dutch

doctrine · Oct 28, 2020 · 119dfde · noud · Oct 28, 2020 · greg0ire
1 parent cc041e1
commit 119dfde
Show file tree

Hide file tree

Showing 9 changed files with 346 additions and 0 deletions.
diff --git a/docs/en/index.rst b/docs/en/index.rst
@@ -38,6 +38,7 @@ you want to create an inflector for to the ``createForLanguage()`` method:
 
 The supported languages are as follows:
 
+- ``Language::DUTCH``
 - ``Language::ENGLISH``
 - ``Language::FRENCH``
 - ``Language::NORWEGIAN_BOKMAL``

diff --git a/lib/Doctrine/Inflector/InflectorFactory.php b/lib/Doctrine/Inflector/InflectorFactory.php
@@ -4,6 +4,7 @@
 
 namespace Doctrine\Inflector;
 
+use Doctrine\Inflector\Rules\Dutch;
 use Doctrine\Inflector\Rules\English;
 use Doctrine\Inflector\Rules\French;
 use Doctrine\Inflector\Rules\NorwegianBokmal;
@@ -24,6 +25,9 @@ public static function create(): LanguageInflectorFactory
     public static function createForLanguage(string $language): LanguageInflectorFactory
     {
         switch ($language) {
+            case Language::DUTCH:
+                return new Dutch\InflectorFactory();
+
             case Language::ENGLISH:
                 return new English\InflectorFactory();
 

diff --git a/lib/Doctrine/Inflector/Language.php b/lib/Doctrine/Inflector/Language.php
@@ -6,6 +6,7 @@
 
 final class Language
 {
+    public const DUTCH            = 'dutch';
     public const ENGLISH          = 'english';
     public const FRENCH           = 'french';
     public const NORWEGIAN_BOKMAL = 'norwegian-bokmal';

diff --git a/lib/Doctrine/Inflector/Rules/Dutch/Inflectible.php b/lib/Doctrine/Inflector/Rules/Dutch/Inflectible.php
@@ -0,0 +1,164 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Inflector\Rules\Dutch;
+
+use Doctrine\Inflector\Rules\Pattern;
+use Doctrine\Inflector\Rules\Substitution;
+use Doctrine\Inflector\Rules\Transformation;
+use Doctrine\Inflector\Rules\Word;
+
+/**
+ *  // http://nl.wikipedia.org/wiki/Klinker_(klank)
+ *  private $klinker = '(a|e|i|o|u|ij)';
+ *  private $korteKlinker = '(u|i|e|a|o)';          // @todo '(ie|oe)'. $plofKlank
+ *
+ *  // http://nl.wikipedia.org/wiki/Medeklinker
+ *  private $plofKlank = '(p|t|k|b|d)';
+ *  private $wrijfKlank = '(f|s|ch|sj|v|z|g|j)';    // journaal
+ *  private $neusKlank = '(m|n|ng)';
+ *  private $vloeiKlank = '(l|r)';
+ *  private $glijKlank = '(j|w)';                   // jaar
+ *  private $affricate = '(ts|zz|tsj|g)';           // /d3/ gin
+ *  private $missingFromWiki = '(c|h|p|q|x|y|z)';
+ *
+ *  public function __construct()
+ *  {
+ *      $this->medeklinker = '(' . $this->missingFromWiki . '|' . $this->plofKlank . '|' . $this->wrijfKlank . '|' . $this->neusKlank . '|' . $this->vloeiKlank . '|' . $this->glijKlank . '|' . $this->affricate . ')';
+ *      $this->medeklinker = '((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g))';
+ *  }
+ */
+class Inflectible
+{
+    /**
+     * @return Transformation[]
+     */
+    public static function getSingular(): iterable
+    {
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Klinkerverandering
+        yield new Transformation(new Pattern('()heden$'), '\1heid');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Beroepen_eindigend_op_-man
+        yield new Transformation(new Pattern('()mannen$'), '\1man');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Latijnse_meervoudsvormen
+        yield new Transformation(new Pattern('()ices$'), '\1ex');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Stapelmeervoud
+        yield new Transformation(new Pattern('^(ei|gemoed|goed|hoen|kind|lied|rad|rund)eren$'), '\1');
+
+        // http://nl.wikipedia.org/wiki/Nederlandse_grammatica
+        yield new Transformation(new Pattern('()ijen$'), '\1ij');
+
+        yield new Transformation(new Pattern('()ieen$'), '\1ie');       // ën
+
+        yield new Transformation(new Pattern('()((a|e|i|o|u|ij))s$'), '\1\2');
+
+        yield new Transformation(new Pattern('()((s)s)en$'), '\1s');
+
+        yield new Transformation(new Pattern('()((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g))en$'), '\1\2');
+    }
+
+    /**
+     * @return Transformation[]
+     */
+    public static function getPlural(): iterable
+    {
+        // @todo already in plural (?)
+        // @todo refine
+        yield new Transformation(new Pattern('()(e)(s)$'), '\1\2\3\3en');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Klinkerverandering
+        yield new Transformation(new Pattern('()heid$'), '\1heden');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Beroepen_eindigend_op_-man
+        yield new Transformation(new Pattern('()man$'), '\1mannen');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Latijnse_meervoudsvormen
+        yield new Transformation(new Pattern('()ix$'), '\1ices');
+
+        yield new Transformation(new Pattern('()ex$'), '\1ices');
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Stapelmeervoud
+        yield new Transformation(new Pattern('^(ei|gemoed|goed|hoen|kind|lied|rad|rund)$'), '\1eren');
+
+        // http://nl.wikipedia.org/wiki/Nederlandse_grammatica
+        yield new Transformation(new Pattern('()ij$'), '\1ijen');
+
+        yield new Transformation(new Pattern('()orie$'), '\1orieen');   // ën klemtoon
+
+        yield new Transformation(new Pattern('()io$'), '\1io\'s');
+
+        yield new Transformation(new Pattern('()(a|e|i|o|u|ij)$'), '\1\2s');
+
+        yield new Transformation(new Pattern('()(((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g))e((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g)))$'), '\1\2s');
+
+        yield new Transformation(new Pattern('()(((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g))(u|i|e|a|o)s)$'), '\1\2sen');
+
+        yield new Transformation(new Pattern('()(((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g))s)$'), '\1\2en');
+
+        yield new Transformation(new Pattern('()s$'), '\1zen');
+
+        yield new Transformation(new Pattern('()((c|h|p|q|x|y|z)|(p|t|k|b|d)|(f|s|ch|sj|v|z|g|j)|(m|n|ng)|(l|r)|(j|w)|(ts|zz|tsj|g))$'), '\1\2en');
+    }
+
+    /**
+     * @return Substitution[]
+     */
+    public static function getIrregular(): iterable
+    {
+        // http://nl.wikipedia.org/wiki/Klemtoon
+        yield new Substitution(new Word('olie'), new Word('oliën'));
+
+        yield new Substitution(new Word('industrie'), new Word('industrieën'));
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Klinkerverandering
+        yield new Substitution(new Word('lid'), new Word('leden'));
+
+        yield new Substitution(new Word('smid'), new Word('smeden'));
+
+        yield new Substitution(new Word('schip'), new Word('schepen'));
+
+        yield new Substitution(new Word('stad'), new Word('steden'));
+
+        yield new Substitution(new Word('gelid'), new Word('gelederen'));
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Stapelmeervoud
+        yield new Substitution(new Word('gelid'), new Word('gelederen'));
+
+        yield new Substitution(new Word('kalf'), new Word('kalveren'));
+
+        yield new Substitution(new Word('lam'), new Word('lammeren'));
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Onregelmatige_meervoudsvorming
+        yield new Substitution(new Word('koe'), new Word('koeien'));
+
+        yield new Substitution(new Word('vlo'), new Word('vlooien'));
+
+        yield new Substitution(new Word('leerrede'), new Word('leerredenen'));
+
+        yield new Substitution(new Word('lende'), new Word('lendenen'));
+
+        yield new Substitution(new Word('epos'), new Word('epen'));
+
+        yield new Substitution(new Word('genius'), new Word('geniën'));
+
+        yield new Substitution(new Word('aanbod'), new Word('aanbiedingen'));
+
+        yield new Substitution(new Word('beleg'), new Word('belegeringen'));
+
+        yield new Substitution(new Word('dank'), new Word('dankbetuigingen'));
+
+        yield new Substitution(new Word('gedrag'), new Word('gedragingen'));
+
+        yield new Substitution(new Word('genot'), new Word('genietingen'));
+
+        yield new Substitution(new Word('lof'), new Word('lofbetuigingen'));
+
+        // http://nl.wikipedia.org/wiki/Meervoud_(Nederlands)#Latijnse_meervoudsvormen
+        yield new Substitution(new Word('qaestrices'), new Word('quaestrix'));
+
+        yield new Substitution(new Word('matrices'), new Word('matrix'));
+    }
+}
diff --git a/lib/Doctrine/Inflector/Rules/Dutch/InflectorFactory.php b/lib/Doctrine/Inflector/Rules/Dutch/InflectorFactory.php
@@ -0,0 +1,21 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Inflector\Rules\Dutch;
+
+use Doctrine\Inflector\GenericLanguageInflectorFactory;
+use Doctrine\Inflector\Rules\Ruleset;
+
+final class InflectorFactory extends GenericLanguageInflectorFactory
+{
+    protected function getSingularRuleset(): Ruleset
+    {
+        return Rules::getSingularRuleset();
+    }
+
+    protected function getPluralRuleset(): Ruleset
+    {
+        return Rules::getPluralRuleset();
+    }
+}
diff --git a/lib/Doctrine/Inflector/Rules/Dutch/Rules.php b/lib/Doctrine/Inflector/Rules/Dutch/Rules.php
@@ -0,0 +1,31 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Inflector\Rules\Dutch;
+
+use Doctrine\Inflector\Rules\Patterns;
+use Doctrine\Inflector\Rules\Ruleset;
+use Doctrine\Inflector\Rules\Substitutions;
+use Doctrine\Inflector\Rules\Transformations;
+
+final class Rules
+{
+    public static function getSingularRuleset(): Ruleset
+    {
+        return new Ruleset(
+            new Transformations(...Inflectible::getSingular()),
+            new Patterns(...Uninflected::getSingular()),
+            (new Substitutions(...Inflectible::getIrregular()))->getFlippedSubstitutions()
+        );
+    }
+
+    public static function getPluralRuleset(): Ruleset
+    {
+        return new Ruleset(
+            new Transformations(...Inflectible::getPlural()),
+            new Patterns(...Uninflected::getPlural()),
+            new Substitutions(...Inflectible::getIrregular())
+        );
+    }
+}
diff --git a/lib/Doctrine/Inflector/Rules/Dutch/Uninflected.php b/lib/Doctrine/Inflector/Rules/Dutch/Uninflected.php
@@ -0,0 +1,66 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Inflector\Rules\Dutch;
+
+use Doctrine\Inflector\Rules\Pattern;
+
+final class Uninflected
+{
+    /**
+     * @return Pattern[]
+     */
+    public static function getSingular(): iterable
+    {
+        yield from self::getDefault();
+
+        // http://nl.wikipedia.org/wiki/Plurale_tantum
+        yield new Pattern('hersenen');
+        yield new Pattern('ingewanden');
+        yield new Pattern('mazelen');
+        yield new Pattern('pokken');
+        yield new Pattern('waterpokken');
+        yield new Pattern('financiën');
+        yield new Pattern('activa');
+        yield new Pattern('passiva');
+        yield new Pattern('onkosten');
+        yield new Pattern('kosten');
+        yield new Pattern('bescheiden');
+        yield new Pattern('paperassen');
+        yield new Pattern('notulen');
+        yield new Pattern('Roma');
+        yield new Pattern('Sinti');
+        yield new Pattern('Inuit');
+        yield new Pattern('taliban');
+        yield new Pattern('illuminati');
+        yield new Pattern('aanstalten');
+        yield new Pattern('hurken');
+        yield new Pattern('lurven');
+        yield new Pattern('luren');
+    }
+
+    /**
+     * @return Pattern[]
+     */
+    public static function getPlural(): iterable
+    {
+        yield from self::getDefault();
+
+        // http://nl.wikipedia.org/wiki/Singulare_tantum
+        yield new Pattern('letterkunde');
+        yield new Pattern('muziek');
+        yield new Pattern('heelal');
+        yield new Pattern('vastgoed');
+        yield new Pattern('have');
+        yield new Pattern('nageslacht');
+    }
+
+    /**
+     * @return Pattern[]
+     */
+    private static function getDefault(): iterable
+    {
+        yield new Pattern('twitter');
+    }
+}
diff --git a/tests/Doctrine/Tests/Inflector/InflectorFactoryTest.php b/tests/Doctrine/Tests/Inflector/InflectorFactoryTest.php
@@ -7,6 +7,7 @@
 use Doctrine\Inflector\InflectorFactory;
 use Doctrine\Inflector\Language;
 use Doctrine\Inflector\LanguageInflectorFactory;
+use Doctrine\Inflector\Rules\Dutch\InflectorFactory as DutchInflectorFactory;
 use Doctrine\Inflector\Rules\English\InflectorFactory as EnglishInflectorFactory;
 use Doctrine\Inflector\Rules\French\InflectorFactory as FrenchInflectorFactory;
 use Doctrine\Inflector\Rules\NorwegianBokmal\InflectorFactory as NorwegianBokmalInflectorFactory;
@@ -38,6 +39,7 @@ public function testCreateForLanguageWithCustomLanguage(string $expectedClass, s
      */
     public static function provideLanguages(): Generator
     {
+        yield 'Dutch' => [DutchInflectorFactory::class, Language::DUTCH];
         yield 'English' => [EnglishInflectorFactory::class, Language::ENGLISH];
         yield 'French' => [FrenchInflectorFactory::class, Language::FRENCH];
         yield 'Norwegian Bokmal' => [NorwegianBokmalInflectorFactory::class, Language::NORWEGIAN_BOKMAL];

diff --git a/tests/Doctrine/Tests/Inflector/Rules/Dutch/DutchFunctionalTest.php b/tests/Doctrine/Tests/Inflector/Rules/Dutch/DutchFunctionalTest.php
@@ -0,0 +1,56 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Doctrine\Tests\Inflector\Rules\Dutch;
+
+use Doctrine\Inflector\Inflector;
+use Doctrine\Inflector\InflectorFactory;
+use Doctrine\Inflector\Language;
+use Doctrine\Tests\Inflector\Rules\LanguageFunctionalTest;
+
+class DutchFunctionalTest extends LanguageFunctionalTest
+{
+    /**
+     * @return string[][]
+     */
+    public function dataSampleWords(): array
+    {
+        return [
+            ['aanbod', 'aanbiedingen'],
+            ['adres', 'adressen'],
+            ['beleg', 'belegeringen'],
+            ['dank', 'dankbetuigingen'],
+            ['ei', 'eieren'],
+            ['epos', 'epen'],
+            ['gedrag', 'gedragingen'],
+            ['gelid', 'gelederen'],
+            ['genius', 'geniën'],
+            ['genot', 'genietingen'],
+            ['industrie', 'industrieën'],
+            ['kalf', 'kalveren'],
+            ['koe', 'koeien'],
+            ['lam', 'lammeren'],
+            ['leerrede', 'leerredenen'],
+            ['lende', 'lendenen'],
+            ['lid', 'leden'],
+            ['lof', 'lofbetuigingen'],
+            ['matrices', 'matrix'],
+            ['mogelijkheid', 'mogelijkheden'],
+            ['olie', 'oliën'],
+            ['qaestrices', 'quaestrix'],
+            ['rij', 'rijen'],
+            ['schip', 'schepen'],
+            ['smid', 'smeden'],
+            ['stad', 'steden'],
+            ['twitter', 'twitter'],
+            ['vlo', 'vlooien'],
+            ['weerman', 'weermannen'],
+        ];
+    }
+
+    protected function createInflector(): Inflector
+    {
+        return InflectorFactory::createForLanguage(Language::DUTCH)->build();
+    }
+}