clean_company_names_to_slugs.php

<?php

function python_slug( $str ) {
  $str = Normalizer::normalize( $str, Normalizer::FORM_KD );
  $str = preg_replace( '~[\x00-\x1F\x80-\xFF]~', '', $str );
  $str = preg_replace( '~[^\w\s-]~', '', $str );
  $str = strtolower( trim( $str ) );
  return preg_replace( '~[-\s]+~', '-', $str );
}

function clean_company_names_to_slugs( $str ) {
  $str = python_slug( $str );

  $endings = array(
    '',
    'ag',
    'as',
    'bv',
    'co',
    'coltd',
    'com',
    'company',
    'corp',
    'corporation',
    'de\-cv',
    'gmbh',
    'group',
    'inc',
    'kg',
    'kk',
    'ltd',
    'ltda',
    'limited',
    'llc',
    'llp',
    'plc',
    'pt',
    'pte',
    'pty',
    'pvt',
    'sdn\-bhd',
    'sa',
    'usa',
  );

  do {
    $prev = $str;
    $str = preg_replace( '~\-(' . implode( '|', $endings ) . ')$~', '', $prev );
  } while ( $prev !== $str );

  return $str;
}