    return $num * $num;
}
echo square(4);   // outputs '16'.
?>
]]>
      </programlisting>
     </example>
    </para>

    <para>
     A function can not return multiple values, but similar results can be
     obtained by returning an array.
    </para>
    <para>
     <example>
      <title>Returning an array to get multiple values</title>
      <programlisting role="php">
<![CDATA[
<?php
function small_numbers()
{
    return [0, 1, 2];
}
// Array destructuring will collect each member of the array individually
[$zero, $one, $two] = small_numbers();

// Prior to 7.1.0, the only equivalent alternative is using list() construct
list($zero, $one, $two) = small_numbers();

?>
]]>
      </programlisting>
     </example>
    </para>
    <para>
     To return a reference from a function, use the reference operator &amp; in
     both the function declaration and when assigning the returned value to a
     variable:
    </para>
    <para>
     <example>
      <title>Returning a reference from a function</title>
      <programlisting role="php">
<![CDATA[
<?php
function &returns_reference()
{
    return $someref;
}

$newref =& returns_reference();
?>
]]>
      </programlisting>
     </example>
    </para>
    <simpara>
     For more information on references, please check out <link
     linkend="language.references">References Explained</link>.
    </simpara>
   </sect2>
  </sect1>

  <sect1 xml:id="functions.variable-functions">
   <title>Variable functions</title>

   <para>
    PHP supports the concept of variable functions. This means that if
    a variable name has parentheses appended to it, PHP will look for
    a function with the same name as whatever the variable evaluates
    to, and will attempt to execute it. Among other things, this can
    be used to implement callbacks, function tables, and so forth.
   </para>
   <para>
    Variable functions won't work with language constructs such 
    as <function>echo</function>, <function>print</function>,
    <function>unset</function>, <function>isset</function>,
    <function>empty</function>, <function>include</function>,
    <function>require</function> and the like. Utilize wrapper functions to make
    use of any of these constructs as variable functions.
   </para>
   <para>
    <example>
     <title>Variable function example</title>
     <programlisting role="php">
<![CDATA[
<?php
function foo() {
    echo "In foo()<br />\n";
}

function bar($arg = '')
{
    echo "In bar(); argument was '$arg'.<br />\n";
}

// This is a wrapper function around echo
function echoit($string)
{
    echo $string;
}

$func = 'foo';
$func();        // This calls foo()

$func = 'bar';
$func('test');  // This calls bar()

$func = 'echoit';
$func('test');  // This calls echoit()
?>
]]>
     </programlisting>
    </example>
   </para>
   <para>
    Object methods can also be called with the variable functions syntax.
    <example>
     <title>Variable method example</title>
     <programlisting role="php">
<![CDATA[
<?php
class Foo
{
    function Variable()
    {
        $name = 'Bar';
        $this->$name(); // This calls the Bar() method
    }
    
    function Bar()
    {
        echo "This is Bar";
    }
}

$foo = new Foo();
$funcname = "Variable";
$foo->$funcname();  // This calls $foo->Variable()

?>
]]>
     </programlisting>
    </example>
   </para>
   <para>
    When calling static methods, the function call is stronger than the static property operator:
    <example>
     <title>Variable method example with static properties</title>
     <programlisting role="php">
<![CDATA[
<?php
class Foo
{
    static $variable = 'static property';
    static function Variable()
    {
        echo 'Method Variable called';
    }
}

echo Foo::$variable; // This prints 'static property'. It does need a $variable in this scope.
$variable = "Variable";
Foo::$variable();  // This calls $foo->Variable() reading $variable in this scope.

?>
]]>
     </programlisting>
    </example>
   </para>
   <para>
    <example>
     <title>Complex callables</title>
     <programlisting role="php">
<![CDATA[
<?php
class Foo
{
    static function bar()
    {
        echo "bar\n";
    }
    function baz()
    {
        echo "baz\n";
    }
}

$func = array("Foo", "bar");
$func(); // prints "bar"
$func = array(new Foo, "baz");
$func(); // prints "baz"
$func = "Foo::bar";
$func(); // prints "bar"
?>
]]>
     </programlisting>
    </example>
   </para>

   <sect2 role="seealso">
    &reftitle.seealso;
    <para>
     <simplelist>
      <member><function>is_callable</function></member>
      <member><function>call_user_func</function></member>
      <member><function>function_exists</function></member>
      <member><link linkend="language.variables.variable">variable variables</link></member>
     </simplelist>
    </para>
   </sect2>
  </sect1>
  
  <sect1 xml:id="functions.internal">
   <title>Internal (built-in) functions</title>
   
   <para>
    PHP comes standard with many functions and constructs. There are also
    functions that require specific PHP extensions compiled in, otherwise 
    fatal "undefined function" errors will appear. For example, to use 
    <link linkend="ref.image">image</link> functions such as 
    <function>imagecreatetruecolor</function>, PHP must be compiled with
    <productname>GD</productname> support. Or, to use
    <function>mysqli_connect</function>, PHP must be compiled with
    <link linkend="book.mysqli">MySQLi</link> support. There are many core functions
    that are included in every version of PHP, such as the
    <link linkend="ref.strings">string</link> and 
    <link linkend="ref.var">variable</link> functions. A call 
    to <function>phpinfo</function> or
    <function>get_loaded_extensions</function> will show which extensions are
    loaded into PHP. Also note that many extensions are enabled by default and
    that the PHP manual is split up by extension. See the
    <link linkend="configuration">configuration</link>,
    <link linkend="install">installation</link>, and individual
    extension chapters, for information on how to set up PHP.
   </para>
   <para>
    Reading and understanding a function's prototype is explained within the
    manual section titled <link linkend="about.prototypes">how to read a
    function definition</link>. It's important to realize what a function
    returns or if a function works directly on a passed in value. For example,
    <function>str_replace</function> will return the modified string while 
    <function>usort</function> works on the actual passed in variable
    itself. Each manual page also has specific information for each
    function like information on function parameters, behavior changes,
    return values for both success and failure, and availability information.
    Knowing these important (yet often subtle) differences is crucial for
    writing correct PHP code.
   </para>
   <note>
    <simpara>
     If the parameters given to a function are not what it expects, such as 
     passing an <type>array</type> where a <type>string</type> is expected, 
     the return value of the function is undefined. In this case it will
     likely return &null; but this is just a convention, and cannot be relied 
     upon.
     As of PHP 8.0.0, a <classname>TypeError</classname> exception is supposed to
     be thrown in this case.
    </simpara>
   </note>

   <sect2 role="seealso">
    &reftitle.seealso;
    <para>
     <simplelist>
      <member><function>function_exists</function></member>
      <member><link linkend="funcref">the function reference</link></member>
      <member><function>get_extension_funcs</function></member>
      <member><function>dl</function></member>
     </simplelist>
    </para>
   </sect2>
  </sect1>
  
  <sect1 xml:id="functions.anonymous">
   <title>Anonymous functions</title>

   <simpara>
    Anonymous functions, also known as <literal>closures</literal>, allow the
    creation of functions which have no specified name. They are most useful as
    the value of <type>callable</type>
    parameters, but they have many other uses.
   </simpara>
   <simpara>
    Anonymous functions are implemented using the <link linkend="class.closure">
    <classname>Closure</classname></link> class.
   </simpara>

   <example>
    <title>Anonymous function example</title>
    <programlisting role="php">
<![CDATA[
<?php
echo preg_replace_callback('~-([a-z])~', function ($match) {
    return strtoupper($match[1]);
}, 'hello-world');
// outputs helloWorld
?>
]]>
    </programlisting>
   </example>

   <simpara>
    Closures can also be used as the values of variables; PHP automatically 
    converts such expressions into instances of the
    <classname>Closure</classname> internal class. Assigning a closure to a
    variable uses the same syntax as any other assignment, including the
    trailing semicolon:
   </simpara>

   <example>
    <title>Anonymous function variable assignment example</title>
    <programlisting role="php">
<![CDATA[
<?php
$greet = function($name)
{
    printf("Hello %s\r\n", $name);
};

$greet('World');
$greet('PHP');
?>
]]>
    </programlisting>
   </example>
   
   <simpara>
    Closures may also inherit variables from the parent scope. Any such
    variables must be passed to the <literal>use</literal> language construct.
    As of PHP 7.1, these variables must not include &link.superglobals;,
    <varname>$this</varname>, or variables with the same name as a parameter.
    A return type declaration of the function has to be placed
    <emphasis>after</emphasis> the <literal>use</literal> clause.
   </simpara>

   <example>
    <title>Inheriting variables from the parent scope</title>
    <programlisting role="php">
<![CDATA[
<?php
$message = 'hello';

// No "use"
$example = function () {
    var_dump($message);
};
$example();

// Inherit $message
$example = function () use ($message) {
    var_dump($message);
};
$example();

// Inherited variable's value is from when the function
// is defined, not when called
$message = 'world';
$example();

// Reset message
$message = 'hello';

// Inherit by-reference
$example = function () use (&$message) {
    var_dump($message);
};
$example();

// The changed value in the parent scope
// is reflected inside the function call
$message = 'world';
$example();

// Closures can also accept regular arguments
$example = function ($arg) use ($message) {
    var_dump($arg . ' ' . $message);
};
$example("hello");

// Return type declaration comes after the use clause
$example = function () use ($message): string {
    return "hello $message";
};
var_dump($example());
?>
]]>
    </programlisting>
    &example.outputs.similar;
    <screen>
<![CDATA[
Notice: Undefined variable: message in /example.php on line 6
NULL
string(5) "hello"
string(5) "hello"
string(5) "hello"
string(5) "world"
string(11) "hello world"
string(11) "hello world"
]]>
    </screen>
   </example>

   <para>
    As of PHP 8.0.0, the list of scope-inherited variables may include a trailing
    comma, which will be ignored.
   </para>
   <simpara>
    Inheriting variables from the parent scope is <emphasis>not</emphasis> 
    the same as using global variables. 
    Global variables exist in the global scope, which is the same no
    matter what function is executing. The parent scope of a closure is the
    function in which the closure was declared (not necessarily the function it
    was called from). See the following example:
   </simpara>

   <example>
    <title>Closures and scoping</title>
    <programlisting role="php">
<![CDATA[
<?php
// A basic shopping cart which contains a list of added products
// and the quantity of each product. Includes a method which
// calculates the total price of the items in the cart using a
// closure as a callback.
class Cart
{
    const PRICE_BUTTER  = 1.00;
    const PRICE_MILK    = 3.00;
    const PRICE_EGGS    = 6.95;

    protected $products = array();
    
    public function add($product, $quantity)
    {
        $this->products[$product] = $quantity;
    }
    
    public function getQuantity($product)
    {
        return isset($this->products[$product]) ? $this->products[$product] :
               FALSE;
    }
    
    public function getTotal($tax)
    {
        $total = 0.00;
        
        $callback =
            function ($quantity, $product) use ($tax, &$total)
            {
                $pricePerItem = constant(__CLASS__ . "::PRICE_" .
                    strtoupper($product));
                $total += ($pricePerItem * $quantity) * ($tax + 1.0);
            };
        
        array_walk($this->products, $callback);
        return round($total, 2);
    }
}

$my_cart = new Cart;

// Add some items to the cart
$my_cart->add('butter', 1);
$my_cart->add('milk', 3);
$my_cart->add('eggs', 6);

// Print the total with a 5% sales tax.
print $my_cart->getTotal(0.05) . "\n";
// The result is 54.29
?>
]]>
    </programlisting>
   </example>

   <example>
    <title>Automatic binding of <literal>$this</literal></title>
    <programlisting role="php">
<![CDATA[
<?php

class Test
{
    public function testing()
    {
        return function() {
            var_dump($this);
        };
    }
}

$object = new Test;
$function = $object->testing();
$function();
    
?>
]]>
    </programlisting>
    &example.outputs;
    <screen>
<![CDATA[
object(Test)#1 (0) {
}
]]>
    </screen>
   </example>

   <para>
    When declared in the context of a class, the current class is
    automatically bound to it, making <literal>$this</literal> available
    inside of the function's scope. If this automatic binding of the current
    class is not wanted, then
    <link linkend="functions.anonymous-functions.static">static anonymous
    functions</link> may be used instead.
   </para>

   <sect2 xml:id="functions.anonymous-functions.static">
    <title>Static anonymous functions</title>
    <para>
     Anonymous functions may be declared statically. This
     prevents them from having the current class automatically bound to
     them. Objects may also not be bound to them at runtime.
    </para>
    <para>
     <example>
      <title>Attempting to use <literal>$this</literal> inside a static anonymous function</title>
      <programlisting role="php">
<![CDATA[
<?php

class Foo
{
    function __construct()
    {
        $func = static function() {
            var_dump($this);
        };
        $func();
    }
};
new Foo();

?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
Notice: Undefined variable: this in %s on line %d
NULL
]]>
      </screen>
     </example>
    </para>

    <para>
     <example>
      <title>Attempting to bind an object to a static anonymous function</title>
      <programlisting role="php">
<![CDATA[
<?php

$func = static function() {
    // function body
};
$func = $func->bindTo(new StdClass);
$func();

?>
]]>
      </programlisting>
      &example.outputs;
      <screen>
<![CDATA[
Warning: Cannot bind an instance to a static closure in %s on line %d
]]>
      </screen>
     </example>
    </para>
   </sect2>
   
   <sect2 role="changelog">
    &reftitle.changelog;
    <para>
     <informaltable>
      <tgroup cols="2">
       <thead>
        <row>
         <entry>&Version;</entry>
         <entry>&Description;</entry>
        </row>
       </thead>
       <tbody>
        <row>
         <entry>7.1.0</entry>
         <entry>
          Anonymous functions may not close over &link.superglobals;,
          <varname>$this</varname>, or any variable with the same name as a
          parameter.
         </entry>
        </row>
       </tbody>
      </tgroup>
     </informaltable>
    </para>
   </sect2>

   <sect2 role="notes">
    &reftitle.notes;
    <note>
     <simpara>
      It is possible to use <function>func_num_args</function>,
      <function>func_get_arg</function>, and <function>func_get_args</function>
      from within a closure.
     </simpara>
    </note>
   </sect2>

  </sect1>
  <sect1 xml:id="functions.arrow">
   <title>Arrow Functions</title>

   <simpara>
    Arrow functions were introduced in PHP 7.4 as a more concise syntax for 
    <link linkend="functions.anonymous">anonymous functions</link>.
   </simpara>
   <simpara>
    Both anonymous functions and arrow functions are implemented using the 
    <link linkend="class.closure"><classname>Closure</classname></link> class.
   </simpara>

   <simpara>
    Arrow functions have the basic form 
    <code>fn (argument_list) =&gt; expr</code>.
   </simpara>

   <simpara>
    Arrow functions support the same features as 
    <link linkend="functions.anonymous">anonymous functions</link>, 
    except that using variables from the parent scope is always automatic.
   </simpara>

   <simpara>
    When a variable used in the expression is defined in the parent scope 
    it will be implicitly captured by-value.
    In the following example, the functions <varname>$fn1</varname> and 
    <varname>$fn2</varname> behave the same way. 
   </simpara>

   <para>
    <example>
     <title>Arrow functions capture variables by value automatically</title>
     <programlisting role="php">
<![CDATA[
<?php

$y = 1;
 
$fn1 = fn($x) => $x + $y;
// equivalent to using $y by value:
$fn2 = function ($x) use ($y) {
    return $x + $y;
};

var_export($fn1(3));
?>
]]>
     </programlisting>
     &example.outputs;
     <screen>
<![CDATA[
4
]]>
      </screen>
    </example>
   </para>
   <simpara>
    This also works if the arrow functions are nested: 
   </simpara>
   <para>
    <example>
     <title>Arrow functions capture variables by value automatically, even when nested</title>
     <programlisting role="php">
<![CDATA[
<?php

$z = 1;
$fn = fn($x) => fn($y) => $x * $y + $z;
// Outputs 51
var_export($fn(5)(10));
?>
]]>
     </programlisting>
    </example>
   </para>
   <simpara>
    Similarly to anonymous functions, 
    the arrow function syntax allows arbitrary function signatures,
    including parameter and return types, default values, variadics,
    as well as by-reference passing and returning.
    All of the following are valid examples of arrow functions: 
   </simpara>
   <para>
    <example>
     <title>Examples of arrow functions</title>
     <programlisting role="php">
<![CDATA[
<?php

fn(array $x) => $x;
static fn(): int => $x;
fn($x = 42) => $x;
fn(&$x) => $x;
fn&($x) => $x;
fn($x, ...$rest) => $rest;

?>
]]>
     </programlisting>
    </example>
   </para>
   <simpara>
    Arrow functions use by-value variable binding.
    This is roughly equivalent to performing a <code>use($x)</code> for every 
    variable <varname>$x</varname> used inside the arrow function.
    A by-value binding means that it is not possible to modify any values 
    from the outer scope. 
    <link linkend="functions.anonymous">Anonymous functions</link> 
    can be used instead for by-ref bindings.
   </simpara>
   <para>
    <example>
     <title>Values from the outer scope cannot be modified by arrow functions</title>
     <programlisting role="php">
<![CDATA[
<?php

$x = 1;
$fn = fn() => $x++; // Has no effect
$fn();
var_export($x);  // Outputs 1

?>
]]>
     </programlisting>
    </example>
   </para>
   
   <sect2 role="changelog">
    &reftitle.changelog;
    <para>
     <informaltable>
      <tgroup cols="2">
       <thead>
        <row>
         <entry>&Version;</entry>
         <entry>&Description;</entry>
        </row>
       </thead>
       <tbody>
        <row>
         <entry>7.4.0</entry>
         <entry>
          Arrow functions became available.
         </entry>
        </row>
       </tbody>
      </tgroup>
     </informaltable>
    </para>
   </sect2>

   <sect2 role="notes">
    &reftitle.notes;
    <note>
     <simpara>
      It is possible to use <function>func_num_args</function>,
      <function>func_get_arg</function>, and <function>func_get_args</function>
      from within an arrow function.
     </simpara>
    </note>
   </sect2>
  </sect1>

  <sect1 xml:id="functions.first_class_callable_syntax">
   <title>First class callable syntax</title>

   <para>
    The first class callable syntax is introduced as of PHP 8.1.0, as a way of creating <link linkend="functions.anonymous">anonymous functions</link> from <link linkend="language.types.callable">callable</link>.
    It supersedes existing callable syntax using strings and arrays.
    The advantage of this syntax is that it is accessible to static analysis, and uses the scope at the point where the callable is acquired.
   </para>

   <para>
    <code>CallableExpr(...)</code> syntax is used to create a <classname>Closure</classname> object from callable. <code>CallableExpr</code> accepts any expression that can be directly called in the PHP grammar:
    <example>
     <title>Simple first class callable syntax</title>
     <programlisting role="php">
<![CDATA[
<?php

class Foo {
   public function method() {}
   public static function staticmethod() {}
   public function __invoke() {}
}

$obj = new Foo();
$classStr = 'Foo';
$methodStr = 'method';
$staticmethodStr = 'staticmethod';


$f1 = strlen(...);
$f2 = $obj(...);  // invokable object
$f3 = $obj->method(...);
$f4 = $obj->$methodStr(...);
$f5 = Foo::staticmethod(...);
$f6 = $classStr::$staticmethodStr(...);

// traditional callable using string, array
$f7 = 'strlen'(...);
$f8 = [$obj, 'method'](...);
$f9 = [Foo::class, 'staticmethod'](...);
?>
]]>
     </programlisting>
    </example>
   </para>

   <note>
    <para>
      The <code>...</code> is part of the syntax, and not an omission.
    </para>
   </note>

   <para>
    <code>CallableExpr(...)</code> has the same semantics as <methodname>Closure::fromCallable</methodname>.
    That is, unlike callable using strings and arrays, <code>CallableExpr(...)</code> respects the scope at the point where it is created:
    <example>
     <title>Scope comparison of <code>CallableExpr(...)</code> and traditional callable</title>
     <programlisting role="php">
<![CDATA[
<?php

class Foo {
    public function getPrivateMethod() {
        return [$this, 'privateMethod'];
    }

    private function privateMethod() {
        echo __METHOD__, "\n";
    }
}

$foo = new Foo;
$privateMethod = $foo->getPrivateMethod();
$privateMethod();
// Fatal error: Call to private method Foo::privateMethod() from global scope
// This is because call is performed outside from Foo and visibility will be checked from this point.

class Foo1 {
    public function getPrivateMethod() {
        // Uses the scope where the callable is acquired.
        return $this->privateMethod(...); // identical to Closure::fromCallable([$this, 'privateMethod']);
    }

    private function privateMethod() {
        echo __METHOD__, "\n";
    }
}

$foo1 = new Foo1;
$privateMethod = $foo1->getPrivateMethod();
$privateMethod();  // Foo1::privateMethod
?>
]]>
     </programlisting>
    </example>

   </para>

   <note>
    <para>
     Object creation by this syntax (e.g <code>new Foo(...)</code>) is not supported, because <code>new Foo()</code> syntax is not considered a call.
    </para>
   </note>

   <note>
    <para>
     The first-class callable syntax cannot be combined with the nullsafe operator. Both of the following result in a compile-time error:
     <example>
      <programlisting role="php">
<![CDATA[
<?php
$obj?->method(...);
$obj?->prop->method(...);
?>
]]>
      </programlisting>
     </example>
    </para>
   </note>
  </sect1>

 </chapter>
 
<!-- Keep this comment at the end of the file
Local variables:
mode: sgml
sgml-omittag:t
sgml-shorttag:t
sgml-minimize-attributes:nil
sgml-always-quote-attributes:t
sgml-indent-step:1
sgml-indent-data:t
indent-tabs-mode:nil
sgml-parent-document:nil
sgml-default-dtd-file:"~/.phpdoc/manual.ced"
sgml-exposed-tags:nil
sgml-local-catalogs:nil
sgml-local-ecat-files:nil
End:
vim600: syn=xml fen fdm=syntax fdl=2 si
vim: et tw=78 syn=sgml
vi: ts=1 sw=1
-->