Merge pull request #29926 from nextcloud/fix/noid/short-accounts-data-value

lib/private/Accounts/AccountManager.php

@@ -343,6 +343,10 @@ class AccountManager implements IAccountManager {
 	}
 
 	public function searchUsers(string $property, array $values): array {
+		// the value col is limited to 255 bytes. It is used for searches only.
+		$values = array_map(function (string $value) {
+			return Util::shortenMultibyteString($value, 255);
+		}, $values);
 		$chunks = array_chunk($values, 500);
 		$query = $this->connection->getQueryBuilder();
 		$query->select('*')
@@ -625,8 +629,11 @@ class AccountManager implements IAccountManager {
 				continue;
 			}
 
+			// the value col is limited to 255 bytes. It is used for searches only.
+			$value = $property['value'] ? Util::shortenMultibyteString($property['value'],  255) : '';
+
 			$query->setParameter('name', $property['name'])
-				->setParameter('value', $property['value'] ?? '');
+				->setParameter('value', $value);
 			$query->executeStatement();
 		}
 	}

lib/public/Util.php

@@ -513,4 +513,28 @@ class Util {
 		}
 		return self::$needUpgradeCache;
 	}
+
+	/**
+	 * Sometimes a string has to be shortened to fit within a certain maximum
+	 * data length in bytes. substr() you may break multibyte characters,
+	 * because it operates on single byte level. mb_substr() operates on
+	 * characters, so does not ensure that the shortend string satisfies the
+	 * max length in bytes.
+	 *
+	 * For example, json_encode is messing with multibyte characters a lot,
+	 * replacing them with something along "\u1234".
+	 *
+	 * This function shortens the string with by $accurancy (-5) from
+	 * $dataLength characters, until it fits within $dataLength bytes.
+	 *
+	 * @since 23.0.0
+	 */
+	public static function shortenMultibyteString(string $subject, int $dataLength, int $accuracy = 5): string {
+		$temp = mb_substr($subject, 0, $dataLength);
+		// json encodes encapsulates the string in double quotes, they need to be substracted
+		while ((strlen(json_encode($temp)) - 2) > $dataLength) {
+			$temp = mb_substr($temp, 0, -$accuracy);
+		}
+		return $temp;
+	}
 }

tests/lib/UtilTest.php

@@ -310,4 +310,11 @@ class UtilTest extends \Test\TestCase {
 			'myApp/vendor/myFancyCSSFile2',
 		], \OC_Util::$styles);
 	}
+
+	public function testShortenMultibyteString() {
+		$this->assertEquals('Short nuff', \OCP\Util::shortenMultibyteString('Short nuff', 255));
+		$this->assertEquals('ABC', \OCP\Util::shortenMultibyteString('ABCDEF', 3));
+		// each of the characters is 12 bytes
+		$this->assertEquals('🙈', \OCP\Util::shortenMultibyteString('🙈🙊🙉', 16, 2));
+	}
 }