From a24aa5943c4d0f0a5c074b0a5ef9a386709aaf4a Mon Sep 17 00:00:00 2001
From: Jeroen De Dauw <jeroendedauw@gmail.com>
Date: Tue, 14 May 2019 20:09:39 +0200
Subject: [PATCH] Add FingerprintDiffer

---
 src/Diff/FingerprintDiffer.php            |  62 ++++++++++++
 tests/unit/Diff/FingerprintDifferTest.php | 118 ++++++++++++++++++++++
 2 files changed, 180 insertions(+)
 create mode 100644 src/Diff/FingerprintDiffer.php
 create mode 100644 tests/unit/Diff/FingerprintDifferTest.php

diff --git a/src/Diff/FingerprintDiffer.php b/src/Diff/FingerprintDiffer.php
new file mode 100644
index 00000000..f817cf3c
--- /dev/null
+++ b/src/Diff/FingerprintDiffer.php
@@ -0,0 +1,62 @@
+<?php
+
+namespace Wikibase\DataModel\Services\Diff;
+
+use Wikibase\DataModel\Term\AliasGroup;
+use Wikibase\DataModel\Term\Fingerprint;
+
+/**
+ * @license GPL-2.0-or-later
+ * @author Jeroen De Dauw < jeroendedauw@gmail.com >
+ */
+class FingerprintDiffer {
+
+	/**
+	 * @param Fingerprint $first
+	 * @param Fingerprint $second
+	 *
+	 * @return Fingerprint Contains the terms present in $first that are not present in $second
+	 */
+	public function diff( Fingerprint $first, Fingerprint $second ): Fingerprint {
+		$difference = new Fingerprint();
+
+		foreach ( $first->getLabels() as $term ) {
+			if ( !$second->getLabels()->hasTerm( $term ) ) {
+				$difference->getLabels()->setTerm( $term );
+			}
+		}
+
+		foreach ( $first->getDescriptions() as $term ) {
+			if ( !$second->getDescriptions()->hasTerm( $term ) ) {
+				$difference->getDescriptions()->setTerm( $term );
+			}
+		}
+
+		foreach ( $first->getAliasGroups() as $aliasGroup ) {
+			$difference->setAliasGroup(
+				$aliasGroup->getLanguageCode(),
+				$this->diffAliasGroup( $aliasGroup, $second )
+			);
+		}
+
+		return $difference;
+	}
+
+	private function diffAliasGroup( AliasGroup $aliasGroup, Fingerprint $second ) {
+		if ( $second->hasAliasGroup( $aliasGroup->getLanguageCode() ) ) {
+			$secondAliases = $second->getAliasGroup( $aliasGroup->getLanguageCode() )->getAliases();
+			$differenceAliases = [];
+
+			foreach ( $aliasGroup->getAliases() as $aliasText ) {
+				if ( !in_array( $aliasText, $secondAliases ) ) {
+					$differenceAliases[] = $aliasText;
+				}
+			}
+
+			return $differenceAliases;
+		}
+
+		return $aliasGroup->getAliases();
+	}
+
+}
diff --git a/tests/unit/Diff/FingerprintDifferTest.php b/tests/unit/Diff/FingerprintDifferTest.php
new file mode 100644
index 00000000..c8c1ecc3
--- /dev/null
+++ b/tests/unit/Diff/FingerprintDifferTest.php
@@ -0,0 +1,118 @@
+<?php
+
+namespace Wikibase\DataModel\Services\Tests\Diff;
+
+use PHPUnit\Framework\TestCase;
+use Wikibase\DataModel\Services\Diff\FingerprintDiffer;
+use Wikibase\DataModel\Term\AliasGroup;
+use Wikibase\DataModel\Term\AliasGroupList;
+use Wikibase\DataModel\Term\Fingerprint;
+use Wikibase\DataModel\Term\Term;
+use Wikibase\DataModel\Term\TermList;
+
+/**
+ * @covers \Wikibase\DataModel\Services\Diff\FingerprintDiffer
+ */
+class FingerprintDifferTest extends TestCase {
+
+	public function testEmptyFingerprints() {
+		$this->assertEquals(
+			new Fingerprint(),
+			$this->diff( new Fingerprint(), new Fingerprint() )
+		);
+	}
+
+	private function diff( Fingerprint $first, Fingerprint $second ): Fingerprint {
+		return ( new FingerprintDiffer() )->diff( $first, $second );
+	}
+
+	public function testTermsNotInSecondAreReturned() {
+		$first = new Fingerprint(
+			new TermList( [ new Term( 'en', 'FirstLabel' ) ] ),
+			new TermList( [ new Term( 'en', 'FirstDesc' ) ] ),
+			new AliasGroupList( [
+				new AliasGroup( 'en', [ 'FirstAliasOne', 'FirstAliasTwo' ] ),
+				new AliasGroup( 'de', [ 'DeAliasOne', 'DeAliasTwo' ] ),
+			] )
+		);
+
+		$this->assertEquals(
+			$first,
+			$this->diff(
+				$first,
+				new Fingerprint(
+					new TermList( [ new Term( 'en', 'SecondLabel' ) ] ),
+					new TermList( [ new Term( 'en', 'SecondDesc' ) ] ),
+					new AliasGroupList( [
+						new AliasGroup( 'en', [ 'SecondAliasOne', 'SecondAliasTwo' ] ),
+						new AliasGroup( 'different', [ 'DeAliasOne', 'DeAliasTwo' ] ),
+					] )
+				)
+			)
+		);
+	}
+
+	public function testTermsInSecondAreNotReturned() {
+		$fingerprint = new Fingerprint(
+			new TermList( [ new Term( 'en', 'FirstLabel' ) ] ),
+			new TermList( [ new Term( 'en', 'FirstDesc' ) ] ),
+			new AliasGroupList( [
+				new AliasGroup( 'en', [ 'FirstAliasOne', 'FirstAliasTwo' ] ),
+			] )
+		);
+
+		$this->assertEquals(
+			new Fingerprint(),
+			$this->diff(
+				$fingerprint,
+				$fingerprint
+			)
+		);
+	}
+
+	public function testAllTheThings() {
+		$this->assertEquals(
+			new Fingerprint(
+				new TermList( [
+					new Term( 'fr', 'OnlyInFirst' )
+				] ),
+				new TermList( [
+					new Term( 'de', 'MismatchLanguage' )
+				] ),
+				new AliasGroupList( [
+					new AliasGroup( 'en', [ 'AliasOnlyInFirst' ] ),
+					new AliasGroup( 'de', [ 'AliasesOnlyInFirst' ] )
+				] )
+			),
+			$this->diff(
+				new Fingerprint(
+					new TermList( [
+						new Term( 'en', 'MatchingTerm' ),
+						new Term( 'fr', 'OnlyInFirst' )
+					] ),
+					new TermList( [
+						new Term( 'de', 'MismatchLanguage' )
+					] ),
+					new AliasGroupList( [
+						new AliasGroup( 'en', [ 'MatchingAlias', 'AliasOnlyInFirst' ] ),
+						new AliasGroup( 'de', [ 'AliasesOnlyInFirst' ] ),
+					] )
+				),
+				new Fingerprint(
+					new TermList( [
+						new Term( 'en', 'MatchingTerm' ),
+						new Term( 'ru', 'OnlyInSecond' )
+					] ),
+					new TermList( [
+						new Term( 'different', 'MismatchLanguage' )
+					] ),
+					new AliasGroupList( [
+						new AliasGroup( 'en', [ 'MatchingAlias', 'AliasOnlyInSecond' ] ),
+						new AliasGroup( 'ru', [ 'AliasesOnlyInSecond' ] ),
+					] )
+				)
+			)
+		);
+	}
+
+}