PHP port of https://github.com/plurals/pluralize

Pluralize.php

<?php

class Pluralize
{
  function pluralize($word)
  {
    return $this->replaceWord(
      $word,
      $this->irregularSingles,
      $this->irregularPlurals,
      $this->pluralRules
    );
  }

  function singularize($word)
  {
    return $this->replaceWord(
      $word,
      $this->irregularPlurals,
      $this->irregularSingles,
      $this->singularRules
    );
  }

  private function replaceWord($word, $replaceMap, $keepMap, $rules)
  {
    $token = strtolower($word);
    if (isset($keepMap[$token])) {
      return $word;
    }
    if (isset($replaceMap[$token])) {
      return $replaceMap[$token];
    }
    if (isset($this->uncountables[$token])) {
      return $word;
    }
    foreach ($rules as $pattern => $replacement) {
      if (preg_match($pattern, $token)) {
        return preg_replace($pattern, $replacement, $token);
      }
    }
    return $word;
  }

  private $pluralRules = [
    '/sheep$/' => '$0',
    '/pox$/' => '$0',
    '/o[iu]s$/' => '$0',
    '/measles$/' => '$0',
    '/fish$/' => '$0',
    '/deer$/' => '$0',
    '/[^aeiou]ese$/' => '$0',
    '/pok[eé]mon$/' => '$0',
    '/^thou$/' => 'you',
    '/m[ae]n$/' => 'men',
    '/eaux$/' => '$0',
    '/(child)(?:ren)?$/' => '$1ren',
    '/(pe)(?:rson|ople)$/' => '$1ople',
    '/\b((?:tit)?m|l)(?:ice|ouse)$/' => '$1ice',
    '/(matr|cod|mur|sil|vert|ind|append)(?:ix|ex)$/' => '$1ices',
    '/(x|ch|ss|sh|zz)$/' => '$1es',
    '/([^ch][ieo][ln])ey$/' => '$1ies',
    '/([^aeiouy]|qu)y$/' => '$1ies',
    '/(?:(kni|wi|li)fe|(ar|l|ea|eo|oa|hoo)f)$/' => '$1$2ves',
    '/sis$/' => 'ses',
    '/(apheli|hyperbat|periheli|asyndet|noumen|phenomen|criteri|organ|prolegomen|hedr|automat)(?:a|on)$/' => '$1a',
    '/(agend|addend|millenni|dat|extrem|bacteri|desiderat|strat|candelabr|errat|ov|symposi|curricul|automat|quor)(?:a|um)$/' => '$1a',
    '/(her|at|gr)o$/' => '$1oes',
    '/(seraph|cherub)(?:im)?$/' => '$1im',
    '/(alumn|alg|vertebr)(?:a|ae)$/' => '$1ae',
    '/(alumn|syllab|vir|radi|nucle|fung|cact|stimul|termin|bacill|foc|uter|loc|strat)(?:us|i)$/' => '$1i',
    '/([^l]ias|[aeiou]las|[ejzr]as|[iu]am)$/' => '$1',
    '/(e[mn]u)s?$/' => '$1s',
    '/(alias|[^aou]us|t[lm]as|gas|ris)$/' => '$1es',
    '/(ax|test)is$/' => '$1es',
    '/([^aeiou]ese)$/' => '$1',
    '/[^\x{0000}-\x{007F}]$/u' => '$0',
    '/s?$/' => 's'
  ];

  private $singularRules = [
    '/sheep$/' => '$0',
    '/pox$/' => '$0',
    '/o[iu]s$/' => '$0',
    '/measles$/' => '$0',
    '/fish$/' => '$0',
    '/deer$/' => '$0',
    '/[^aeiou]ese$/' => '$0',
    '/pok[eé]mon$/' => '$0',
    '/men$/' => 'man',
    '/(eau)x?$/' => '$1',
    '/(child)ren$/' => '$1',
    '/(pe)(rson|ople)$/' => '$1rson',
    '/(matr|append)ices$/' => '$1ix',
    '/(cod|mur|sil|vert|ind)ices$/' => '$1ex',
    '/(alumn|alg|vertebr)ae$/' => '$1a',
    '/(apheli|hyperbat|periheli|asyndet|noumen|phenomen|criteri|organ|prolegomen|hedr|automat)a$/' => '$1on',
    '/(agend|addend|millenni|dat|extrem|bacteri|desiderat|strat|candelabr|errat|ov|symposi|curricul|quor)a$/' => '$1um',
    '/(alumn|syllab|vir|radi|nucle|fung|cact|stimul|termin|bacill|foc|uter|loc|strat)(?:us|i)$/' => '$1us',
    '/(test)(?:is|es)$/' => '$1is',
    '/(movie|twelve|abuse|e[mn]u)s$/' => '$1',
    '/(analy|diagno|parenthe|progno|synop|the|empha|cri|ne)(?:sis|ses)$/' => '$1sis',
    '/(x|ch|ss|sh|zz|tto|go|cho|alias|[^aou]us|t[lm]as|gas|(?:her|at|gr)o|[aeiou]ris)(?:es)?$/' => '$1',
    '/(seraph|cherub)im$/' => '$1',
    '/\b((?:tit)?m|l)ice$/' => '$1ouse',
    '/\b(mon|smil)ies$/' => '$1ey',
    '/\b(l|(?:neck|cross|hog|aun)?t|coll|faer|food|gen|goon|group|hipp|junk|vegg|(?:pork)?p|charl|calor|cut)ies$/' => '$1ie',
    '/(dg|ss|ois|lk|ok|wn|mb|th|ch|ec|oal|is|ck|ix|sser|ts|wb)ies$/' => '$1ie',
    '/ies$/' => 'y',
    '/(ar|(?:wo|[ae])l|[eo][ao])ves$/' => '$1f',
    '/(wi|kni|(?:after|half|high|low|mid|non|night|[^\w]|^)li)ves$/' => '$1fe',
    '/(ss)$/' => '$1',
    '/s$/' => ''
  ];

  private $uncountables = [
    'adulthood' => 1,
    'advice' => 1,
    'agenda' => 1,
    'aid' => 1,
    'aircraft' => 1,
    'alcohol' => 1,
    'ammo' => 1,
    'analytics' => 1,
    'anime' => 1,
    'athletics' => 1,
    'audio' => 1,
    'bison' => 1,
    'blood' => 1,
    'bream' => 1,
    'buffalo' => 1,
    'butter' => 1,
    'carp' => 1,
    'cash' => 1,
    'chassis' => 1,
    'chess' => 1,
    'clothing' => 1,
    'cod' => 1,
    'commerce' => 1,
    'cooperation' => 1,
    'corps' => 1,
    'debris' => 1,
    'diabetes' => 1,
    'digestion' => 1,
    'elk' => 1,
    'energy' => 1,
    'equipment' => 1,
    'excretion' => 1,
    'expertise' => 1,
    'firmware' => 1,
    'flounder' => 1,
    'fun' => 1,
    'gallows' => 1,
    'garbage' => 1,
    'graffiti' => 1,
    'hardware' => 1,
    'headquarters' => 1,
    'health' => 1,
    'herpes' => 1,
    'highjinks' => 1,
    'homework' => 1,
    'housework' => 1,
    'information' => 1,
    'jeans' => 1,
    'justice' => 1,
    'kudos' => 1,
    'labour' => 1,
    'literature' => 1,
    'machinery' => 1,
    'mackerel' => 1,
    'mail' => 1,
    'manga' => 1,
    'media' => 1,
    'mews' => 1,
    'moose' => 1,
    'mud' => 1,
    'music' => 1,
    'news' => 1,
    'only' => 1,
    'personnel' => 1,
    'pike' => 1,
    'plankton' => 1,
    'pliers' => 1,
    'police' => 1,
    'pollution' => 1,
    'premises' => 1,
    'rain' => 1,
    'research' => 1,
    'rice' => 1,
    'salmon' => 1,
    'scissors' => 1,
    'series' => 1,
    'sewage' => 1,
    'shambles' => 1,
    'shrimp' => 1,
    'software' => 1,
    'staff' => 1,
    'swine' => 1,
    'tennis' => 1,
    'traffic' => 1,
    'transportation' => 1,
    'trout' => 1,
    'tuna' => 1,
    'wealth' => 1,
    'welfare' => 1,
    'whiting' => 1,
    'wildebeest' => 1,
    'wildlife' => 1,
    'you' => 1
  ];

  private $irregularPlurals = [
    'anathemata' => 1,
    'are' => 1,
    'axes' => 1,
    'canvases' => 1,
    'carves' => 1,
    'dice' => 1,
    'dingoes' => 1,
    'dogmata' => 1,
    'eaves' => 1,
    'echoes' => 1,
    'feet' => 1,
    'geese' => 1,
    'genera' => 1,
    'grooves' => 1,
    'have' => 1,
    'humans' => 1,
    'lemmata' => 1,
    'looies' => 1,
    'our' => 1,
    'ourselves' => 1,
    'oxen' => 1,
    'passersby' => 1,
    'pickaxes' => 1,
    'proofs' => 1,
    'quizzes' => 1,
    'schemata' => 1,
    'stigmata' => 1,
    'stomata' => 1,
    'teeth' => 1,
    'their' => 1,
    'them' => 1,
    'themselves' => 1,
    'these' => 1,
    'they' => 1,
    'thieves' => 1,
    'those' => 1,
    'tornadoes' => 1,
    'torpedoes' => 1,
    'us' => 1,
    'valves' => 1,
    'viscera' => 1,
    'volcanoes' => 1,
    'we' => 1,
    'were' => 1,
    'yeses' => 1,
    'yourselves' => 1
  ];

  private $irregularSingles = [
    'anathema' => 1,
    'axe' => 1,
    'canvas' => 1,
    'carve' => 1,
    'die' => 1,
    'dingo' => 1,
    'dogma' => 1,
    'eave' => 1,
    'echo' => 1,
    'foot' => 1,
    'genus' => 1,
    'goose' => 1,
    'groove' => 1,
    'has' => 1,
    'he' => 1,
    'her' => 1,
    'herself' => 1,
    'himself' => 1,
    'his' => 1,
    'human' => 1,
    'i' => 1,
    'is' => 1,
    'its' => 1,
    'itself' => 1,
    'lemma' => 1,
    'looey' => 1,
    'me' => 1,
    'my' => 1,
    'myself' => 1,
    'ox' => 1,
    'passerby' => 1,
    'pickaxe' => 1,
    'proof' => 1,
    'quiz' => 1,
    'schema' => 1,
    'she' => 1,
    'stigma' => 1,
    'stoma' => 1,
    'that' => 1,
    'them' => 1,
    'themself' => 1,
    'thief' => 1,
    'this' => 1,
    'tooth' => 1,
    'tornado' => 1,
    'torpedo' => 1,
    'valve' => 1,
    'viscus' => 1,
    'volcano' => 1,
    'was' => 1,
    'yes' => 1,
    'yourself' => 1
  ];
}