Note: FILTER_NULL_ON_FAILURE, FILTER_REQUIRE_SCALAR, FILTER_REQUIRE_ARRAY and FILTER_FORCE_ARRAY dont have prefix "FILTER_FLAG_" but they are FLAGS!Note: FILTER_NULL_ON_FAILURE, FILTER_REQUIRE_SCALAR, FILTER_REQUIRE_ARRAY and FILTER_FORCE_ARRAY dont have prefix "FILTER_FLAG_" but they are FLAGS!<?php
+ // Handle FILTER_FLAG_STRIP_BACKTICK
+ // Fixes bug PHP_VERSION < 5.5.24
+ if ($flags & FILTER_FLAG_STRIP_BACKTICK) {
+ $output = str_replace('`', '', $output);
+ }
?>
Revised version:
<?php
function filter_sanitize_string(mixed $value, int $flags = 0): string|false|null {
if ($flags & FILTER_FLAG_EMPTY_STRING_NULL && $value === "") {
return null;
}
if (!(is_scalar($value) || is_null($value))) {
return false;
}
// Strip HTML tags and remove NUL bytes
$value = (string)$value;
$value = strip_tags($value);
$value = str_replace("\0", "", $value);
// High/Low ASCII handling (Byte-by-byte to match legacy behavior)
$output = "";
for ($i = 0, $n = strlen($value); $i < $n; $i++) {
$char = $value[$i];
$ord = ord($char);
if ($ord < 32) {
if ($flags & FILTER_FLAG_STRIP_LOW) continue;
if ($flags & FILTER_FLAG_ENCODE_LOW) {
$output .= "&#$ord;"; // Legacy decimal encoding
continue;
}
}
if ($ord > 127) {
if ($flags & FILTER_FLAG_STRIP_HIGH) continue;
if ($flags & FILTER_FLAG_ENCODE_HIGH) {
$output .= "&#$ord;"; // This turns © into © (©)
continue;
}
}
$output .= $char;
}
// Strip backticks
// Fixes bug PHP_VERSION < 5.5.24
if ($flags & FILTER_FLAG_STRIP_BACKTICK) {
$output = str_replace('`', '', $output);
}
// Legacy ONLY encoded ampersands if FILTER_FLAG_ENCODE_AMP was set
if ($flags & FILTER_FLAG_ENCODE_AMP) {
$output = str_replace('&', '&', $output);
}
// Replicate legacy quote encoding exactly (' and ")
if (!($flags & FILTER_FLAG_NO_ENCODE_QUOTES)) {
$output = str_replace(["'", '"'], [''', '"'], $output);
}
return $output;
}We can add FILTER_FLAG_EMPTY_STRING_NULL:
<?php
- function filter_sanitize_string(mixed $value, int $flags = 0): string|false {
+ function filter_sanitize_string(mixed $value, int $flags = 0): string|false|null {
+ if ($flags & FILTER_FLAG_EMPTY_STRING_NULL && $value === "") {
+ return null;
+ }
?>
For FILTER_FLAG_STRIP_BACKTICK, there is a bug history
(https://bugs.php.net/bug.php?id=69202)
so I did not attempt to code compatility through version history.We're all looking for a polyfill to replace the FILTER_SANITIZE_STRING flag, deprecated as of PHP 8.1.0,
that behaves exactly the same for scalar and non-scalar values.
(Using htmlspecialchars() and strip_tags() are not returning exactly the same output.)
<?php
function filter_sanitize_string(mixed $value, int $flags = 0): string|false {
if (!(is_scalar($value) || is_null($value))) {
return false;
}
// Strip HTML tags and remove NUL bytes
$value = (string)$value;
$value = strip_tags($value);
$value = str_replace("\0", "", $value);
// High/Low ASCII handling (Byte-by-byte to match legacy behavior)
$output = "";
for ($i = 0, $n = strlen($value); $i < $n; $i++) {
$char = $value[$i];
$ord = ord($char);
if ($ord < 32) {
if ($flags & FILTER_FLAG_STRIP_LOW) continue;
if ($flags & FILTER_FLAG_ENCODE_LOW) {
$output .= "&#$ord;"; // Legacy decimal encoding
continue;
}
}
if ($ord > 127) {
if ($flags & FILTER_FLAG_STRIP_HIGH) continue;
if ($flags & FILTER_FLAG_ENCODE_HIGH) {
$output .= "&#$ord;"; // This turns © into © (©)
continue;
}
}
$output .= $char;
}
// Legacy ONLY encoded & if FILTER_FLAG_ENCODE_AMP was set.
if ($flags & FILTER_FLAG_ENCODE_AMP) {
$output = str_replace('&', '&', $output);
}
// Replicate legacy quote encoding exactly (' and ")
if (!($flags & FILTER_FLAG_NO_ENCODE_QUOTES)) {
$output = str_replace(["'", '"'], [''', '"'], $output);
}
return $output;
}
// Tests
$string = "It's a \"test\" \x02 <script> > & ©";
// Deprecated example:
$sanitized = filter_var($string, FILTER_SANITIZE_STRING, FILTER_FLAG_STRIP_LOW | FILTER_FLAG_ENCODE_HIGH);
var_dump($sanitized);
// Polyfill example:
$sanitized = filter_sanitize_string($string, FILTER_FLAG_STRIP_LOW | FILTER_FLAG_ENCODE_HIGH);
var_dump($sanitized);
// Polyfill in filter_var() callback:
$sanitized = filter_var($string, FILTER_CALLBACK, [
'options' => function($value) {
return filter_sanitize_string($value, FILTER_FLAG_STRIP_LOW | FILTER_FLAG_ENCODE_HIGH);
}
]);
var_dump($sanitized);
?>
Output:
string(44) "It's a "test" > & ©"
string(44) "It's a "test" > & ©"
string(44) "It's a "test" > & ©"