<?xml version="1.0" encoding="iso-8859-1"?>
<!-- $Revision: 1.9 $ -->
<refentry id="function.sqlite-create-aggregate">
 <refnamediv>
  <refname>sqlite_create_aggregate</refname>
  <refpurpose>Register an aggregating UDF for use in SQL statements</refpurpose>
 </refnamediv>
 <refsect1>
  <title>Description</title>
  <methodsynopsis>
   <type>bool</type><methodname>sqlite_create_aggregate</methodname>
   <methodparam><type>resource</type><parameter>dbhandle</parameter></methodparam>
   <methodparam><type>string</type><parameter>function_name</parameter></methodparam>
   <methodparam><type>callback</type><parameter>step_func</parameter></methodparam>
   <methodparam><type>callback</type><parameter>finalize_func</parameter></methodparam>
   <methodparam choice="opt"><type>int</type><parameter>num_args</parameter></methodparam>
  </methodsynopsis>
  <para>
   <function>sqlite_create_aggregate</function> is similar to
   <function>sqlite_create_function</function> except that it registers
   functions that can be used to calculate a result aggregated across all the
   rows of a query.
  </para>
  <para>
   The key difference between this function and
   <function>sqlite_create_function</function> is that two functions are
   required to manage the aggregate; <parameter>step_func</parameter> is
   called for each row of the result set.  Your PHP function should
   accumulate the result and store it into the aggregation context.
   Once all the rows have been processed,
   <parameter>finalize_func</parameter> will be called and it should then
   take the data from the aggregation context and return the result.
   Callback functions should return a type understood by SQLite (not
   <type>array</type> for example).
  </para>
  <para>
   <example>
    <title>max_length aggregation function example</title>
    <programlisting role="php">
<![CDATA[
<?php
$data = array(
   'one',
   'two',
   'three',
   'four',
   'five',
   'six',
   'seven',
   'eight',
   'nine',
   'ten',
   );
$dbhandle = sqlite_open(':memory:');
sqlite_query($dbhandle, "CREATE TABLE strings(a)");
foreach ($data as $str) {
    $str = sqlite_escape_string($str);
    sqlite_query($dbhandle, "INSERT INTO strings VALUES ('$str')");
}

function max_len_step(&$context, $string) 
{
    if (strlen($string) > $context) {
        $context = strlen($string);
    }
}

function max_len_finalize(&$context) 
{
    return $context;
}

sqlite_create_aggregate($dbhandle, 'max_len', 'max_len_step', 'max_len_finalize');

var_dump(sqlite_array_query($dbhandle, 'SELECT max_len(a) from strings'));

?>
]]>
    </programlisting>
   </example>
  </para>
  <para>
   In this example, we are creating an aggregating function that will
   calculate the length of the longest string in one of the columns of the
   table.  For each row, the <literal>max_len_step</literal> function is
   called and passed a <parameter>context</parameter> parameter.  The context
   parameter is just like any other PHP variable and be set to hold an array
   or even an object value.  In this example, we are simply using it to hold
   the maximum length we have seen so far; if the
   <parameter>string</parameter> has a length longer than the current
   maximum, we update the context to hold this new maximum length.
  </para>
  <para>
   After all of the rows have been processed, SQLite calls the
   <literal>max_len_finalize</literal> function to determine the aggregate
   result.  Here, we could perform some kind of calculation based on the
   data found in the <parameter>context</parameter>.  In our simple example
   though, we have been calculating the result as the query progressed, so we
   simply need to return the context value.
  </para>
  <note>
   <para>
    The example above will not work correctly if the column contains binary
    data.  Take a look at the manual page for
    <function>sqlite_udf_decode_binary</function> for an explanation of why
    this is so, and an example of how to make it respect the binary encoding.
   </para>
  </note>
  <tip>
   <para>
    It is NOT recommended for you to store a copy of the values in the context
    and then process them at the end, as you would cause SQLite to use a lot of
    memory to process the query - just think of how much memory you would need
    if a million rows were stored in memory, each containing a string 32 bytes
    in length.
   </para>
  </tip>
  <tip>
   <para>
    You can use <function>sqlite_create_function</function> and
    <function>sqlite_create_aggregate</function> to override SQLite native
    SQL functions.
   </para>
  </tip>
  <para>
   See also <function>sqlite_create_function</function>,
   <function>sqlite_udf_encode_binary</function> and
   <function>sqlite_udf_decode_binary</function>.
  </para>
 </refsect1>
</refentry>
<!-- 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
 sgml-parent-document:nil
 sgml-default-dtd-file:"../../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
 -->