This function takes the last comma or dot (if any) to make a clean float, ignoring thousand separator, currency or any other letter :

function tofloat($num) {
    $dotPos = strrpos($num, '.');
    $commaPos = strrpos($num, ',');
    $sep = (($dotPos > $commaPos) && $dotPos) ? $dotPos : 
        ((($commaPos > $dotPos) && $commaPos) ? $commaPos : false);
   
    if (!$sep) {
        return floatval(preg_replace("/[^0-9]/", "", $num));
    } 

    return floatval(
        preg_replace("/[^0-9]/", "", substr($num, 0, $sep)) . '.' .
        preg_replace("/[^0-9]/", "", substr($num, $sep+1, strlen($num)))
    );
}

$num = '1.999,369€';
var_dump(tofloat($num)); // float(1999.369)
$otherNum = '126,564,789.33 m²';
var_dump(tofloat($otherNum)); // float(126564789.33)

Demo : http://codepad.org/NW4e9hQH

you can also use typecasting instead of functions:

(float) $value;

There is much easier way to deal with formatted numbers:

<?php
$str = '13,232.95';
$var = (double)filter_var($str, FILTER_SANITIZE_NUMBER_FLOAT, FILTER_FLAG_ALLOW_FRACTION);
var_dump($var);
?>
double(13232.95)

To view the very large and very small numbers (eg from a database DECIMAL), without displaying scientific notation, or leading zeros.

FR : Pour afficher les très grand et très petits nombres (ex. depuis une base de données DECIMAL), sans afficher la notation scientifique, ni les zéros non significatifs.

<?php
function floattostr( $val )
{
    preg_match( "#^([\+\-]|)([0-9]*)(\.([0-9]*?)|)(0*)$#", trim($val), $o );
    return $o[1].sprintf('%d',$o[2]).($o[3]!='.'?$o[3]:'');
}
?>

<?php
echo floattostr("0000000000000001");
echo floattostr("1.00000000000000");
echo floattostr("0.00000000001000");
echo floattostr("0000.00010000000");
echo floattostr("000000010000000000.00000000000010000000000");
echo floattostr("-0000000000000.1");
echo floattostr("-00000001.100000");

// result
// 1
// 1
// 0.00000000001
// 0.0001
// 10000000000.0000000000001
// -0.1
// -1.1

?>

Easier-to-grasp-function for the ',' problem.

<?php
function Getfloat($str) {
  if(strstr($str, ",")) {
    $str = str_replace(".", "", $str); // replace dots (thousand seps) with blancs
    $str = str_replace(",", ".", $str); // replace ',' with '.'
  }
  
  if(preg_match("#([0-9\.]+)#", $str, $match)) { // search for number that may contain '.'
    return floatval($match[0]);
  } else {
    return floatval($str); // take some last chances with floatval
  }
}

echo Getfloat("$ 19.332,35-"); // will print: 19332.35
?>

locale aware floatval:

<?php
function ParseFloat($floatString){
    $LocaleInfo = localeconv();
    $floatString = str_replace($LocaleInfo["mon_thousands_sep"] , "", $floatString);
    $floatString = str_replace($LocaleInfo["mon_decimal_point"] , ".", $floatString);
    return floatval($floatString);
}
?>

Use this snippet to extract any float out of a string. You can choose how a single dot is treated with the (bool) 'single_dot_as_decimal' directive.
This function should be able to cover almost all floats that appear in an european environment.

<?php

    function float($str, $set=FALSE) 
    {            
        if(preg_match("/([0-9\.,-]+)/", $str, $match))
        {
            // Found number in $str, so set $str that number
            $str = $match[0];
            
            if(strstr($str, ',')) 
            {
                // A comma exists, that makes it easy, cos we assume it separates the decimal part.
                $str = str_replace('.', '', $str);    // Erase thousand seps
                $str = str_replace(',', '.', $str);    // Convert , to . for floatval command
                
                return floatval($str);
            }
            else
            {
                // No comma exists, so we have to decide, how a single dot shall be treated
                if(preg_match("/^[0-9]*[\.]{1}[0-9-]+$/", $str) == TRUE && $set['single_dot_as_decimal'] == TRUE)
                {
                    // Treat single dot as decimal separator
                    return floatval($str);
                    
                }
                else
                {
                    // Else, treat all dots as thousand seps
                    $str = str_replace('.', '', $str);    // Erase thousand seps
                    return floatval($str);
                }                
            }
        }
        
        else
        {
            // No number found, return zero
            return 0;
        }
    }

// Examples

echo float('foo 123,00 bar'); // returns 123.00
echo float('foo 123.00 bar' array('single_dot_as_decimal'=> TRUE)); //returns 123.000
echo float('foo 123.00 bar' array('single_dot_as_decimal'=> FALSE)); //returns 123000
echo float('foo 222.123.00 bar' array('single_dot_as_decimal'=> TRUE)); //returns 222123000
echo float('foo 222.123.00 bar' array('single_dot_as_decimal'=> FALSE)); //returns 222123000

// The decimal part can also consist of '-'
echo float('foo 123,-- bar'); // returns 123.00

?>

Big Up.
Philipp

@pillepop2003 at yahoo dot de

<?php
float('-100.00', array('single_dot_as_decimal' => true)); // whoops, returns -10000
?>

use: "/^[0-9-]*[\.]{1}[0-9-]+$/"
instead of: "/^[0-9]*[\.]{1}[0-9-]+$/"

Most of the functions listed here that deal with $ and , are unnecessarily complicated. You can use ereg_replace() to strip out ALL of the characters that will cause floatval to fail in one simple line of code:

<?php $output = floatval(ereg_replace("[^-0-9\.]","",$input)); ?>

setlocale() and floatval() duo could break your DB queries in a very simple way:

<?php
setlocale(LC_ALL, 'bg_BG', 'bgr_BGR'); 
echo floatval(0.15); // output 0,15
?>

You would need simple workaround like:

<?php
function number2db($value)
{
    $larr = localeconv();
    $search = array(
        $larr['decimal_point'], 
        $larr['mon_decimal_point'], 
        $larr['thousands_sep'], 
        $larr['mon_thousands_sep'], 
        $larr['currency_symbol'], 
        $larr['int_curr_symbol']
    );
    $replace = array('.', '.', '', '', '', '');

    return str_replace($search, $replace, $value);
}

setlocale(LC_ALL, 'bg_BG', 'bgr_BGR'); 
$testVal = floatval(0.15); // result 0,15
var_dump($testVal, number2db($testVal));

// Result:
// float(0,15) 
// string(4) "0.15" 
?>

The last getFloat() function is not completely correct.

1.000.000 and 1,000,000 and its negative variants are not correctly parsed. For the sake of comparing and to make myself clear I use the name parseFloat in stead of getFloat for the new function:

<?php
function parseFloat($ptString) {
            if (strlen($ptString) == 0) {
                    return false;
            }
            
            $pString = str_replace(" ", "", $ptString);
            
            if (substr_count($pString, ",") > 1)
                $pString = str_replace(",", "", $pString);
            
            if (substr_count($pString, ".") > 1)
                $pString = str_replace(".", "", $pString);
            
            $pregResult = array();
        
            $commaset = strpos($pString,',');
            if ($commaset === false) {$commaset = -1;}
        
            $pointset = strpos($pString,'.');
            if ($pointset === false) {$pointset = -1;}
        
            $pregResultA = array();
            $pregResultB = array();
        
            if ($pointset < $commaset) {
                preg_match('#(([-]?[0-9]+(\.[0-9])?)+(,[0-9]+)?)#', $pString, $pregResultA);
            }
            preg_match('#(([-]?[0-9]+(,[0-9])?)+(\.[0-9]+)?)#', $pString, $pregResultB);
            if ((isset($pregResultA[0]) && (!isset($pregResultB[0])
                    || strstr($preResultA[0],$pregResultB[0]) == 0
                    || !$pointset))) {
                $numberString = $pregResultA[0];
                $numberString = str_replace('.','',$numberString);
                $numberString = str_replace(',','.',$numberString);
            }
            elseif (isset($pregResultB[0]) && (!isset($pregResultA[0])
                    || strstr($pregResultB[0],$preResultA[0]) == 0
                    || !$commaset)) {
                $numberString = $pregResultB[0];
                $numberString = str_replace(',','',$numberString);
            }
            else {
                return false;
            }
            $result = (float)$numberString;
            return $result;
}
?>

Comparing of float parsing functions with the following function:

<?php
function testFloatParsing() {
    $floatvals = array(
        "22 000,76",
        "22.000,76",
        "22,000.76",
        "22 000",
        "22,000",
        "22.000",
        "22000.76",
        "22000,76",
        "1.022.000,76",
        "1,022,000.76",
        "1,000,000",
        "1.000.000",
        "1022000.76",
        "1022000,76",
        "1022000",
        "0.76",
        "0,76",
        "0.00",
        "0,00",
        "1.00",
        "1,00",
        "-22 000,76",
        "-22.000,76",
        "-22,000.76",
        "-22 000",
        "-22,000",
        "-22.000",
        "-22000.76",
        "-22000,76",
        "-1.022.000,76",
        "-1,022,000.76",
        "-1,000,000",
        "-1.000.000",
        "-1022000.76",
        "-1022000,76",
        "-1022000",
        "-0.76",
        "-0,76",
        "-0.00",
        "-0,00",
        "-1.00",
        "-1,00"
    );
    
    echo "<table>
        <tr>
            <th>String</th>
            <th>floatval()</th>
            <th>getFloat()</th>
            <th>parseFloat()</th>
        </tr>";
        
    foreach ($floatvals as $fval) {
        echo "<tr>";
        echo "<td>" . (string) $fval . "</td>";
        
        echo "<td>" . (float) floatval($fval) . "</td>";
        echo "<td>" . (float) getFloat($fval) . "</td>";
        echo "<td>" . (float) parseFloat($fval) . "</td>";
        echo "</tr>";
    }
    echo "</table>";
}
?>

Be aware the last tofloat($num).

In theory it is very useful to have a function "separator-agnostic" (I think "locale based" solutions are useless if you have to parse a user file that can have a locale different to the server).

But this can lead to misinterpretations; in short: "123,456" is "123.456" (so comma used as decimal separator) or "123456" (comma used as thousand separator).

In any case, if you really want to use it, please don't forget that this function doesn't manage negative numbers.

Float value less than 0.0001 (0.0000999999999999995) will be converted by floatval to scientific notation (exponential notation):
<?php

var_dump(floatval(0.0000999999999999995)); # >> float(0,0001)

var_dump(floatval("0.000099")); # >> float(9.9E-5)

var_dump((string)floatval(0.000099)); # >> string(6) "9.9E-5"

More elegant function with selection of decimal point (deafault ,):
<?php
function floatvaldec($v, $dec=',') { return floatval(ereg_replace("," , "." , ereg_replace("[^-0-9$dec]","",$v))); }

// examples:
echo '<br>'.floatvaldec('somthing123.456.789,12Euro') ; 
echo '<br>'.floatvaldec('x123,456 789.12 Euro', '.') ; 
echo '<br>'.floatvaldec('123.456 789,12$') ;
?>

This function converts a string to a float no matter is the decimal separator dot (.) or comma (,). It also converts integers correctly. It takes the digits from the beginning of the string and ignores all other characters.

<?php
function floatval($strValue) {
   $floatValue = ereg_replace("(^[0-9]*)(\\.|,)([0-9]*)(.*)", "\\1.\\3", $strValue);
   if (!is_numeric($floatValue)) $floatValue = ereg_replace("(^[0-9]*)(.*)", "\\1", $strValue);
   if (!is_numeric($floatValue)) $floatValue = 0;
   return $floatValue;
  }
?>

-Zipi (Finland)

Instead of using floatval which only appeared in PHP 4.2 you could juse use $variable = (float)$variable

This function doesn't seem to add any functionality that wasn't already there.

(float) would be more performant here (up to 6x times faster).

intval, floatval, doubleval, strva for PHP4 functions (intval, floatval, doubleval, strval), in PHP5 use type casting construction (i.e. '(type) parameter').

For those of you, who are looking for a function that rips the first,
but longest possible float (or at least integer) from a string,
like 123.45 from the string "Price: 123,45$"

If no useable value is found, the function returns false.

Checks for both comma and dot as decimal-separator,
but does not check for 3 digits between thousands,
so 1,234.5 is as valid as 1,23,4.5 (both will return 1234.5)

12,.3 will return 12
1,000,000 will return 1000.0 !

(if thousands separator is defined,
decimals should be defined too ... 
in fact I was too lazy to check for that too)

Here you go, and feel free to optimize the function ;)
<?php
function getFloat($pString) {
    if (strlen($pString) == 0) {
            return false;
    }
    $pregResult = array();

    $commaset = strpos($pString,',');
    if ($commaset === false) {$commaset = -1;}

    $pointset = strpos($pString,'.');
    if ($pointset === false) {$pointset = -1;}

    $pregResultA = array();
    $pregResultB = array();

    if ($pointset < $commaset) {
        preg_match('#(([-]?[0-9]+(\.[0-9])?)+(,[0-9]+)?)#', $pString, $pregResultA);
    }
    preg_match('#(([-]?[0-9]+(,[0-9])?)+(\.[0-9]+)?)#', $pString, $pregResultB);
    if ((isset($pregResultA[0]) && (!isset($pregResultB[0]) 
            || strstr($preResultA[0],$pregResultB[0]) == 0 
            || !$pointset))) {
        $numberString = $pregResultA[0];
        $numberString = str_replace('.','',$numberString);
        $numberString = str_replace(',','.',$numberString);
    }
    elseif (isset($pregResultB[0]) && (!isset($pregResultA[0]) 
            || strstr($pregResultB[0],$preResultA[0]) == 0 
            || !$commaset)) {
        $numberString = $pregResultB[0];
        $numberString = str_replace(',','',$numberString);
    }
    else {
        return false;
    }
    $result = (float)$numberString;
    return $result;
}
?>

<?php    
       $price = '1.299,00 EUR';
    //$price = 'EUR 1.299,00';
    //$price = '$1,745.09';
    //$price = '$14';
    //$price = '$.14';
    

    function floatValue($str){
        if(preg_match("/([0-9\.,-]+)/", $str, $match)){
            $value = $match[0];
            if( preg_match("/(\.\d{1,2})$/", $value, $dot_delim) ){
                $value = (float)str_replace(',', '', $value);
            }
            else if( preg_match("/(,\d{1,2})$/", $value, $comma_delim) ){
                $value = str_replace('.', '', $value); 
                $value = (float)str_replace(',', '.', $value);
            }
            else
                $value = (int)$value;
        }
        else {
            $value = 0;
        }
        return $value;
    }
    
    echo floatValue($price);
/*
    1.299,00
    1.299,00
    1,745.09
    14
    0.14
*/
?>

I get the following disturbing results:
var_dump string(10) "0.01333"
echo the string=0.01333
echo (float)string=0
echo floatval(string)=0

The string is an outcome of array_map('str_getcsv', file(...
I can't find the characters 8-10

thanks

<?php
function floatvalue($value) {
     return floatval(preg_replace('#^([-]*[0-9\.,\' ]+?)((\.|,){1}([0-9-]{1,2}))*$#e', "str_replace(array('.', ',', \"'\", ' '), '', '\\1') . '.\\4'", $value));
}
?>

It is much shorter and able to handle those one, too:
xx,-
xx,--
xx'xxx,xx

After using floatvalue() you can go forward with number_format() as usual.

floatval() does not work with "$35,234.43", as it could not handle the '$' and the ','.  The following takes care of all values, such that only numeric and the decimal sign are input into floatval().  (It probably shows I'm an old 'c' guy)...this function only lightly tested.

<?php
function strtflt($str) {
    $il = strlen($str);
    $flt = "";
    $cstr = "";
    
    for($i=0;$i<$il;$i++) {
        $cstr = substr($str, $i, 1);
        if(is_numeric($cstr) || $cstr == ".")
            $flt = $flt.$cstr;
    }
    return floatval($flt);
}
?>

Richard Vickers
[email protected]