Add first unit tests for language analysis and guid language detection

tests/phpunit/FieldLanguagePatternTest.php

@@ -0,0 +1,44 @@
+<?php
+
+use WPML\ElasticPress\Field\Sync;
+
+class FieldLanguagePatternTestDouble extends Sync {
+	public function callBuildLanguagePattern() {
+		return $this->buildLanguagePattern();
+	}
+}
+
+class FieldLanguagePatternTest extends OTGS_TestCase {
+
+	public function guidProvider() {
+		return [
+			'language as directory'   => [ 'http://example.com/de/some-post/', 'de' ],
+			'language as parameter'   => [ 'http://example.com/?p=1&lang=fr', 'fr' ],
+			'language as subdomain'   => [ 'http://de.example.com/some-post/', 'de' ],
+			'default language guid'   => [ 'http://example.com/some-post/', null ],
+			'unrelated language code' => [ 'http://example.com/it/some-post/', null ],
+		];
+	}
+
+	/**
+	 * @dataProvider guidProvider
+	 *
+	 * @param string      $guid
+	 * @param string|null $expectedLanguage
+	 */
+	public function test_pattern_extracts_language_from_guid( $guid, $expectedLanguage ) {
+		$subject = new FieldLanguagePatternTestDouble( '7.10', [ 'en', 'de', 'fr' ], 'en', 'en' );
+
+		$pattern = $subject->callBuildLanguagePattern();
+		$matched = preg_match( $pattern, $guid, $match );
+
+		if ( null === $expectedLanguage ) {
+			$this->assertSame( 0, $matched );
+			return;
+		}
+
+		$this->assertSame( 1, $matched );
+		// getPostLanguage() reads the last captured group.
+		$this->assertSame( $expectedLanguage, end( $match ) );
+	}
+}

tests/phpunit/TranslateLanguagesTest.php

@@ -0,0 +1,58 @@
+<?php
+
+use WPML\ElasticPress\Traits\TranslateLanguages;
+
+class TranslateLanguagesTestDouble {
+	use TranslateLanguages;
+
+	public function callGenerateAnalysisLanguages( $languageCode ) {
+		return $this->generateAnalysisLanguages( $languageCode );
+	}
+
+	public function callLanguageHasStemmer( $language ) {
+		return $this->languageHasStemmer( $language );
+	}
+}
+
+class TranslateLanguagesTest extends OTGS_TestCase {
+
+	public function analysisLanguagesProvider() {
+		return [
+			'german has analyzer and snowball' => [ 'de', 'german', 'German' ],
+			'english is fully supported'       => [ 'en', 'english', 'English' ],
+			'arabic has no snowball'           => [ 'ar', 'arabic', 'English' ],
+			'thai has no snowball'             => [ 'th', 'thai', 'English' ],
+			'brazilian portuguese maps'        => [ 'pt-br', 'brazilian', 'English' ],
+			'norwegian nynorsk maps'           => [ 'nn', 'norwegian', 'Norwegian' ],
+			'unknown falls back to english'    => [ 'xx', 'english', 'English' ],
+		];
+	}
+
+	/**
+	 * @dataProvider analysisLanguagesProvider
+	 *
+	 * @param string $languageCode
+	 * @param string $expectedAnalyzer
+	 * @param string $expectedSnowball
+	 */
+	public function test_generate_analysis_languages( $languageCode, $expectedAnalyzer, $expectedSnowball ) {
+		$subject = new TranslateLanguagesTestDouble();
+
+		$this->assertSame(
+			[
+				'analyzer' => $expectedAnalyzer,
+				'snowball' => $expectedSnowball,
+			],
+			$subject->callGenerateAnalysisLanguages( $languageCode )
+		);
+	}
+
+	public function test_language_has_stemmer() {
+		$subject = new TranslateLanguagesTestDouble();
+
+		$this->assertTrue( $subject->callLanguageHasStemmer( 'german' ) );
+		$this->assertTrue( $subject->callLanguageHasStemmer( 'arabic' ) );
+		$this->assertFalse( $subject->callLanguageHasStemmer( 'thai' ) );
+		$this->assertFalse( $subject->callLanguageHasStemmer( 'unknown' ) );
+	}
+}