Make StatementGuid preserve case

Due to the `EntityId` constructor normalizing ids to upper case, `StatementGuid` couldn't guarantee that the statement id is in the correct case as it is stored in the database. This is an issue for the WB REST API as statement ids in the response were always normalized to have upper case entity id prefixes.

This commit fixes this by adding an optional `$originalStatementId` parameter to `StatementGuid` for providing the original, non-normalized, statement id and updating `StatementGuidParser` to always use this 3rd parameter.

Bug: T354262
Change-Id: Ie2e16a188d9aa329e14a26c4e6bff4b22e796717

Author: outdooracorn

src/Statement/StatementGuid.php

@@ -2,6 +2,7 @@
 
 namespace Wikibase\DataModel\Statement;
 
+use InvalidArgumentException;
 use Wikibase\DataModel\Entity\EntityId;
 
 /**
@@ -25,10 +26,23 @@ class StatementGuid {
 	private string $guidPart;
 	private string $serialization;
 
-	public function __construct( EntityId $entityId, string $guid ) {
-		$this->serialization = $entityId->getSerialization() . self::SEPARATOR . $guid;
+	/**
+	 * @param EntityId $entityId the normalized entity id
+	 * @param string $guidPart the guid part of the statement id that appears after the separator
+	 * @param string|null $originalStatementId the original, non-normalized, statement id.
+	 * Defaults to `null` for compatability.
+	 */
+	public function __construct( EntityId $entityId, string $guidPart, string $originalStatementId = null ) {
+		$constructedStatementId = $entityId->getSerialization() . self::SEPARATOR . $guidPart;
+		if ( $originalStatementId !== null
+			 && strtolower( $originalStatementId ) !== strtolower( $constructedStatementId ) ) {
+			throw new InvalidArgumentException( '$originalStatementId does not match $entityId and/or $guidPart' );
+		}
+
+		// use the original serialization when available to avoid normalizing the entity id prefix
+		$this->serialization = $originalStatementId ?? $constructedStatementId;
 		$this->entityId = $entityId;
-		$this->guidPart = $guid;
+		$this->guidPart = $guidPart;
 	}
 
 	public function getEntityId(): EntityId {
@@ -43,9 +57,9 @@ public function getGuidPart(): string {
 	}
 
 	/**
-	 * @deprecated The value returned by this method might differ in case from the original, unparsed statement GUID
-	 * (the entity ID part might have been lowercase originally, but is always normalized in the return value here),
-	 * which means that the value should not be compared to other statement GUID serializations,
+	 * If the `$originalStatementId` parameter is not used when constructing the StatementGuid object,
+	 * then this method will return a statement id where the entity id prefix is normalized to upper case.
+	 * This could cause issues when comparing to other statement id serializations,
 	 * e.g. to look up a statement in a StatementList.
 	 */
 	public function getSerialization(): string {

tests/unit/Statement/StatementGuidTest.php

@@ -19,41 +19,62 @@
 class StatementGuidTest extends TestCase {
 
 	/**
-	 * @dataProvider provideConstructionData
+	 * @dataProvider provideValidConstructionData
 	 */
-	public function testConstructor( EntityId $entityId, string $guid, string $expected ): void {
-		$statementGuid = new StatementGuid( $entityId, $guid );
+	public function testConstructor( EntityId $entityId, string $guidPart, ?string $originalSerialization, string $expected ): void {
+		$statementGuid = new StatementGuid( $entityId, $guidPart, $originalSerialization );
 
 		$this->assertSame( $expected, $statementGuid->getSerialization() );
 		$this->assertEquals( $entityId, $statementGuid->getEntityId() );
-		$this->assertSame( $guid, $statementGuid->getGuidPart() );
+		$this->assertSame( $guidPart, $statementGuid->getGuidPart() );
 	}
 
-	public static function provideConstructionData(): array {
+	public static function provideValidConstructionData(): array {
 		return [
 			[
 				new ItemId( 'q42' ),
 				'D8404CDA-25E4-4334-AF13-A3290BCD9C0N',
+				null,
 				'Q42$D8404CDA-25E4-4334-AF13-A3290BCD9C0N',
 			],
+			[
+				new ItemId( 'q42' ),
+				'D8404CDA-25E4-4334-AF13-A3290BCD9C0N',
+				'q42$D8404CDA-25E4-4334-AF13-A3290BCD9C0N',
+				'q42$D8404CDA-25E4-4334-AF13-A3290BCD9C0N',
+			],
+			[
+				new ItemId( 'Q1234567' ),
+				'D4FDE516-F20C-4154-ADCE-7C5B609DFDFF',
+				null,
+				'Q1234567$D4FDE516-F20C-4154-ADCE-7C5B609DFDFF',
+			],
 			[
 				new ItemId( 'Q1234567' ),
 				'D4FDE516-F20C-4154-ADCE-7C5B609DFDFF',
 				'Q1234567$D4FDE516-F20C-4154-ADCE-7C5B609DFDFF',
+				'Q1234567$D4FDE516-F20C-4154-ADCE-7C5B609DFDFF',
+			],
+			[
+				new ItemId( 'Q1' ),
+				'foo',
+				null,
+				'Q1$foo',
 			],
 			[
 				new ItemId( 'Q1' ),
 				'foo',
 				'Q1$foo',
+				'Q1$foo',
 			],
 		];
 	}
 
 	public function provideStatementGuids(): array {
 		$argLists = [];
 
-		foreach ( $this->provideConstructionData() as $data ) {
-			$argLists[] = [ new StatementGuid( $data[0], $data[1] ) ];
+		foreach ( $this->provideValidConstructionData() as $data ) {
+			$argLists[] = [ new StatementGuid( ...$data ) ];
 		}
 
 		return $argLists;