minimalnfc.php

<?
/*

  minimalnfc.php - Fast and limited Unicode normalizer written in pure PHP.

  Use utf8_normalize() instead of Normalizer::normalize when you do not need
  full Unicode normalization and prefer something tiny with less dependancies.

  The function operates directly on UTF-8 and replaces decomposed unicode
  characters found in regular latin text (basically ISO8859-1) with their
  precomposed counterpart using Normalization Form C.

  Written in 2014 by Johannes Lundberg.
  Released to the public domain.

*/

function utf8_normalize ($s) {

  $latin_nfc = array(
    "A\xcc\x80" => "\xc3\x80",
    "A\xcc\x81" => "\xc3\x81",
    "A\xcc\x82" => "\xc3\x82",
    "A\xcc\x83" => "\xc3\x83",
    "A\xcc\x88" => "\xc3\x84",
    "A\xcc\x8a" => "\xc3\x85",
    "C\xcc\xa7" => "\xc3\x87",
    "E\xcc\x80" => "\xc3\x88",
    "E\xcc\x81" => "\xc3\x89",
    "E\xcc\x82" => "\xc3\x8a",
    "E\xcc\x88" => "\xc3\x8b",
    "I\xcc\x80" => "\xc3\x8c",
    "I\xcc\x81" => "\xc3\x8d",
    "I\xcc\x82" => "\xc3\x8e",
    "I\xcc\x88" => "\xc3\x8f",
    "N\xcc\x83" => "\xc3\x91",
    "O\xcc\x80" => "\xc3\x92",
    "O\xcc\x81" => "\xc3\x93",
    "O\xcc\x82" => "\xc3\x94",
    "O\xcc\x83" => "\xc3\x95",
    "O\xcc\x88" => "\xc3\x96",
    "U\xcc\x80" => "\xc3\x99",
    "U\xcc\x81" => "\xc3\x9a",
    "U\xcc\x82" => "\xc3\x9b",
    "U\xcc\x88" => "\xc3\x9c",
    "Y\xcc\x81" => "\xc3\x9d",
    "a\xcc\x80" => "\xc3\xa0",
    "a\xcc\x81" => "\xc3\xa1",
    "a\xcc\x82" => "\xc3\xa2",
    "a\xcc\x83" => "\xc3\xa3",
    "a\xcc\x88" => "\xc3\xa4",
    "a\xcc\x8a" => "\xc3\xa5",
    "c\xcc\xa7" => "\xc3\xa7",
    "e\xcc\x80" => "\xc3\xa8",
    "e\xcc\x81" => "\xc3\xa9",
    "e\xcc\x82" => "\xc3\xaa",
    "e\xcc\x88" => "\xc3\xab",
    "i\xcc\x80" => "\xc3\xac",
    "i\xcc\x81" => "\xc3\xad",
    "i\xcc\x82" => "\xc3\xae",
    "i\xcc\x88" => "\xc3\xaf",
    "n\xcc\x83" => "\xc3\xb1",
    "o\xcc\x80" => "\xc3\xb2",
    "o\xcc\x81" => "\xc3\xb3",
    "o\xcc\x82" => "\xc3\xb4",
    "o\xcc\x83" => "\xc3\xb5",
    "o\xcc\x88" => "\xc3\xb6",
    "u\xcc\x80" => "\xc3\xb9",
    "u\xcc\x81" => "\xc3\xba",
    "u\xcc\x82" => "\xc3\xbb",
    "u\xcc\x88" => "\xc3\xbc",
    "y\xcc\x81" => "\xc3\xbd",
    "y\xcc\x88" => "\xc3\xbf"
  );

  $out = array();

  /* Every interesting character is at least 3 bytes long. */
  $lastpos = 0;
  for ($pos = 0; $pos < strlen($s)-2; $pos++) {
    if ($s[$pos+1] == "\xCC") {
      $letter = substr($s,$pos,3);
      if (isset($latin_nfc[$letter])) {
        $out[] = substr($s,$lastpos,$pos-$lastpos);
        $out[] = $latin_nfc[$letter];
        $pos += 2;
        $lastpos = $pos + 1;
      }
    }
  }
  $out[] = substr($s,$lastpos,$pos-$lastpos+2);

  return implode('',$out);
}

?>