<?php
class Requests_IDNAEncoder
{
const ACE_PREFIX = 'xn--';
const BOOTSTRAP_BASE = 36;
const BOOTSTRAP_TMIN = 1;
const BOOTSTRAP_TMAX = 26;
const BOOTSTRAP_SKEW = 38;
const BOOTSTRAP_DAMP = 700;
const BOOTSTRAP_INITIAL_BIAS = 72;
const BOOTSTRAP_INITIAL_N = 128;
public static function encode($string)
{
$parts = explode('.', $string);
foreach ($parts as &$part) {
$part = self::to_ascii($part);
}
return implode('.', $parts);
}
public static function to_ascii($string)
{
if (self::is_ascii($string)) {
if (strlen($string) < 64) {
return $string;
}
throw new Requests_Exception('Provided string is too long', 'idna.provided_too_long', $string);
}
$string = self::nameprep($string);
if (self::is_ascii($string)) {
if (strlen($string) < 64) {
return $string;
}
throw new Requests_Exception('Prepared string is too long', 'idna.prepared_too_long', $string);
}
if (strpos($string, self::ACE_PREFIX) === 0) {
throw new Requests_Exception('Provided string begins with ACE prefix', 'idna.provided_is_prefixed', $string);
}
$string = self::punycode_encode($string);
$string = self::ACE_PREFIX . $string;
if (strlen($string) < 64) {
return $string;
}
throw new Requests_Exception('Encoded string is too long', 'idna.encoded_too_long', $string);
}
protected static function is_ascii($string)
{
return preg_match('/(?:[^\\x00-\\x7F])/', $string) !== 1;
}
protected static function nameprep($string)
{
return $string;
}
protected static function utf8_to_codepoints($input)
{
$codepoints = array();
$strlen = strlen($input);
for ($position = 0; $position < $strlen; $position++) {
$value = ord($input[$position]);
if ((~$value & 0x80) === 0x80) {
$character = $value;
$length = 1;
$remaining = 0;
} elseif (($value & 0xe0) === 0xc0) {
$character = ($value & 0x1f) << 6;
$length = 2;
$remaining = 1;
} elseif (($value & 0xf0) === 0xe0) {
$character = ($value & 0xf) << 12;
$length = 3;
$remaining = 2;
} elseif (($value & 0xf8) === 0xf0) {
$character = ($value & 0x7) << 18;
$length = 4;
$remaining = 3;
} else {
throw new Requests_Exception('Invalid Unicode codepoint', 'idna.invalidcodepoint', $value);
}
if ($remaining > 0) {
if ($position + $length > $strlen) {
throw new Requests_Exception('Invalid Unicode codepoint', 'idna.invalidcodepoint', $character);
}
for ($position++; $remaining > 0; $position++) {
$value = ord($input[$position]);
if (($value & 0xc0) !== 0x80) {
throw new Requests_Exception('Invalid Unicode codepoint', 'idna.invalidcodepoint', $character);
}
$character |= ($value & 0x3f) << --$remaining * 6;
}
$position--;
}
if ($length > 1 && $character <= 0x7f || $length > 2 && $character <= 0x7ff || $length > 3 && $character <= 0xffff || ($character & 0xfffe) === 0xfffe || $character >= 0xfdd0 && $character <= 0xfdef || ($character > 0xd7ff && $character < 0xf900 || $character < 0x20 || $character > 0x7e && $character < 0xa0 || $character > 0xefffd)) {
throw new Requests_Exception('Invalid Unicode codepoint', 'idna.invalidcodepoint', $character);
}
$codepoints[] = $character;
}
return $codepoints;
}
public static function punycode_encode($input)
{
$output = '';
$n = self::BOOTSTRAP_INITIAL_N;
$delta = 0;
$bias = self::BOOTSTRAP_INITIAL_BIAS;
$h = $b = 0;
$codepoints = self::utf8_to_codepoints($input);
$extended = array();
foreach ($codepoints as $char) {
if ($char < 128) {
$output .= chr($char);
$h++;
} elseif ($char < $n) {
throw new Requests_Exception('Invalid character', 'idna.character_outside_domain', $char);
} else {
$extended[$char] = true;
}
}
$extended = array_keys($extended);
sort($extended);
$b = $h;
if (strlen($output) > 0) {
$output .= '-';
}
while ($h < count($codepoints)) {
$m = array_shift($extended);
$delta += ($m - $n) * ($h + 1);
$n = $m;
for ($num = 0; $num < count($codepoints); $num++) {
$c = $codepoints[$num];
if ($c < $n) {
$delta++;
} elseif ($c === $n) {
$q = $delta;
for ($k = self::BOOTSTRAP_BASE;; $k += self::BOOTSTRAP_BASE) {
if ($k <= $bias + self::BOOTSTRAP_TMIN) {
$t = self::BOOTSTRAP_TMIN;
} elseif ($k >= $bias + self::BOOTSTRAP_TMAX) {
$t = self::BOOTSTRAP_TMAX;
} else {
$t = $k - $bias;
}
if ($q < $t) {
break;
}
$digit = $t + ($q - $t) % (self::BOOTSTRAP_BASE - $t);
$output .= self::digit_to_char($digit);
$q = floor(($q - $t) / (self::BOOTSTRAP_BASE - $t));
}
$output .= self::digit_to_char($q);
$bias = self::adapt($delta, $h + 1, $h === $b);
$delta = 0;
$h++;
}
}
$delta++;
$n++;
}
return $output;
}
protected static function digit_to_char($digit)
{
if ($digit < 0 || $digit > 35) {
throw new Requests_Exception(sprintf('Invalid digit %d', $digit), 'idna.invalid_digit', $digit);
}
$digits = 'abcdefghijklmnopqrstuvwxyz0123456789';
return substr($digits, $digit, 1);
}
protected static function adapt($delta, $numpoints, $firsttime)
{
if ($firsttime) {
$delta = floor($delta / self::BOOTSTRAP_DAMP);
} else {
$delta = floor($delta / 2);
}
$delta += floor($delta / $numpoints);
$k = 0;
$max = floor((self::BOOTSTRAP_BASE - self::BOOTSTRAP_TMIN) * self::BOOTSTRAP_TMAX / 2);
while ($delta > $max) {
$delta = floor($delta / (self::BOOTSTRAP_BASE - self::BOOTSTRAP_TMIN));
$k += self::BOOTSTRAP_BASE;
}
return $k + floor((self::BOOTSTRAP_BASE - self::BOOTSTRAP_TMIN + 1) * $delta / ($delta + self::BOOTSTRAP_SKEW));
}
}